IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User'

IBM Tivoli OMEGAMON XE on z/OS
Version 5.1.0
EPILOG for MVS User's Guide
SC27-4034-00
IBM Tivoli OMEGAMON XE on z/OS
Version 5.1.0
EPILOG for MVS User's Guide
SC27-4034-00
Note:
Before using this information and the product it supports, read the information in “Notices” on page 369.
March 2012
This edition applies to version 5, release 1, modification 0 of IBM Tivoli OMEGAMON XE on z/OS (product number
5698-T01) and to all subsequent releases and modifications until otherwise indicated in new editions.
© Copyright IBM Corporation 1990, 1992, 1997, 2009, 2012.
US Government Users Restricted Rights – Use, duplication or disclosure restricted by GSA ADP Schedule Contract
with IBM Corp.
Contents
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
About this guide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Part 1. Basic reporting features . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 1. Introduction to EPILOG . . .
Prevention without the guesswork . . . .
Answers to your performance questions .
Who can use EPILOG for MVS . . . . .
Product overview . . . . . . . . . .
The EPILOG collector . . . . . . .
The EPILOG reporter . . . . . . .
The EPILOG Workload Profiling Facility .
Bottleneck analysis . . . . . . . . .
Performance reporting . . . . . . . .
Profile and workload comparisons . . .
EPILOG reports . . . . . . . . .
Retrospective problem resolution . . . .
Change evaluation . . . . . . . . .
Performance impact on applications . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
3
3
3
4
4
4
5
5
6
6
7
8
9
9
9
Chapter 2. Problem solving . . . . .
The logical tuning approach. . . . . .
Helping your capacity planning effort . .
Service analysis . . . . . . . . .
Cause analysis . . . . . . . . .
Action. . . . . . . . . . . . .
Slowdowns after consolidating workloads.
Service analysis . . . . . . . . .
Cause analysis . . . . . . . . .
Action. . . . . . . . . . . . .
Getting rid of the 4GL blues . . . . .
Service analysis . . . . . . . . .
Cause analysis . . . . . . . . .
Action. . . . . . . . . . . . .
Dealing with shared-DASD contention . .
Service analysis . . . . . . . . .
Cause analysis . . . . . . . . .
Action. . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
11
11
11
11
11
12
12
13
13
14
14
14
15
16
17
17
17
18
Chapter 3. Getting started. . .
Invoking the reporter . . . . .
Mode of operation . . . . .
Reporter option defaults . . .
Datastore buffers . . . . .
Split-screen mode under ISPF .
The initial entry screens . . .
Command syntax and conventions
Delimiters . . . . . . . .
The continuation character . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
19
19
19
20
20
20
22
24
24
24
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
iii
Comment lines . . . . .
Short and long forms . . .
Using masks for operands .
Online help . . . . . . . .
Invoking TSO commands . .
Using the DISPLAY command .
Using the sample datastore . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
24
25
25
25
25
26
27
Chapter 4. Displaying workload information. . . . .
System-wide degradation . . . . . . . . . . . .
Performance group degradation . . . . . . . . . .
Multiple performance groups . . . . . . . . . .
Symbolic performance groups . . . . . . . . . .
Performance period analysis . . . . . . . . . .
Cross-performance group analysis . . . . . . . .
Batch job, started task, and TSO user degradation . . .
Examples . . . . . . . . . . . . . . . . .
Job-, task-, and session-level displays . . . . . . .
Step-level displays . . . . . . . . . . . . . .
Interval-level displays . . . . . . . . . . . . .
Multiple operands . . . . . . . . . . . . . .
Shared-DASD analysis . . . . . . . . . . . . .
Keywords . . . . . . . . . . . . . . . . .
Examples . . . . . . . . . . . . . . . . .
Navigation . . . . . . . . . . . . . . . . .
Workload display options. . . . . . . . . . . . .
Detail and summary displays (DETAIL, SUMMARY) . .
Detail wait reasons . . . . . . . . . . . . . .
Summary wait categories (SUMWAIT) . . . . . . .
Limiting the display by wait reason severity (PLOTMIN)
Average and total performance group figures . . . .
Setting the display scale (MAXSCALE) . . . . . .
Excluding wait reasons from the display . . . . . .
Terminal wait swaps (TWAITON) . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
29
30
31
32
33
33
33
35
37
39
41
42
44
44
44
44
45
46
46
49
51
54
54
55
57
60
Chapter 5. Displaying resource information .
Resource display keywords . . . . . . . .
Sample resource displays . . . . . . . .
Cache subsystem statistics (RCCH). . . .
Channel activity (RCHN) . . . . . . . .
CPU activity (RCPU) . . . . . . . . .
DASD device activity (RDAS) . . . . . .
SRM domain statistics (RDOM) . . . . .
General system information (RINF) . . . .
I/O queuing (RLCU) . . . . . . . . .
Paging and storage (RPAG) . . . . . .
Page data set activity (RPDS) . . . . . .
Performance group statistics (RPGN) . . .
Swap data set activity (RSDS). . . . . .
SRM MPL adjustment values (RSRM) . . .
System swap activity (RSWA) . . . . . .
System swap reason (RSWR) . . . . . .
Virtual Lookaside Facility statistics (RVLF) .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
63
63
64
64
65
66
68
73
74
76
78
81
82
84
84
85
87
89
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Chapter 6. Using the reporter . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Selecting data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
iv
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Setting dates and times . . . . . . . . . .
Setting exception thresholds . . . . . . . .
Setting display defaults . . . . . . . . . .
Combining time intervals . . . . . . . . . . .
Performance groups and system resources . . .
Batch jobs, started tasks, and TSO user sessions
Screen display functions . . . . . . . . . .
Scrolling through reporter output . . . . . .
Displaying the current command . . . . . .
Logging displays . . . . . . . . . . . .
Controlling uppercase output . . . . . . . .
Setting up a color terminal. . . . . . . . .
Reporting in a multiple datastore environment . .
Environmental considerations . . . . . . .
The datastore list . . . . . . . . . . . .
The SYSID keyword . . . . . . . . . . .
Merging shared-DASD data . . . . . . . .
Initializing the active datastore list . . . . . .
Product display functions . . . . . . . . . .
Displaying datastore information. . . . . . .
Displaying product information . . . . . . .
The PRODUCTS command . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 91
. 95
. 95
. 97
. 97
. 100
. 101
. 102
. 103
. 104
. 104
. 104
. 106
. 106
. 106
. 107
. 108
. 111
. 111
. 111
. 114
. 114
Chapter 7. Service level management . . . .
Standard reports . . . . . . . . . . . .
Service levels reports (KEPSRDIR) . . . .
Jobs and groups reports (KEPSROPM) . . .
Performance and tuning reports (KEPSRSPG)
Workload profile comparisons (KEPSRCMP) .
TSO response time . . . . . . . . . . .
Batch jobs . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
115
115
115
115
116
117
117
120
Chapter 8. Fast-path navigation . . .
Using fast-path navigation . . . . . .
The RESOURCE command . . . . .
Navigating in a multisystem environment
Navigating to a shared-DASD display
Expanding the time period . . . . .
Viewing workload utilization directly .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
125
126
130
131
131
133
134
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Part 2. Advanced reporting features . . . . . . . . . . . . . . . . . . . . . 135
Chapter 9. Advanced reporting functions . .
Setting reporter options . . . . . . . . . .
Data set naming conventions. . . . . . .
PF key definitions . . . . . . . . . . .
Summary wait categories . . . . . . . .
Productivity index definitions . . . . . . .
Using the reporter in batch . . . . . . . .
The PAGESEP command . . . . . . . .
The LIMIT keyword . . . . . . . . . .
Workload exception filters . . . . . . . . .
Examples . . . . . . . . . . . . . .
Response time and elapsed time (RESPONSE
Detail wait reasons . . . . . . . . . .
Summary wait reason codes . . . . . . .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
and
. .
. .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
ELAPSED)
. . . .
. . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Contents
137
137
137
137
140
141
142
142
143
143
144
145
146
148
v
I/O and enqueue exceptions . . . .
SELECTIF and REPORTIF processing
Resource exception filters . . . . . .
The REPORTIF keyword . . . . .
RIF command syntax . . . . . .
How exception criteria are applied . .
SET processing. . . . . . . . .
Automatic analysis . . . . . . . .
The Automatic Analysis matrix . . .
A-Matrix definitions . . . . . . .
AUTOMATIC resource displays . . .
The RESOURCE command . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Chapter 10. Workload Profiling Facility (WPF) . . . . . .
WPF overview . . . . . . . . . . . . . . . . . . .
Using WPF . . . . . . . . . . . . . . . . . . . .
Before you begin . . . . . . . . . . . . . . . . . .
Creating profiles . . . . . . . . . . . . . . . . . .
A sample batch job for creating profiles . . . . . . . . .
The PROFILE command: creating profiles . . . . . . . .
Workload keywords . . . . . . . . . . . . . . . .
Date and time keywords . . . . . . . . . . . . . .
Selecting meaningful profiles . . . . . . . . . . . . .
Examples of using the PROFILE command . . . . . . .
Displaying profiles . . . . . . . . . . . . . . . . . .
The profile display. . . . . . . . . . . . . . . . .
The DISPLAY command: generating profile displays or reports
Examples of using the DISPLAY command . . . . . . .
Comparing workloads with profiles . . . . . . . . . . . .
COMPARE displays . . . . . . . . . . . . . . . .
The COMPARE command: comparing workloads with profiles
Examples of using the COMPARE command . . . . . . .
The SETP command: setting default values for the COMPARE
Examples of using the SETP command . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
149
151
155
155
155
158
161
161
161
162
163
165
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
command
. . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
167
167
167
168
168
169
169
171
171
173
176
179
179
180
182
183
184
186
188
189
190
Chapter 11. Exporting EPILOG data . . .
The OBTAIN utility . . . . . . . . . .
Using the OBTAIN utility . . . . . . .
Syntax . . . . . . . . . . . . . .
Output file layouts . . . . . . . . . .
Output file examples . . . . . . . . .
Graphic reporting with sequential EPILOG data
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
193
193
193
193
195
197
199
Chapter 12. Reporting with SAS graphics
How the SAS interface can help you . . .
Performance problem resolution . . .
Management reporting . . . . . . .
Capacity planning and trending . . . .
Reporting overview . . . . . . . .
Designing your own reports . . . . .
Getting started . . . . . . . . . . .
Overview . . . . . . . . . . . .
Running reports . . . . . . . . .
Using the starter kit . . . . . . . . .
Allocating a permanent SAS file. . . .
Modifying the user overrides . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
201
201
201
201
201
202
202
202
202
203
209
210
211
vi
.
.
.
.
.
.
.
.
.
.
.
.
.
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Modifying the procedure KEPILGPM . . . . . . . . . .
The graphic replay procedure . . . . . . . . . . . .
Macros . . . . . . . . . . . . . . . . . . . . .
SAS considerations . . . . . . . . . . . . . . . .
Sample CLIST and batch job for running the SAS interface .
Starter kit reports . . . . . . . . . . . . . . . . . .
User overrides . . . . . . . . . . . . . . . . . .
Channel Utilization report . . . . . . . . . . . . . .
Channel I/O Rate report . . . . . . . . . . . . . .
Channel Peak Utilization report . . . . . . . . . . . .
Channel Peak I/O Rate report . . . . . . . . . . . .
Demand Paging Activity report . . . . . . . . . . . .
Swap Paging Activity report . . . . . . . . . . . . .
CPU Utilization by Performance Group report. . . . . . .
Real Storage Utilization by Performance Group report . . .
Performance Group Response Time Degradation report . . .
Performance Group Trending and Forecasting report . . . .
Batch Job Comparison report . . . . . . . . . . . .
Batch Job Trending and Forecasting report . . . . . . .
Channel Service Rate report . . . . . . . . . . . . .
DASD Utilization report . . . . . . . . . . . . . . .
Batch Program Resource Consumption report . . . . . .
The EXTRACT command . . . . . . . . . . . . . . .
The COMPEXT command . . . . . . . . . . . . . . .
COMPEXT command syntax . . . . . . . . . . . . .
Examples of using the COMPEXT command . . . . . . .
The SETP command: setting default values for the COMPEXT
Chapter 13. Report element tables . . . . . .
Report table organization . . . . . . . . . .
Using resource data . . . . . . . . . . . .
The SAS data sets . . . . . . . . . . .
Table contents . . . . . . . . . . . . .
Using cache resource data (RCCH) . . . . .
Using channel resource data (RCHN) . . . .
Using CPU resource data (RCPU) . . . . . .
Using DASD resource data (RDAS) . . . . .
Using SRM domain activity data (RDOM) . . .
Using general system resource data (RINF) . .
Using I/O queuing data (RLCU) . . . . . . .
Using paging and storage resource data (RPAG)
Using page data set resource data (RPDS) . .
Using performance group resource data (RPGN)
Using swap data set resource data (RSDS) . .
Using SRM MPL statistics (RSRM) . . . . .
Using swap activity data (RSWA) . . . . . .
Using swap reason data (RSWR) . . . . . .
Using virtual lookaside facility statistics (RVLF) .
Using workload degradation data . . . . . . .
The workload tables . . . . . . . . . . .
Performance group degradation data (PGNDETL)
Batch job data (BTCHDETL) . . . . . . . .
Started task degradation data (STCDETL) . . .
TSO user session degradation data (TSODETL)
Comparing workloads and profiles . . . . . . .
The COMPEXT tables . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
command
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
211
212
214
214
214
215
215
216
216
217
217
218
218
219
219
220
221
221
222
223
225
226
227
229
229
230
231
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
233
233
233
234
234
235
239
241
247
250
252
255
259
263
264
267
269
270
273
278
279
279
280
284
286
288
290
290
Contents
vii
Performance group and profile comparisons (PGNCOMP)
Batch job and profile comparisons (BTCHCOMP) . . .
Started task and profile comparisons (STCCOMP) . . .
TSO user session and profile comparisons (TSOCOMP)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
290
293
294
296
Part 3. Appendixes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Appendix A. Data dictionary . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
Appendix B. Documentation library .
OMEGAMON XE on z/OS library . . .
OMEGAMON XE and Tivoli Management
OMEGAMON II for MVS V520 library. .
IBM Tivoli Monitoring library . . . . .
Other sources of documentation . . .
Support information
. . . . . . .
. . . . . . .
Services on z/OS
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . .
. . . . . .
common library
. . . . . .
. . . . . .
. . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
363
363
364
364
365
365
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
viii
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Figures
1. Sample SUMMARY display panel . . . . . . . . . . . .
2. Sample Profile/Workload Comparison report . . . . . . . .
3. SAS graphic display of TSO response components . . . . .
4. System views in the CICS region . . . . . . . . . . . .
5. EPILOG display of SRM MPL adjustment values . . . . . .
6. Performance group statistics for batch jobs (PGN1) . . . . .
7. Largest response time components for performance group 120 .
8. Detail degradation analysis for performance group 120 . . . .
9. DASD device activity display for volume MP220B . . . . . .
10. Cross performance group display . . . . . . . . . . . .
11. EPILOG display for started task CICSPROD. . . . . . . .
12. DASD device activity display for device CICS01 . . . . . .
13. Shared-DASD usage in a multi-system environment . . . . .
14. Primary Option Menu . . . . . . . . . . . . . . . .
15. PF key defaults . . . . . . . . . . . . . . . . . .
16. DISPLAY command syntax . . . . . . . . . . . . . .
17. Sample workload display . . . . . . . . . . . . . . .
18. System-wide degradation display . . . . . . . . . . . .
19. Performance group degradation display . . . . . . . . .
20. Multiple performance group degradation display . . . . . .
21. Performance period degradation display . . . . . . . . .
22. Cross-performance group degradation display . . . . . . .
23. Cross-performance group display with summary wait categories
24. Sample job-level display . . . . . . . . . . . . . . .
25. Started task degradation display . . . . . . . . . . . .
26. TSO session degradation display . . . . . . . . . . . .
27. Batch job step level display . . . . . . . . . . . . . .
28. Batch job interval level display . . . . . . . . . . . . .
29. Started task interval display . . . . . . . . . . . . . .
30. TSO session interval display . . . . . . . . . . . . .
31. Shared-DASD degradation display (by workload) . . . . . .
32. Shared-DASD degradation display (by performance group) . .
33. DETAIL performance group degradation display . . . . . .
34. SUMMARY performance group degradation display . . . . .
35. DETAIL batch job degradation display . . . . . . . . . .
36. SUMMARY batch job degradation display. . . . . . . . .
37. Degradation display with detailed wait reasons . . . . . . .
38. Degradation display with summary wait reasons . . . . . .
39. AVERAGE performance group degradation display . . . . .
40. TOTAL performance group degradation display. . . . . . .
41. Degradation display scale adjustment . . . . . . . . . .
42. Degradation display with excluded waits . . . . . . . . .
43. Degradation Display with Terminal Input Waits . . . . . . .
44. RCCH resource display (MVS/ESA) . . . . . . . . . . .
45. Sample channel resource display (MVS/370) . . . . . . .
46. Sample channel resource display (MVS/XA and MVS/ESA) . .
47. Sample CPU resource display (MVS/370). . . . . . . . .
48. Sample CPU resource display (MVS/XA and MVS/ESA) . . .
49. Sample CPU resource display (with logical partitioning) . . .
50. DASD resource display (MVS/370) . . . . . . . . . . .
51. DASD resource display (MVS/XA non-3090) . . . . . . . .
52. DASD resource display (MVS/XA 3090) . . . . . . . . .
53. DASD resource display (MVS/ESA) . . . . . . . . . . .
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. 7
. 8
. 12
. 13
. 14
. 14
. 15
. 15
. 16
. 16
. 17
. 17
. 18
. 21
. 23
. 26
. 29
. 31
. 32
. 32
. 33
. 34
. 35
. 39
. 41
. 41
. 42
. 43
. 43
. 44
. 45
. 45
. 47
. 47
. 48
. 49
. 52
. 52
. 55
. 55
. 56
. 57
. 61
. 65
. 66
. 66
. 67
. 68
. 68
. 70
. 71
. 72
. 73
ix
54. Sample SRM domain resource display (Pre-MVS/SP 4.2) . . . . .
55. Typical system resource display . . . . . . . . . . . . . .
56. Typical system resource display (with logical partitioning) . . . . .
57. Sample I/O queuing resource display (non-3090 processor) . . . .
58. Sample I/O queuing resource display (3090 processor) . . . . . .
59. Paging and storage resource display (MVS/370) . . . . . . . .
60. Paging and storage resource display (MVS/XA and MVS/ESA) . . .
61. Sample page data set resource display . . . . . . . . . . .
62. Sample performance group resource display . . . . . . . . .
63. RPGN display with performance periods . . . . . . . . . . .
64. Sample swap data set resource display . . . . . . . . . . .
65. SRM resource display . . . . . . . . . . . . . . . . . .
66. Swap activity resource display (Non-3090 processors) . . . . . .
67. Swap activity resource display (3090 processor) . . . . . . . .
68. Swap reason resource display (non-expanded storage). . . . . .
69. Swap reason resource display (with expanded storage) . . . . .
70. RVLF resource display (MVS/ESA) . . . . . . . . . . . . .
71. Combined performance group summary display . . . . . . . .
72. Combined resource display . . . . . . . . . . . . . . . .
73. Sample combined job level display . . . . . . . . . . . . .
74. Sample combined step level display . . . . . . . . . . . .
75. Sample display panel with a title block . . . . . . . . . . .
76. CONTROL COLOR command display . . . . . . . . . . .
77. The RDAS display in a multi-system environment (without MERGE)
78. The RDAS display in a multisystem environment (with MERGE) . .
79. The RDAS display using MERGE and COMBINE . . . . . . .
80. Summary report showing profile of TSO time . . . . . . . . .
81. TSO response time comparison . . . . . . . . . . . . . .
82. Effect of queued I/O on TSO response time . . . . . . . . .
83. Average job step analysis for class A . . . . . . . . . . . .
84. Analysis of jobs running in the wrong class. . . . . . . . . .
85. Analysis of tape mounts . . . . . . . . . . . . . . . . .
86. Reserve analysis . . . . . . . . . . . . . . . . . . .
87. Analysis of long running jobs . . . . . . . . . . . . . . .
88. COMBINEd interval SUMMARY display . . . . . . . . . . .
89. SINGLE interval SUMMARY display . . . . . . . . . . . .
90. SINGLE interval DETAIL display . . . . . . . . . . . . . .
91. SINGLE interval DETAIL display . . . . . . . . . . . . . .
92. SINGLE interval resource display . . . . . . . . . . . . .
93. COMBINEd interval DETAIL display . . . . . . . . . . . .
94. COMBINEd interval resource display . . . . . . . . . . . .
95. Navigating to the shared-DASD degradation display (part 1 of 4). .
96. Navigating to the shared-DASD degradation display (part 2 of 4). .
97. Navigating to the shared-DASD degradation display (part 3 of 4). .
98. Navigating to the shared-DASD degradation display (part 4 of 4). .
99. PF key defaults . . . . . . . . . . . . . . . . . . . .
100. Productivity index calculation formula . . . . . . . . . . . .
101. Sample PAGESEP output . . . . . . . . . . . . . . . .
102. Exception Analysis Keyword Syntax . . . . . . . . . . . .
103. Exception processing on combined performance group data . . .
104. REPORTIF keyword syntax . . . . . . . . . . . . . . .
105. DASD resource display (MVS/ESA) . . . . . . . . . . . .
106. Typical system resource display (with logical partitioning) . . . .
107. Sample CPU resource display (with logical partitioning) . . . . .
108. Sample I/O queuing resource display (3090 processor) . . . . .
109. Automatic analysis example (part 1 of 2) . . . . . . . . . .
x
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 74
. 75
. 76
. 77
. 78
. 79
. 80
. 82
. 83
. 84
. 84
. 85
. 86
. 87
. 88
. 89
. 90
. 98
. 98
. 101
. 101
. 103
. 105
. 109
. 109
. 110
. 118
. 119
. 120
. 121
. 122
. 123
. 123
. 124
. 126
. 126
. 127
. 127
. 128
. 129
. 130
. 131
. 132
. 132
. 133
. 138
. 141
. 142
. 143
. 153
. 155
. 159
. 159
. 160
. 161
. 164
110. Automatic analysis example (part 2 of 2) . . . . . . . .
111. SINGLE interval DETAIL display . . . . . . . . . . . .
112. Overview of Using the Workload Profiling Facility . . . . .
113. PROFILE Command Format . . . . . . . . . . . . .
114. Sample elapsed time plot . . . . . . . . . . . . . .
115. Sample Profile Report display . . . . . . . . . . . .
116. Sample Profile Report with device detail . . . . . . . . .
117. DISPLAY Command Format . . . . . . . . . . . . .
118. Sample Profile/Workload Comparison report . . . . . . .
119. Sample Profile/Workload Comparison report with device detail
120. Sample Profile/Workload Comparison Summary report . . .
121. COMPARE command syntax . . . . . . . . . . . . .
122. SETP command format . . . . . . . . . . . . . . .
123. OBTAIN and SET command syntax . . . . . . . . . .
124. Output file for OBTAIN using FORMAT(INT) . . . . . . .
125. Output file for OBTAIN using FORMAT(COL) . . . . . . .
126. Output file for OBTAIN using FORMAT(PC) . . . . . . .
127. Sample Job to Allocate a Graphics Replay File . . . . . .
128. Sample Job to Allocate SAS File . . . . . . . . . . .
129. The graphic replay procedure. . . . . . . . . . . . .
130. The Channel Service Rate report . . . . . . . . . . .
131. The DASD Utilization report . . . . . . . . . . . . .
132. The Batch Program Resource Consumption report . . . . .
133. EXTRACT command syntax . . . . . . . . . . . . .
134. COMPEXT command format . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
164
166
168
169
176
179
180
180
184
185
186
187
190
193
198
199
199
203
211
213
224
225
227
228
229
Figures
xi
xii
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Tables
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
Default reporter options . . . . . . . . . . . . . .
DISPLAY command keyword summary . . . . . . . . .
Performance group keywords . . . . . . . . . . . .
Batch workload keywords. . . . . . . . . . . . . .
Workload display options keywords . . . . . . . . . .
Detail wait reasons . . . . . . . . . . . . . . . .
Default summary wait categories . . . . . . . . . . .
Detail wait reason codes . . . . . . . . . . . . . .
Resource display keywords . . . . . . . . . . . . .
Date and time keywords . . . . . . . . . . . . . .
Keywords valid with SET . . . . . . . . . . . . . .
Fast-Path Navigation Commands . . . . . . . . . .
Batch report page-separator keywords . . . . . . . .
Detail wait reason codes . . . . . . . . . . . . .
Default summary wait categories . . . . . . . . . .
I/O and enqueue operands. . . . . . . . . . . . .
Enqueue major name defaults . . . . . . . . . . .
Workload keywords . . . . . . . . . . . . . . .
Date and time Keywords . . . . . . . . . . . . .
User Macros . . . . . . . . . . . . . . . . . .
EXTRACT command keyword summary . . . . . . . .
Resource Tables for the OBTAIN and EXTRACT Commands
Cache resource activity . . . . . . . . . . . . . .
Channel resource activity . . . . . . . . . . . . .
CPU resource activity . . . . . . . . . . . . . .
DASD device resource activity . . . . . . . . . . .
SRM domain resource activity . . . . . . . . . . .
General system resource activity . . . . . . . . . .
I/O queuing activity . . . . . . . . . . . . . . .
Paging and Storage Resource Activity . . . . . . . .
Page data set Resource Activity . . . . . . . . . . .
Performance group resource activity . . . . . . . . .
Swap data set resource activity . . . . . . . . . . .
SRM happy value statistics . . . . . . . . . . . .
Swapping activity . . . . . . . . . . . . . . . .
Swap reasons . . . . . . . . . . . . . . . . .
VLF class statistics . . . . . . . . . . . . . . .
Workload tables for the OBTAIN and EXTRACT commands
Performance group degradation data . . . . . . . . .
Batch job degradation data . . . . . . . . . . . .
Started task degradation data. . . . . . . . . . . .
TSO user session degradation data . . . . . . . . .
Workload tables for the COMPEXT command. . . . . .
Performance group comparison data . . . . . . . . .
Batch job comparison data . . . . . . . . . . . . .
Started task comparison data . . . . . . . . . . . .
TSO user session comparison data . . . . . . . . .
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 20
. 26
. 31
. 36
. 46
. 49
. 53
. 58
. 63
. 92
. 96
. 128
. 142
. 146
. 148
. 150
. 150
. 171
. 171
. 214
. 228
. 233
. 236
. 239
. 242
. 247
. 250
. 252
. 256
. 260
. 263
. 264
. 267
. 269
. 271
. 274
. 278
. 280
. 281
. 284
. 286
. 288
. 290
. 291
. 293
. 294
. 296
xiii
xiv
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
About this guide
EPILOG for MVS™ is an autonomous component of the IBM® Tivoli® OMEGAMON® XE on z/OS® product.
In OMEGAMON XE on z/OS, however, the historical data collection and reporting functions that EPILOG
provides have been superceded. OMEGAMON XE monitoring agents collect their own data, and use the
historical reporting facilities of the Tivoli Enterprise Portal and Tivoli Common Reporting.
This book is a combination of two previously released books, EPILOG for MVS Beginning User's Guide
and EPILOG for MVS Advanced User's Guide, that have not been available since the purchase of the
former Candle Corporation by IBM. The book is being released as a convenience for long-time users who
prefer to use the older product. The book has been rebranded, but the information it contains has not been
updated: it contains information that is no longer valid and describes features that are no longer supported.
For example, history collection for Sysplexes using the EPILOG collector is no longer supported. Any
references to Candle Support are obsolete. For support questions, contact IBM Software Support.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
xv
xvi
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Part 1. Basic reporting features
Part 1 describes the basic reporting features of EPILOG for MVS. The introduction provides a product
overview and a discussion of the EPILOG for MVS approach to performance management. The rest of
Part 1 explains how to use the EPILOG reporter, including the various types of reports and the use of the
DISPLAY command.
Special topics, such as advanced reporting options, the Workload Profile Facility, exception filtering,
exporting historical data, and reporting with SAS graphics are documented in Part 2.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
1
2
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 1. Introduction to EPILOG
EPILOG is a performance management tool for the MVS operating system. In contrast to a real time
performance monitor such as OMEGAMON for MVS, EPILOG collects pertinent performance data, saves it
in a historical database, and lets you make online and batch queries against it.
Prevention without the guesswork
Your performance goal is high service levels. When setting out to prevent the recurrence of system
problems, you cannot rely on educated guesswork or hunches. And you cannot wait for the problem to
happen again to begin diagnosis. You need to document the severity, duration, cause, and affected
workloads so you can take preventive measures.
EPILOG for MVS, an historical performance monitor, provides tabular displays and graphic reports that
show you the cause of performance problems. EPILOG helps you safeguard service levels by revealing
inter-workload interference and pinpointing resource contention, so you can fix problems before they have
a detrimental effect.
EPILOG removes the mystery from tuning and capacity planning. It helps everyone on the staff to
understand the relationships between workload response, throughput, and the MVS tuning parameters. A
variety of graphic reports (which can be generated through SAS graphic reporting tools) help convey the
results of your work to management.
Answers to your performance questions
Once EPILOG is installed, you can begin to ask such questions as:
v Were there periods yesterday when TSO was not meeting its service objective? If so, why? Is this a
one-time problem, or a condition that has been developing for some time?
v The nightly batch job stream, which must complete before CICS® can be brought up in the morning, ran
late three times this week; as a result, the CICS availability commitment was not met. What is the
problem? Does this indicate a trend towards poor performance in an important workload?
v A new string of DASD is arriving soon. Would TSO performance be best helped by using them for data
set storage, or should some of them be added to the paging configuration?
v Applications Development is making extensive changes to a major system. How does the performance
of the modified version compare with that of the earlier production version?
v The technical staff believes a cached paging subsystem is needed to maintain online service. Is there a
report to show paging analysis to management? Once the device is installed, does it make a
difference? Is there a report to illustrate the difference to management?
Questions like these generally break down into a few categories:
v Performance reporting and profiling
v Retrospective problem resolution
v Evaluation of change in tuning and configuration
v Application system performance analysis
v Early detection of potential problems
v Management reporting and capacity planning
EPILOG can solve problems in these categories easily and economically for two key reasons:
v EPILOG represents a focused approach to performance issues. Rather than collecting every possible
system variable, it allows you to select only those which are useful and relevant to performance
management.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
3
v EPILOG is not just a report generator; it is the key to a tuning methodology. Its design enables you to
use a logical tuning approach, which is a structured, easy-to-follow procedure for investigating
performance problems. (The logical tuning approach is described in Chapter 2, “Problem solving,” on
page 11.)
Who can use EPILOG for MVS
EPILOG provides a flexible performance management system for data center personnel. It lets each
department investigate their own performance issues.
Performance Analysts can identify and correct significant bottlenecks or spot trends towards poor
performance among workloads that they feel are critical. For example, the PAYROLL job, which normally
takes two hours, ran for four hours yesterday. With an EPILOG performance degradation display, it is easy
to tell why the job ran as long as it did. And with the Workload Profiling Facility of EPILOG, an analyst can
compare the last run of this job with a profile of its past performance.
Systems programmers can understand why jobs are running slower than usual, why CPUs are being
under-utilized, and which disk packs are poorly configured. EPILOG can compare a system before and
after a tuning or configuration change.
Application programmers can get feedback about how efficient their programs are and how they can
achieve better performance.
Production control personnel can recognize jobs that should be rescheduled so they will not interfere with
other work.
Computer operations managers can understand why jobs are running slower than usual and why TSO
users are experiencing poor response time. Exception reports can spotlight incidents of specific interest to
data center management.
Capacity planners can examine the results of stress tests and more accurately estimate the benefits of a
particular hardware upgrade.
Product overview
EPILOG for MVS has three major components: the EPILOG collector, the EPILOG reporter, and the
Workload Profiling Facility.
The following sections describe each of these components.
The EPILOG collector
The EPILOG collector collects several different types of performance related information from RMF™ and
from selected control blocks in the MVS operating system.
v Service levels (elapsed times and response times)
v Resource utilization data
v Degradation data (by batch job and performance group)
The collector is a started task that should be started automatically after each IPL. (There can only be one
EPILOG collector active in an MVS system at a time.)
For resources and performance groups, the collector writes data to the EPILOG datastore at the end of
each collection interval (typically, the standard RMF interval of 15 minutes to 1 hour). For batch jobs, data
is written at the end of every job step. You can also define collection filters which allow only a desired
subset of the data to be written to the EPILOG datastore. Data in the EPILOG datastore is available to the
reporter immediately after it has been collected.
4
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
In addition to writing data to the EPILOG datastore, you can also have the collector write the data it
collects from MVS control blocks to an SMF record type. This data, however, is not affected by your
collection filters. (For more details about collection filters, see the IBM Tivoli OMEGAMON XE on z/OS:
Planning and Configuration Guide.)
Service level and resource utilization data are a subset of the data normally collected by RMF, such as
TSO response times, CPU and DASD device utilizations, and paging rates. Most of this data is extracted
from the records that RMF writes to SMF. The remainder of the resource data (such as SRM MPL
adjustment values and real storage utilization by performance group) is collected by EPILOG itself.
Degradation data is used to show the execution states and the bottlenecks that make up transaction
response time.
The EPILOG collector tolerates dynamic reconfiguration, which allows the addition and deletion of I/O
devices on an active MVS system. When a system I/O reconfiguration occurs, an informational message is
written to the collector log stating that the collector detected an I/O configuration change. The
OMEGAMON Subsystem must be installed in order for EPILOG to gather information on devices defined
as dynamic. For further information about the OMEGAMON Subsystem, see the IBM Tivoli OMEGAMON
XE on z/OS: Planning and Configuration Guide.
The EPILOG reporter
You can use the EPILOG reporter component either as a batch report generator or for interactive inquiry
under TSO. A powerful set of commands lets you request information for a spectrum of times and
workloads. These commands can answer a wide variety of questions. You can request data by specific
time periods, combine and average collection intervals together, select intervals by exceptional conditions,
or supply report titles. The EPILOG reporter can typically:
v Display all collection intervals from last week during prime shift when the TSO first period response time
was greater than two seconds.
v Show all the class A jobs from yesterday that ran for more than 30 minutes.
v Display all batch jobs this week that waited over five minutes for a tape mount.
v Show the intervals during which DASD RESERVEs were a major contributor to poor TSO response
time.
v Compare the degradation profile of averaged TSO transactions from Monday to Tuesday.
v Show average DASD device activity last Friday from 9:00 AM to 12:00 PM.
The EPILOG Workload Profiling Facility
In a data processing environment, performance problems usually develop slowly. They first appear as
exceptions to the normal response time or elapsed run time of a particular workload. Without an easy way
to detect such emerging problems, performance management becomes a difficult task.
Using the Workload Profiling Facility (WPF), you can compare the current performance of particular
workloads to profiles of their average past performance. This enables you to detect differences in the
degradation profile of a workload over time and to spot performance exceptions.
Profiles are generated with a simple batch job. According to your specifications, WPF selects and
averages data for selected workloads from the EPILOG datastore, and saves the results in the Profile
datastore (PRDS). Afterwards, you can access this data with the EPILOG reporter to generate online or
batch reports for profiles and profile/workload comparisons.
These exception management features make it possible for you to:
v Spot exception trends in system performance and throughput of workloads. You are then able to
anticipate performance problems and correct them before they become too severe.
Chapter 1. Introduction to EPILOG
5
v Get a bird’s-eye view of resource consumption and bottlenecks for critical workloads over a period of
time. With this information, you can tune the system to accommodate those workloads that must run
within certain time constraints.
v Establish realistic initial service objectives.
Bottleneck analysis
Traditionally, performance management systems have collected and displayed resource utilization data; for
example, they might tell you that a particular device is 50 percent busy. You would then compare this
value against your installation’s rules of thumb to determine if the workload is performing poorly because
of high disk utilization. This decision may be difficult, since these rules of thumb may be different for
different devices, and could even change daily.
A device being 50 percent busy should not concern us, as long as the important workloads are not
suffering because of it. Bottleneck analysis is a statistical sampling technique used by the EPILOG
collector that tells you directly how much each type of workload is degraded because of each type of
bottleneck.
For every sample taken, the EPILOG collector analyzes every active address space in order to determine
its execution state. EPILOG measures approximately 40 execution states, such as using a CPU, waiting
for I/O to complete, waiting for an enqueue, waiting for tape mount, waiting for CPU, swapped out, or
waiting for a page fault to be resolved.
By sampling at a regular interval, the EPILOG collector can statistically estimate the relative impact of
each wait reason on performance for a particular workload. For example, assume that TSO response time
is two seconds and that execution states are distributed as follows:
Execution State
TSO Response Time
Using CPU
10%
Paging
50%
Unilateral swaps
40%
Total
100%
The execution state Using CPU is not degradation and accounts for 10% of TSO response time. The
remaining 90% of TSO response time is spent in nonproductive states, some of which may be tunable. We
can estimate the impact of the degradation as follows:
Paging
50% of 2.0 seconds = 1.0 second
Unilateral swaps
40% of 2.0 seconds = 0.8 seconds
The advantage of this approach is that you can quickly determine two things: (1) where to begin your
tuning effort and (2) the potential benefit of removing the bottleneck.
EPILOG presents the results of bottleneck analysis in the form of degradation data, which is an analysis of
the portion of time that a given workload was in a nonproductive state. For this reason, the EPILOG
workload analysis displays are referred to as degradation analysis displays.
Performance reporting
You can use EPILOG for bottleneck and resource reporting. EPILOG displays are available both
interactively and in batch. Under TSO, the EPILOG reporter takes advantage of the full-screen capabilities
of the 327x and 3290 terminals, providing scrolling, highlighting of important information, and extended
color support (with a 3179 or 3279 terminal).
6
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
You request EPILOG panels via simple yet powerful commands. For example,
DISPLAY SUMMARY PGN(2) YESTERDAY REPORTIF(RESPONSE(>1S)) STARTTIME(0900) ENDTIME(1700)
generates a panel with a line for each reporting interval yesterday between 9:00 AM and 5:00 PM in which
performance group 2 (TSO) response time was greater than one second. In addition, the main component
of response time is shown along with the total response time. The output might look something like
Figure 1.
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| Period: 09:45 to 16:30 on 11/11/99
|
+-----------------------------------------------------------------------------+
|DATE__START_END___MAIN_REASON(*)_TIME(-)|0S_____1______2______3______4______5|
|11/11 09:45 10:00 CPU Wait
1.21 S|***---->
.
.
.
.|
|
10:00 10:15 CPU Wait
1.53 S|*****----->
.
.
.
.|
|
10:15 10:30 CPU Wait
1.55 S|****------>
.
.
.
.|
|
10:45 11:00 ENQ MISC USR
1.41 S|****----->
.
.
.
.|
|
11:00 11:15 ENQ MISC USR
4.35 S|*******************---------->
.|
|
11:15 11:30 CPU Wait
1.02 S|***--->.
.
.
.
.|
|
11:45 12:00 CPU Wait
1.51 S|***------->
.
.
.
.|
|
12:00 12:15 Long Wait Swp
1.01 S|***--->.
.
.
.
.|
|
12:30 12:45 Long Wait Swp
1.12 S|***---->
.
.
.
.|
|
12:45 13:01 CPU Wait
1.58 S|***------->
.
.
.
.|
|
13:01 13:15 Long Wait Swp
1.38 S|****----->
.
.
.
.|
|
13:15 13:30 Long Wait Swp
1.04 S|***--->.
.
.
.
.|
|
13:30 13:45 CPU Wait
1.50 S|**------->
.
.
.
.|
|
13:45 14:00 Long Wait Swp
1.65 S|***--------> .
.
.
.|
|
14:00 14:15 Long Wait Swp
2.15 S|***----------->
.
.
.|
|
16:15 16:30 ECB/STIMER
2.85 S|*******------------> .
.
.|
|---------------------------------------|
|Average for this Workload Display 1.68 S
|
+=============================================================================+
Figure 1. Sample SUMMARY display panel
EPILOG provides two basic types of degradation displays: a summary display (shown above), which
shows the main wait reason in each interval, and a detail display, which shows all execution states and
wait reasons above the user-defined cutoff point. There are also a number of resource displays that show
resource utilization statistics. A list of reports available from EPILOG is given in “EPILOG reports” on page
8.
Profile and workload comparisons
The EPILOG Workload Profiling Facility enables you to use the reporter to compare a workload with its
profile and display a clear and detailed report of a workload’s current degradation measured against its
performance in the past. For example,
COMPARE JOB(SORTNY) TODAY PNAME
compares the degradation for the current day’s SORTNY job with its last profile.
The output from this command appears in Figure 2 on page 8.
Chapter 1. Introduction to EPILOG
7
+======================== Profile/Workload Comparison ========================+
| Job = SORTNY
Prof Job = SORTNY
Profname = (automatic)
|
| Prof Period: 00:01 to 12:03 on 10/28/99
Sysid = SYSA |
| Wkld Period: 14:55 to 14:57 on 11/11/99
Sysid = SYSA |
+-----------------------------------------------------------------------------+
|
Times
Delta Plot (in Seconds)
|
|Wait_Reason____________Profile_Workload|27.67___13.83_____0_____13.83___27.67|
|Using CPU
2.87 S
2.88 S | .
.
.
.
|> .
.
.
. |
|Waiting for CPU
1:05 M
1:33 M | .
.
.
. |+++++++++++++++> |
|Waiting for MVS Lock
5.88 S 15.59 S | .
.
.
. |++++> . . . |
|Private Page-In Wait
n/a
2.59 S | .
.
.
.
|> .
.
.
. |
|I/O QUEUED
1.17 S
n/a
| .
.
.
. <|
.
.
.
. |
|ECB Wait
5.88 S
n/a
| . . . .<--|
.
.
.
. |
|I/O ACTIVE
25.88 S
5.19 S | . .<----------|
.
.
.
. |
|Elapsed Time
1:47 M
1:59 M | .
.
.
. |++++++>. . . |
+=============================================================================+
Figure 2. Sample Profile/Workload Comparison report
In addition to profile and workload comparison reports, you can also display other types of profile reports:
the standard profile report that shows a workload’s average degradation data for a given time period, and
the summary comparison report that provides a list of those workloads which show the most deviation
from their profiles. For a detailed discussion of the Workload Profiling Facility, see Chapter 10, “Workload
Profiling Facility (WPF),” on page 167.
EPILOG reports
EPILOG performance reporting and workload profiling provide you with the following reports.
v Degradation analysis panels
– Degradation Analysis by Detail Wait Reason
– Degradation Analysis by Main Wait Reason
– Degradation Analysis by Summary Wait Reason
– Cross Performance Group Analysis
– Shared-DASD Degradation Analysis
v Resource panels
– Cache Subsystem Statistics (MVS/XA and MVS/ESA only)
– CPU Activity
– Channel Activity
– DASD Device Activity
– SRM Domains
– General System Information
– I/O Queuing (MVS/XA and MVS/ESA only)
– Paging and Storage
– Page Dataset Activity
– SRM Performance Groups
– Swap Dataset Activity
– SRM MPL Adjustment Values
– System Swap Activity
– Swap Activity By Swap Reason
– Virtual Lookaside Facility (VLF) Statistics (MVS/ESA and above)
v Workload Profile reports
– Standard Profile report
– Profile Report with Device Detail
– Standard Profile/Workload Comparison report
– Profile/Workload Comparison Report with Device Detail
– Summary Comparison report
8
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Retrospective problem resolution
One of the most common performance management tasks is to diagnose a problem that was not resolved
when it actually occurred. EPILOG uses a methodology to solve such problems, which is called the logical
tuning approach. This methodology is defined and illustrated in Chapter 2, “Problem solving,” on page 11.
Change evaluation
Another common performance management task is the evaluation of a tuning or configuration change.
Degradation analysis data from EPILOG greatly simplifies this task.
The goal of most tuning changes is to improve the processing speed of a certain workload. This requires
changes that shift the execution profile of the workload towards productive states such as using CPU and
active I/O, and away from unproductive states (namely, all other execution states). Since EPILOG
measures execution states directly, before-and-after comparisons to evaluate tuning actions are easy to
do. In fact, you can obtain before-and-after comparisons from EPILOG for any change in the configuration
(even one which is not supposed to have any effect on performance, such as the installation of a new
release of a compiler).
EPILOG allows you to capture a workload or resource profile before a change (for an average hour, day,
shift, and so on) and another profile after a change. You can look at both profiles on the same screen
under TSO for easy comparison.
Performance impact on applications
A vital but often neglected aspect of performance management is how to avoid performance problems by
being aware of how new or modified applications affect resource usage.
Modern application development tools such as application generators, fourth generation languages,
high-level database management systems with their associated query languages, and even powerful
options on traditional application implementation tools place heavy demands on system resources. These
demands can become difficult to manage. By providing information about execution states and resource
usage in a form that is easy to understand, EPILOG can help application developers spot and avoid
potential performance problems.
Chapter 1. Introduction to EPILOG
9
10
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 2. Problem solving
This chapter shows how to use EPILOG as a problem-solving tool for effective and accurate tuning when
you encounter performance slowdowns and bottlenecks. The problem-solving examples in this chapter are
based on real-life situations encountered during typical data center operations.
The logical tuning approach
EPILOG for MVS prevents future slowdowns by helping you solve both chronic and fleeting performance
problems. Monitoring workload performance is no longer a time consuming process, but a routine
procedure that employs a three-step logical tuning approach:
Service analysis
Customized summary and detail reports and graphs point out trends in
resource utilization, response, and throughput. These displays are created
interactively from the online performance database, which contains data
ranging from a few minutes to months ago.
Cause analysis
Once you’re alerted to a performance problem, either by reviewing reports
or by a user complaint, you can find the cause by narrowing the time
period and increasing the detail of the reports. This zooming feature lets
you quickly determine the impact of the problem and, when applicable, the
impacting workload. EPILOG reports and graphs pinpoint the problem and
put you on the road to a solution.
Action
The reports generated by EPILOG make the required corrective action
immediately obvious, whether it is rescheduling conflicting workloads,
changing tuning parameters, reconfiguring, or adding more hardware.
Helping your capacity planning effort
Your capacity planning methodology employs the rule of thumb (ROT) method. You observe trends of
resource consumption (CPU, channel utilization, paging) by workload and by the entire system. When
these trend lines intersect your ROT threshold, and service reports show a worsening of response time
and throughput, you request additional hardware.
Your new data center manager is not convinced that a new processor is needed. After all, CPU utilization
is only 80%. You need a clearer, more intuitive way to explain the problem and the solution.
Service analysis
Over the past four months, CPU utilization has increased from 70% to 80% during the peak hours of 9:00
AM to 11:00 AM. TSO response time has gone from 0.7 to 1.5 seconds for first period transactions. Other
online and batch workloads have also suffered. You know what the symptoms are, but you need to identify
the cause of the problem.
Cause analysis
To identify the cause of the problem, you proceed as follows:
1. You decide to create a five month trend of the components of TSO response time using the SAS
interface feature of EPILOG. To extract this data from the EPILOG datastore, you enter the following
command:
EXTRACT PERFGROUP(2) PGPERIOD(1) STARTTIME(9:00) ENDTIME(11:00) STARTDATE(9/1/98) ENDDATE(1/19/99) SUMWAIT COMBINE(2H)
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
11
This command extracts historical data for performance group 2, period 1, for the time period 9:00 AM
to 11:00 AM, from 9/1/98 to 1/19/99, showing summary wait conditions only, and combining
performance data in two hour intervals. For an explanation of the syntax used in this command, see
“Using the DISPLAY command” on page 26.
2. You plot the results using SAS code supplied by EPILOG. This procedure is explained in Chapter 12,
“Reporting with SAS graphics,” on page 201. You can also generate a tabular report without using SAS
graphics. The SAS interface procedure generates the graph in Figure 3.
TSO Response Components
Sept ’88 to Jan ’89
1.4
Seconds
1.2
1.0
0.8
0.6
0.4
0.2
SEP 88
OCT 88
NOV 88
DEC 88
JAN 89
PAGE and SWAP Wait
CPU
I/O
CPU Wait
Figure 3. SAS graphic display of TSO response components
The graph shows that CPU WAIT (the time that workloads are inactive because they are waiting for CPU)
has increased dramatically during the reporting period, although CPU (the time that workloads were
actively using CPU) was relatively constant. Since increased CPU wait time is usually caused by workload
contention for CPU time, you can now establish a connection between performance problems and CPU
resources.
Action
With the support of this graph, you explain to the data center manager that, because of the contention of
workloads for the processor, TSO response time is rapidly deteriorating. You use the analogy that traffic on
the highway slows down even before cars are bumper-to-bumper (100% utilization of the highway).
The data center manager asks for the same analysis for batch and CICS. You are ready with similar
EPILOG graphs for these workloads. Your analysis is accepted, and your manager supports the upgrade
recommendation.
Slowdowns after consolidating workloads
Due to an acquisition of another company, data processing work is being consolidated on your mainframe.
A light batch workload is to be added to a processor running only CICS. Now that the conversion has been
completed and the last increment of batch work has been added, the batch work is running satisfactorily.
But there are other problems.
12
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Service analysis
There are random periods of poor response time for the online CICS applications. You have to pinpoint
and solve this problem quickly.
Cause analysis
Your operations staff has been too busy with the consolidation to use your real time monitor to diagnose
the problem as it has been occurring. You decide to use EPILOG to determine why your CICS applications
have slowed down. You proceed as follows.
1. To get an overview of the problem, display the CICS performance group (performance group 7). To do
this, enter the EPILOG command
DISPLAY PERFGROUP(7) SUMMARY COMBINE(1H) STARTTIME(8:00) ENDTIME(10:00) STARTDATE(1/9/99) ENDDATE(1/13/99)
This command displays historical data for performance group 7, for the time period 8:00 AM to 10:00
AM during the date period 1/9/99 to 1/13/99, combines the data into hourly intervals, and shows only
the most important wait reason in each interval display. In other words, this command shows hourly
system views of activity in the CICS region.
You note that as workloads are being added, page-in waits have become increasingly significant during
the consolidation (see Figure 4).
+=============================================================================+
| PERF. GROUP =
7
SYMBOLIC NAME = MIS CIMS
|
| FROM
08:00 ON 01/09/99 TO 10:00 ON 01/13/99
|
+-----------------------------------------------------------------------------+
| DATE__START_END____MAIN REASON(*)_TIME(-) | OM_____5____10____15____20____25|
| 01/09 08:00 09:00 PAGE-IN WAIT
4.5 M | *----->.
.
.
.
. |
|
09:00 10:00 PAGE-IN WAIT
4.1 M | *----> .
.
.
.
. |
| 01/10 08:00 09:00 PAGE-IN WAIT
5.2 M | **----->
.
.
.
. |
|
09:00 10:00 PAGE-IN WAIT
5.3 M | **----->
.
.
.
. |
| 01/11 08:00 09:00 PAGE-IN WAIT
5.4 M | **----->
.
.
.
. |
|
09:00 10:00 PAGE-IN WAIT
5.3 M | **----->
.
.
.
. |
| 01/12 08:00 09:00 PAGE-IN WAIT
7.3 M | ***------> .
.
.
. |
|
09:00 10:00 PAGE-IN WAIT
7.4 M | ***------> .
.
.
. |
| 01/13 08:00 09:00 PAGE-IN WAIT
14.1 M | ***********------>.
.
. |
|
09:00 10:00 PAGE-IN WAIT
14.0 M | ***********------>.
.
. |
|
.
.
.
.
.
|
|
.
.
.
.
.
|
|
.
.
.
.
.
|
+=============================================================================+
Figure 4. System views in the CICS region
2. You decide the problem may be a real-storage constraint leading to increased paging. To verify this,
enter DISPLAY RSRM, including the date and time parameters shown in step 1. This command generates
a set of resource panels for the period from 1/9/99 to 1/13/99.
The display in Figure 5 on page 14 shows one of the panels generated by the command, showing that
the Unreferenced Interval Count (UIC) is 2. The OBSERVED AVG... line displays this information. By
comparing this panel with other panels generated for the indicated time period, you determine that the
UIC has deteriorated over the time period in question.
Chapter 2. Problem solving
13
+===========================SRM MPL ADJUSTMENT VALUES===========================+
| FROM
08:00 TO 09:00 ON 01/13/99
ELAP = 58:18 M
SYSB |
| COMBINATION OF 4 INTERVALS
|
+-------------------------------------------------------------------------------+
|
PAGE DMAND PAGE % OF FRAMES FIXED |
| SRM PARAMETERS
CPU UIC ASMQ FAULT PGING DELAY
BELOW 16M
TOTAL |
|
--- --- ---- ----- ----- ----- ----------- ----- |
| HIGH THRESHOLD
105
4 1000
85
80 1000
88
72 |
|
|
|
OBSERVED MAX
84
10
346
500
577
288
34
26 |
|
OBSERVED AVG
74
2
232
426
478
214
29
22 |
|
OBSERVED MIN
56
1
101
402
455
135
28
22 |
|
|
| LOW THRESHOLD
101
2 1000
70
65 1000
82
66 |
+===============================================================================+
Figure 5. EPILOG display of SRM MPL adjustment values
3. You decide to view the historical rise in real storage usage by batch (performance group 1). To do this,
you use the command DISPLAY RPGN (1), including the date and time parameters shown in step 1 on
page 13. This command generates a set of resource panels for the period from 1/9/99 to 1/13/99. By
looking at the value of total real storage in the panel for 1/13/99 (as shown under the column TOTAL
STG K in Figure 6), and comparing it with the value of real storage in the panel for 1/9/99, you
determine that total real storage has increased by 15 megabytes.
+========================SRM PERFORMANCE GROUPS=================================+
|FROM
08:00 TO 09:00 ON 01/13/99
ELAP = 1:00 H
SYSB|
|COMBINATION OF 4 INTERVALS
|
+-------------------------------------------------------------------------------+
|
SU'S
AVG
SWAPS
AVG AVG
TOTAL
EXCPS|
|PGN PGP
/SEC
RESP /TRX /TRX
%RES WKST * IN = STG K SRB% TCB% /SEC|
|-------------------------------------------------------------------------------|
| 1
1
2303 5:00M 12.5
3.6
99.4 1559 12.5 19487
1.0 10.5 57.1|
|
SUM
2303 5:00M 12.5
3.6
99.4 1473 12.5 19487
1.0 10.8 58.9|
+===============================================================================+
Figure 6. Performance group statistics for batch jobs (PGN1)
Action
You decide to fence the working set of your CICS region using the PWSS parameter in the IPS and adjust
the MAX MPL level of the batch workload downward. You can use EPILOG to monitor the results.
Meanwhile, you tell management that a hardware upgrade (paging packs or memory) may be necessary,
pending the results of your tuning efforts.
Getting rid of the 4GL blues
The new Vice President of Sales has directed the sales staff to produce monthly financial reports and to
have them on his desk by the sixth day of each month. But the final numbers aren’t available to the sales
staff until the last minute, and the reports have been late for the last two months.
Service analysis
The sales department has started using SAS to produce these reports, so on the sixth day of the month
they load the system with SAS programs to generate the reports. The last time they ran their applications,
they complained of poor response time and missed deadlines. When the vice president wanted to know
why the reports were late, they blamed the data center. Your manager has promised Sales that you will
prevent the recurrence of the perceived poor service.
14
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Cause analysis
You know the TSO performance group and logon IDs of the sales staff and the date and time of their
problem. You examine the response components of their performance group (PGN120) online by signing
on to EPILOG and entering
DISPLAY PERFGROUP(120) STARTTIME(9:00) ENDTIME(10:00) STARTDATE(3/6/99) ENDDATE(3/6/99) SUMMARY
This command produces a display that shows the largest component of response time for performance
group 120 for each 15-minute interval between 9:00 and 10:00 on March 6, 1999. In this case, the
response time component is queuing for I/O on disk pack MP220B. The line with a start time of 09:45
displays this information.
+============================================================================+
| PERF. GROUP =
120
SYMBOLIC NAME = SALETSO
|
| FROM
09:00 TO 10:00 ON 03/06/99
|
+----------------------------------------------------------------------------+
|DATE__START_END____MAIN REASON(*)_TIME(-) | OS____10____20____30____40____50|
|03/06 09:00 09:15 QD83 MP220B
20.34 S | *****-------->
.
.
. |
|
09:15 09:30 QD83 MP220B
30.37 S | *********---------->
.
. |
|
09:30 09:45 QD83 MP220B
20.12 S | ***---------->
.
.
. |
D
09:45 10:00 QD83 MP220B
20.89 S | ****--------->
.
.
. |
|
|
| ---------------------------------------|
|AVG FOR THIS WORKLOAD DISPLAY
22.93 S
|
+============================================================================+
Figure 7. Largest response time components for performance group 120
To get more detailed information on the 9:45 to 10:00 interval, you enter a D on the corresponding display
line. (This feature is explained in Chapter 8, “Fast-path navigation,” on page 125.) EPILOG displays a
detailed degradation analysis panel for this interval.
+===============================================================================+
|PERFORMANCE GROUP = 120
SYMBOLIC NAME = SAL ETSO
|
|FROM
09:45 TO 10:00 ON 03/06/99
ELAP = 14.527 SYSB
|
+-------------------------------------------------------------------------------+
|WAIT REASON __________TIME _____%____|0___1___2___3___4___5___6___7___8___9___0|
|USING CPU
3.37S 17 |------->.
.
.
.
.
.
.
.
.|
VDISK MP220B D83 QUE
5.32S 26 |----------> .
.
.
.
.
.
.
.|
|DISK MP220B D83 ACT
4.64S 23 |---------> .
.
.
.
.
.
.
.|
|WAITING FOR CPU
3.27S 16 |------->.
.
.
.
.
.
.
.
.|
|ECB WAIT
1.69S 08 |--->.
.
.
.
.
.
.
.
.
.|
|MISC USR ENQUEUE
.39S 02 |-> .
.
.
.
.
.
.
.
.
.|
|PRIVATE PAGE-IN WAIT
.11S 01 |
.
.
.
.
.
.
.
.
.
.|
|AVERAGE TRANS TIME
20.12S
|
164 MVS TRANSACTIONS ENDED
|
+===============================================================================+
Figure 8. Detail degradation analysis for performance group 120
This panel shows that waiting for I/O on volume MP220B is the cause of the slow response time. You
wonder if this was caused by another workload or simply contention among the users in performance
group 120.
First, you produce a resource display of DASD unit 13A by entering V on the display line Disk MP220B...
The result is as follows:
Chapter 2. Problem solving
15
+=============================DASD DEVICE ACTIVITY==============================+
| FROM:
09:45 TO 10:00 ON 03/06/99
ELAP = 15:06 M
SYSB |
+-------------------------------------------------------------------------------+
|DEV VOLUME
I/O
_______ TIME IN MILLISECONDS _________ % DEV OPEN |
| #
SERIAL LCU
RATE
TOTAL = IOSQ+PEND+CONN+DISC CUB DB UTIL DSNS |
|--- ------ ------------- ---- ---- ---- --- --- ---- ---- |
|D83 MP220B 023
25.6
53
32
0
4 17
55.6 117.8 |
+===============================================================================+
Figure 9. DASD device activity display for volume MP220B
This display shows that I/O contention for the volume is causing queuing (IOSQ = 32msec/IO). The
queuing is caused by too many tasks accessing the volume (average number of data sets opened during
the interval is 117.8).
To determine if other performance groups are also doing I/O on D83 and causing contention, you request
a cross-performance group display by entering DISPLAY XPG.
+=================================/ /===========================================+
| PERIOD: 09:45 TO 10:00 ON 03/0/ /
ELAP = 14.52 M SYSB |
+--------------------------------\ \--------------------------------------------+
|XPG
CPU ECS
ECB
CPW \ \
PAG
RDY TMP WTO LON QD83 D83|
|*SYSTEM* 6.3
13.0
.1
\ \
1.4
|
|REGBATCH 11.1
.5
.9 29.4
\ \
4.2 92.6 100. 100.
|
|DEV TSO
4.6
1.1
2.3
\ \
4.7
.8
|
|NCCF
1.0 7.5
6.8
2.0
\ \ 6.1
|
|STC TASK 14.3 34.4 40.7 28.9
/ / 18.4
.2
93.3
|
|VTAM
3.0 2.5
.0
.3
/ /
7.0
|
|JES2
2.8 2.7
2.6
/ /
.8
|
|MONITORS 9.0 8.5
.6
.0
/ /
3.6
|
|OMVTAM
6.0
3.7
.6 / /
1.4
|
|W.P. TSO
.3
.1
.1 / /
.8
|
|AUX A.S.
.9 5.1 19.5
\ \
.8
|
|Q BATCH
.4
.5 \ \
3.6
|
|DELTAMON
.1 3.8
1.3
\ \
.2
|
|CIMS TRX 12.7
.8
.0
4.8
\ \
3.0
.5
|
|MSA
1.5 2.1
.2
3.8
\ \
3.0
|
|CLMENU
5.2 13.9
.8
\ \ 6.7
6.6
|
|PROD DB2 2.3 6.6
6.7
5.5
/ / 3.0
|
|SALETSO
2.3
.6
1.6
/ /
5.0
.2
100. 100.|
|R&D W.P.
.0
/ /
.2
|
|R&D Q.M.
.4
.0
/ /
.8
|
|R&D BASE
.5
.2
/ /
1.4
|
|TS L.A. 5.7
1.0
3.9 / /
15.1
.5
|
|TS S.F.
.6
.1
.3 \ \
.5
|
|TS N.Y. 1.6
4.1
.2
1.1 \ \
.5
|
|MIS CIMS 1.1
.1
.6
\ \
1.6
|
|MIS FIN. 1.7
.2
2.5
\ \
2.5
|
|DC SYSP.
.3
.0
.2
\ \
|
|DC OPS.
2.0
.1
4.1
\ \ 2.2
|
|PERF172
.5
2.9
.0
.8
/ / 3.9
.5
|
|PERF176
6.7
.0
/ /
.5
|
+===================================/ /=========================================+
Figure 10. Cross performance group display
The cross-performance group (XPG) display shows that SALETSO accounts for approximately 100% of the
I/O activity to MP220B. The I/O activity for device MP220B is shown under the column heading D83, which
is the device address. In addition, you know that most of the application data sets have been placed on
the same volume, because Figure 9 shows that there are a large number of open data sets allocated to
volume MP220B.
Action
You decide to redistribute the data sets across disks or purchase additional volumes. You can use
EPILOG DASD device activity report of all volumes to determine candidate volumes.
16
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Dealing with shared-DASD contention
You manage a data center with a multi-system environnment (SYSP and SYST) where shared-DASD is
utilized. Users report poor response time on an electronic mail system running under CICS.
Service analysis
By the time you investigate the problem, it has already disappeared, suggesting that the CICS region was
in contention for resources with another workload.
Cause analysis
The CICS region suffering the response time problem is a started task named CICSPROD. To determine
where this region was spending its time during the problem period, you issue the command
DISPLAY STC(CICSPROD) TODAY STIME (10) ETIME (12) COMBINE PMIN(0)
This command produces the following display.
+=============================================================================+
| Started Task = CICSPROD
JES Number = 7033
STC Steps = 1 / 1 |
| From
10:00 to 12:00 on 07/27/99
Elap =
2:00 H SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
2.95 S
.0|.
.
.
.
.
.
.
.
.
.
.|
|ECB Wait (w/ STIMER)
1:19 H 66.1|------------============>>> .
.
.
.|
VDisk CICS01 D8F Que 35.40 M 29.9|------------>
.
.
.
.
.
.
.|
|Waiting for CPU
2:22 M
2.0|-> .
.
.
.
.
.
.
.
.
.|
|Disk CICS01 D8F Act 1:56 M
1.6|.
.
.
.
.
.
.
.
.
.
.|
|Disk PROD17 D85 Act 31.92 S
.4|.
.
.
.
.
.
.
.
.
.
.|
|Disk COM002 D8A Act 4.91 S
.0|.
.
.
.
.
.
.
.
.
.
.|
|ECB Wait
2.45 S
.0|.
.
.
.
.
.
.
.
.
.
.|
|Waiting for MVS Lock
2.45 S
.0|.
.
.
.
.
.
.
.
.
.
.|
|STC Elapsed Time
1:59 H
|
|Productivity Index =
0%
|
+=============================================================================+
Figure 11. EPILOG display for started task CICSPROD
The display shows that there is excessive queued I/O for the workload against device CICS01. The Disk
CICS01 D8F Que... line displays this information. You know that this device is usually dedicated to the
electronic mail system, so you decide to look for the cause of the performance degradation in the area of
DASD contention.
You navigate directly to the DASD resource display (RDAS) by entering a V next to to the I/O wait reason
line for CICS01. (This feature is described in “Navigating in a multisystem environment” on page 131.) The
following panel is displayed.
+========================== DASD Device Activity ==============================+
| From:
10:00 to 12:00 on 07/27/99
Elap = 30:59 M
|
| Sysid:
SYSP SYST
|
| Combination and Merge of 16 Intervals
|
+------------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev
Avg|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Allcs|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- ------|
JCICS01 SYSP D8F 041 12.6
930
851
69
3
7
.4 69.9 12.8
7.0|
|
SYST D8F 008 23.1
42
1
7 18 16
.1 6.4 78.0
7.4|
+==============================================================================+
Figure 12. DASD device activity display for device CICS01
Chapter 2. Problem solving
17
Normally, there is no activity on volume CICS01 from any other system but SYSP. The 69 milliseconds of
PEND time on volume CICS01 in system SYSP indicates that you may be dealing with cross-system
contention for this volume. This contention may also account for the unusually large IOSQ time.
To identify the workloads on system SYST that may have been contending for volume CICS01, enter J
next to the volume identifier on the DASD Device Activity panel. The result is shown in the following panel.
+=============================================================================+
| Volume = CICS01
Period = 10:00 to 12:00 on 07/27/99
|
+-----------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system ACTIVE I-O
|
+-----------------------------------------------------------------------------+
|Workload_Jes #_Sysid_Time_______%__|0___1___2___3___4___5___6___7___8___9___0|
|LOOPTEST J7166 SYST
1:37 H 98.0 |-------------============>>>>>>>>>>>>>>>.|
|CICSPROD S7033 SYSP
1:56 M
2.0 |-> .
.
.
.
.
.
.
.
.
.|
+-----------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system QUEUED I-O
|
+-----------------------------------------------------------------------------+
|Workload_Jes #_Sysid_Time_______%__|0___1___2___3___4___5___6___7___8___9___0|
|CICSPROD S7033 SYSP 35:40 M 98.6 |-------------============>>>>>>>>>>>>>>>.|
|LOOPTEST J7166 SYST 31.22 S
1.4 |
.
.
.
.
.
.
.
.
.
.|
+=============================================================================+
Figure 13. Shared-DASD usage in a multi-system environment
The Shared-DASD Degradation Analysis panel shows the relative usage of the specified DASD device in a
multisystem environment. In this example, you expect to see CICSPROD as the only workload having any
I/O activity against CICS01 during the period when the CICS region was up.
Instead, the Shared-DASD display shows that batch job LOOPTEST has 98% of the observed Active I/O
against volume CICS01. The first LOOPTEST line displays this information. You identify the contending
workload as an analyze/scan job that was run against CICS01 from SYST.
Using EPILOG, you solved this problem in about 10 minutes. Analyzing raw RMF and SMF data could
easily have taken all day.
Action
Based on this data, you advise the DASD management group to stop running DASD maintenance during
the prime shift. You document your request with printed copies of the EPILOG displays.
18
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 3. Getting started
This chapter describes how to invoke the EPILOG reporter, documents the command syntax and
conventions, and provides an overview definition of the DISPLAY command.
Invoking the reporter
Invoke the EPILOG reporter either by using a batch job or a CLIST under TSO or ISPF. During invocation,
the EPILOG datastores are opened for reading, and a number of reporter defaults are established for the
session.
The following sections describe the various ways to invoke the reporter in batch and online, the reporter
defaults, and the index and data buffers used to open the EPILOG datastore. For instructions on using
sample EPILOG data, see “Using the sample datastore” on page 27.
Mode of operation
You can run the reporter in any of the following modes:
v ISPF split-screen
v TSO full-screen
v Batch
These modes are described in the following topics.
ISPF split-screen mode
A sample CLIST to run the reporter under ISPF is provided in rhilev.REDDATA(KEDSPF). (During
installation, this CLIST may have been moved to your CLIST library.) When the CLIST is invoked, a menu
of options is displayed. This menu is described in “Split-screen mode under ISPF” on page 20.
This mode also allows you to execute TSO commands from the EPILOG reporter.
TSO full-screen mode
If you do not have ISPF, you can run the reporter directly under TSO with the sample CLIST provided in
rhilev.REDDATA(KEPTSO) or in rhilev.REDDATA(KEDTSO), if both EPILOG and DELTAMON for MVS are
installed at your site. This facility does not allow you to use the ISPF split-screen function or the EPILOG
TSO command if you are running TSO/E Version 2.0 or higher.
When this CLIST is invoked, the reporter displays the initial entry screen, which is described in “The initial
entry screens” on page 22.
To exit the reporter, you can enter the END command on the command line at any time. STOP and QUIT
are valid synonyms for END.
Batch mode
A sample jobstream that invokes the reporter in batch is distributed in rhilev.REDDATA(KEPPROC) or in
rhilev.REDDATA(KEDPROC), if both EPILOG and DELTAMON are installed at your site. This jobstream
performs the necessary allocations and reads the input commands from a data set pointed to by the
EPSYSIN DD statement. Output is directed to a data set pointed to by the EPREPT DD statement, and
messages are logged to a data set pointed to by the EPMSG DD statement. Because JCL is system-level
dependent, we recommend that you read the KEPPROC member and tailor it to your own needs.
The EPILOG product tape includes standard report jobs that can be run in batch mode. These report jobs
are described in Chapter 7, “Service level management,” on page 115.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
19
Reporter option defaults
A number of default options are set when you invoke the reporter. The sample CLISTs and batch job used
to invoke the reporter point to the rhilev.REDPARM partitioned data set, which contains six members.
These members contain the default options, and are described in Table 1.
Table 1. Default reporter options
Member name
Option
@DEFAULT
Contains the default A-matrix, which associates selected resource panels with detail
wait reasons. The A-matrix is used to navigate from workload degradation displays
to associated resource panels. It is also used to associate resources with wait
reasons with the AUTO keyword and the RESOURCE command.
$DEFAULT
Contains PF key definitions.
DEFAULTP
Groups detailed wait reasons into three categories: productive, nonproductive, and
idle. These categories are used to calculate the productivity index displayed on
workload degradation screens.
DEFAULTS
Groups detailed wait reasons into summary wait categories. The categories are
used to format the SUMWAIT displays.
EDSLIST
Defines the initial datastore list; that is, the list of EPILOG datastores that are
allocated and opened when the reporter is initialized. The active datastore list can
be modified during a reporter session by the DATASTOR command.
LSRBUFFS
Contains specification for the number of VSAM index and data buffers.
You can modify these defaults or create and invoke your own members of rhilev.REDPARM. Instructions
for setting up and invoking your own set of options are described in “Setting reporter options” on page
137.
Datastore buffers
By default, the reporter opens the EPILOG datastore with three data buffers and two index buffers. Under
most circumstances, this should be sufficient. However, if I/O waits are degrading reporter performance,
you can reduce the number of physical I/Os required by increasing the number of buffers.You do so with
the REOPEN command, which accepts the following keywords:
BUFND
number of data buffers
BUFNI
number of index buffers
For example, to increase the number of data and index buffers to 5 and 7, respectively, enter:
REOPEN BUFND(5) BUFNI(7)
The reporter closes the EPILOG datastore and reopens it with the new number of buffers. This command
affects the EPILOG datastore only.
Split-screen mode under ISPF
EPILOG can be run under ISPF (Version 2 and above) in split-screen mode. A CLIST, KEDSPF, has been
provided for this support.
To use the reporter under ISPF, follow these steps:
1. Invoke the KEDSPF CLIST.
The first display is the Primary Option Menu for starting your EPILOG session.
20
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
------------------------- CANDLE PRIMARY OPTION MENU ------------------------OPTION ===>
0
1
X
PARMS
COMMAND
EXIT
- Change user session parms
- Begin COMMAND driven session
- Return to ISPF
Figure 14. Primary Option Menu
Important: The first time that each user invokes the CLIST from a TSO user ID after product
installation, it is necessary to enter option 0 to initialize the PF key assignments. On subsequent
invocations, you can go directly to step 4.
2. Enter option 0.
The following screen is displayed:
--------------------------- CANDLE USER PARMS MENU --------------------------COMMAND===>
1. ASSIGN PF KEYS FOR SPLIT AND SWAP FUNCTIONS
(NOTE: THESE PF KEYS WILL NOT BE AVAILABLE TO THE REPORTER)
SPLIT PFK # ===> 02
SWAP PFK # ====> 09
PRESS ENTER WHEN CORRECT
At this point, you can either press Enter (so that PF2 and PF9 are reserved for split and swap,
respectively), or overtype them with different PF key designations. If you change these defaults, you
should also modify the EPILOG PF key definitions to reflect the change, using the PFK command. This
command is described in “The PFK command” on page 139.
3. Press Enter.
You are returned to the Primary Options Menu.
4. To initiate a reporter session, select option 1.
Once you are in the reporter, the split function behaves exactly as it does in other ISPF functions. In
addition, you should keep in mind that reporter displays require a minimum of eight display lines. If you
attempt to use fewer than eight, the reporter displays the message
INSUF SCRN SIZE
in the upper-right corner. If you receive this message, do the following:
1. Split the screen again to allow EPILOG at least eight display lines.
2. Re-enter the command.
You should avoid splitting the screen to a smaller size while viewing EPILOG logical displays that exceed
the size of your screen (for example, a large RDAS display panel). Doing so may cause a portion of the
display to become hidden. However, if you do split the screen again, you can recover the lost portion of
the display in one of these two ways:
v Return the screen to full-screen mode to reveal the portion of the display that became hidden.
v Re-enter the command and begin to view the display from the beginning.
Display scrolling in EPILOG is discussed in “Scrolling through reporter output” on page 102.
To exit the reporter, enter END, STOP, or QUIT on the command line. This returns you to the split-screen
menu shown above. Enter X on the OPTION line to complete reporter termination.
Chapter 3. Getting started
21
The initial entry screens
When you invoke the reporter, the first display shows you three kinds of information:
v A copyright notice necessary to protect the proprietary nature of the products
v A list of currently installed background monitors
v A message about the status of your data stores
The data store status message is displayed in the following format.
+==============================================================================+
|
|
|
Summary Data Store Statistics
|
|
|
|
Collection
|
|
Date and Time Range
Type Util SYSID Status
Reason
|
|
------------------------------- ---- ---- ---- --------- --------- |
|
1 TDNN01.EDS.SYSA.DEC98
|
|
12/01/98 09:00 - 12/31/98 23:59 EDS
64% SYSA AVAILABLE
|
|
|
|
2 TDNN01.EDS.SYSA.JAN99
|
|
01/01/99 00:01 - 01/31/99 23:00 EDS
73% SYSA SWITCHED MONTH
|
|
|
|
3 TDNN01.EDS.SYSA.FEB99
|
|
02/01/99 00:02 - 02/17/99 08:15 EDS
21% SYSA ACTIVE
|
|
|
|
4 TDNN01.EDS.SYSF.F1
|
|
01/01/99 00:01 - 01/25/99 07:00 EDS 100% SYSF SWITCHED FULL
|
|
|
|
5 TDNN01.EDS.SYSF.F2
|
|
01/25/99 07:16 - 02/17/99 08:15 EDS
62% SYSF ACTIVE
|
|
|
|
6 TDNN01.EDS.SYSF.F3
|
|
............................... .... .... .... UNAVAIL
ALLOC
|
|
|
|
TDNN01.DEV.PRDS
|
|
PRDS 47% SYSA
|
|
|
|
TDNN01.DEV.PRDS
|
|
PRDS 12% SYSF
|
|
|
|
TDNN01.DEV.DDS
|
|
DDS
88% SYSA
|
|
|
|
|
|
Press ENTER to Continue
|
|
|
+==============================================================================+
This display shows the status of the data stores that are currently installed. This information is especially
helpful if you are using EPILOG to monitor multiple systems. For a description of this status display, see
“Displaying datastore information” on page 111. For a description of multiple datastore reporting, see
“Reporting in a multiple datastore environment” on page 106.
When you press Enter, the initial entry screen is displayed.
22
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
CANDLE CORP.
12/11/99 10:54 Mode: PAGE
D1
PFK DISPLAY
CMD==>
*****************************************************************************
P R O G R A M
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
F U N C T I O N
K E Y S
'HELP'
'CONTROL MODE'
'BACK'
'CONTROL RECALL'
'CONTROL TITLE'
'DIS RALL TODAY COMBINE STIME(9) ETIME(17)'
'CONTROL SCROLL UP'
'CONTROL SCROLL DOWN'
'CONTROL LOG'
'DIS PGN(2) TODAY SUMMARY COMBINE(1H)'
'DIS PGN(2) TODAY RIF(RESP(>1S)) AUTO'
'PFK'
'HELP'
'CONTROL MODE'
'BACK'
'CONTROL RECALL'
'CONTROL TITLE'
'DIS RALL TODAY COMBINE STIME(9) ETIME(17)'
'CONTROL SCROLL UP'
'CONTROL SCROLL DOWN'
'CONTROL LOG'
'DIS PGN(2) TODAY SUMMARY COMBINE(1H)'
'DIS PGN(2) TODAY RIF(RESP(>1S)) AUTO'
'PFK'
Figure 15. PF key defaults
The top line of the display header shows you the following information:
v The current date in mm/dd/yy format.
v The current time in 24-hour format.
v The scrolling mode (PAGE) and the number of pages in the display (1 of 2). For details about display
scrolling, see “Scrolling through reporter output” on page 102.
v The current display type (PFK DISPLAY).
The second line of the header is known as the command line. This is where you enter reporter commands.
The rest of the display shows you the default PF Key settings. (If you also have DELTAMON for MVS
installed, you will notice that some of the commands associated with the PF keys are different. For an
explanation of these commands, refer to the DELTAMON Reference Manual.) You can reset the PF keys
at any time during the session by using PF12.
Reporter commands
You can enter any of the reporter commands from this display screen. Valid commands are:
DISPLAY
Displays degradation and resource information. This is the most frequently used
command. It is described in “Using the DISPLAY command” on page 26.
END
Ends the reporter session. STOP and QUIT are valid synonyms for END.
HELP
Invokes the online help facility. This command is described in “Online help” on page 25.
INQUIRE
Displays information about your data storage subsystem. This command is described in
“Displaying datastore information” on page 111.
PRODUCTS
Displays a list of IBM background monitors currently installed on your system. This
command is described in “The PRODUCTS command” on page 114.
Chapter 3. Getting started
23
REOPEN
Closes and reopens the datastores with a specified number of index and data buffers
(default = 3 data buffers and 2 index buffers). This command is described in “Datastore
buffers” on page 20.
RESOURCE
Associates system resources with performance degradation. This command is described in
“The RESOURCE command” on page 130.
SET
Sets defaults for the DISPLAY command. This command is described in “Setting display
defaults” on page 95.
TSO
Invokes TSO CLISTs and command processors during an EPILOG reporter session. This
command is described in “Invoking TSO commands” on page 25.
The following reporter commands are described in Part 2, “Advanced reporting features,” on page 135.
A-matrix Commands
Four commands maintain user-defined A-matrix tables: CREATEM, DELETEM, LISTM,
and REPLACEM.
COMPARE
Compares recent or current workload degradation to a profile of past degradation.
PAGESEP
Formats page separators between batch reports. This command is only used in batch
mode.
PFK
Assigns functions to PF keys.
SETP
Sets criteria for selecting the profile used with the COMPARE command.
Command syntax and conventions
You control the format and content of the reporter displays and reports using a set of commands and
keywords.Keywords as a rule do not have to be entered in any particular order. (Specific cases where
entry order affects command execution are pointed out in the keyword descriptions.) The following
sections describe some general rules for entering commands, keywords, and operands.
Delimiters
Keywords are separated from other keywords and operands by one or more blank spaces, equal signs,
commas, or parentheses. If you use parentheses, they must be balanced, and an input line may not begin
with a right parenthesis. (In other words, EPILOG uses essentially the same syntax conventions as TSO.)
For consistency, in this manual we use blank spaces to separate keywords and parentheses to separate
operands. We suggest that you adopt a similar convention.
Operands that contain blanks or other delimiters must be enclosed in single quotes.Two consecutive single
quotes are treated as a single quote in the operand. Also, an operand enclosed in quotes may not be
continued from one line to another.
The continuation character
To continue a command line, end it with a hyphen (-) preceded by a blank space and press Enter. A new
line is generated for you to continue the command. You can use as many lines as you like to enter the
command. Nothing will be executed until the reporter encounters a line that does not contain a hyphen. If
you leave text to the right of the hyphen, that text is treated as a comment.
Comment lines
During batch processing, it is often useful to include comments with the commands. To do so, place an
asterisk (*) in column 1 of the command line. The comment is written out to the message data set pointed
to by the EPMSG DD statement.
24
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Short and long forms
Most commands and keywords have both a long and short form. In addition, most keywords (but not
commands) will also be recognized if you enter an unambiguous abbreviation of the long form (for
example, COMBI instead of COMBINE). Commands require either the long or short form, and are not
recognized in any other abbreviated form.
Using masks for operands
You can use the asterisk (*) as a wildcard character in certain operands, such as job names, device
addresses, and volume serial numbers. The asterisk is interpreted as a variable character; any character
found in the position of the asterisk meets the selection criteria. For example:
DISPLAY JOB(IE*100)
This command selects any job beginning with IE and having 100 in positions 4 to 6.
When the asterisk is in the last position, it is propagated to fill out the operand. For example,
DISPLAY JOB(IEB*)
selects any job that begins with IEB, regardless of length.
You can use the asterisk in the last position and specify the length by enclosing the argument in quotes.
For example,
DISPLAY JOB('IEB**')
selects any job beginning with IEB that is five characters long.
Online help
Online help is available for all EPILOG commands and topics. When you enter HELP or H without any
operands, you will get a list of valid commands and topic names. When you enter HELP commandname or
HELP topic, the requested HELP information will be displayed.
You can enter the HELP command at any time, even while you have the output of a DISPLAY command
on the screen. When you are finished with the help facility, press Enter to return to the interrupted display.
You can also request help by entering the command or topic name on the input line and pressing PF1 or
PF13 instead of Enter. This activates the specified help facility.
EPILOG comes with an online help PDS which must be allocated to ddname EPHELP. (The sample CLIST
distributed with the product takes care of this for you.)
Invoking TSO commands
You can invoke TSO CLISTs and command processors directly from EPILOG by using the TSO command.
For example, to issue a list catalog command while running EPILOG, enter TSO LISTC.
Note: Since the EPILOG reporter does not run with APF authorization, it is not possible to run
APF-authorized programs underneath it.
The TSO command is not valid when EPILOG for MVS is running as a batch program.
If you want to use this option, and you are running TSO/E Version 2.0 or higher, the EPILOG reporter
must be run in split-screen ISPF mode, as described in “Mode of operation” on page 19.
Chapter 3. Getting started
25
Using the DISPLAY command
The DISPLAY command allows you to report information from the EPILOG and Profile datastores, and can
be used to display information online or to generate batch reports. This manual describes reporting from
the EPILOG datastore. Chapter 10, “Workload Profiling Facility (WPF),” on page 167 describes how to
display profiles from the Profile datastore.
The following isFigure 16 shows the syntax of the DISPLAY command.
DISPLAY
{workload | resource} [time period] [combining factor] [SYSID([MERGE,] cccc, ...)] [display options]
Figure 16. DISPLAY command syntax
The only required keywords for this command syntax are those that specify a workload or resource.
However, if you omit the time period, the reporter displays the entire span of data from the EPILOG
datastore. This usually results in long displays that do not focus on useful information.
The keyword types used with the DISPLAY command are summarized in Table 2, and are described more
fully in the following pages.
Table 2. DISPLAY command keyword summary
Keyword Type
Description
Workload
The workload keywords display degradation data by performance group, batch job,
started task, or TSO user session. These keywords are described in Chapter 4,
“Displaying workload information,” on page 29. You can display degradation data in
either detail or summary format, as described in “Detail and summary displays (DETAIL,
SUMMARY)” on page 46.
The workload utilization keywords display shared-DASD usage in a multi-system
environment. These keywords are described in “Shared-DASD analysis” on page 44.
Resource
The resource keywords display information about the types of system resources that
are monitored by EPILOG. These keywords are described in Chapter 5, “Displaying
resource information,” on page 63.
Time period
Since the EPILOG datastore is an historical database, you can select data from any
time period that the EPILOG collector was running. The date/time keywords for doing so
are described in “Setting dates and times” on page 91.
Combining factor
Typically, resource and performance group collection is synchronized with RMF intervals
at your installation. By default, each display panel usually corresponds to an RMF
interval. However, this interval is not always a convenient or desirable unit for reporting
data; it may provide more detail than is really useful. For reporting purposes, you can
combine RMF-based intervals into a more reasonable interval size, such as 1 hour or 1
day.
Batch jobs, started tasks, and TSO user sessions can either be collected at RMF-based
intervals, or at termination time (end-of-job, end-of-step, end-of-session, or task
termination). The way you combine this data depends on how it was collected. If data
was collected at RMF-based intervals, the data is automatically combined unless you
specify the INTERVAL keyword. If data was collected at termination time, you can use
the COMBINE keyword to combine steps together, or combine together selected job
runs, TSO sessions, or started tasks.
The COMBINE keyword is described in “Combining time intervals” on page 97.
26
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 2. DISPLAY command keyword summary (continued)
Keyword Type
Description
SYSID[MERGE]
If you are reporting on more than one system, DISPLAY generates one panel per
reporting interval for each system.To restrict the data display to a particular system,
include the SYSID(ccc) keyword in the DISPLAY, where ccc is the SMF ID of the
system data you want to display. To restrict the data display to several systems, use the
format SYSID (aaaa, bbbb, cccc,...), where aaaa is the SMF ID of the first system to be
displayed, bbbb is the SMF ID of the second system to be displayed, and so on.
The MERGE keyword allows you to combine DASD device activity from multiple
systems into a cross-system, shared-DASD display. The use of this keyword is
described in “Reporting in a multiple datastore environment” on page 106.
Display options
There are a number of display formatting keywords to help you set up your display
screen or batch report. Certain formatting keywords apply to workload displays only.
These keywords are described in “Workload display options” on page 46. Other
keywords allow you to format either workload or resource displays. These are described
under their own headings (for example, “Setting up a color terminal” on page 104).
For detailed instructions on using the DISPLAY command to display workload information, see Chapter 4,
“Displaying workload information,” on page 29. For instructions on displaying resource information, see
Chapter 5, “Displaying resource information,” on page 63.
There are other ways you can use the DISPLAY command. For example, the PROFNAME keyword can
be used to display a workload profile, and the keywords REPORTIF and SELECTIF can be used in
workload and resource exception filtering. In addition, there are other ways you can process historical
data. The OBTAIN command allows you to export EPILOG data to a sequential file; the SAS interface
uses the EXTRACT and COMPEXT commands to write EPILOG data to SAS data sets for graphic
reporting. These functions and keywords are described Part 2, “Advanced reporting features,” on page
135.
Using the sample datastore
Since EPILOG is an historical monitor, it will take one or two weeks for your collector to gather enough
data to generate a realistic range of historical reports. If you want to try out the reports and keyword
features illustrated in this manual, you can use the two sample EPILOG datastores (EDS) that are
provided on the product tape.
To access the sample EPILOG datastores, follow these steps:
1. Verify that each sample EDS has been loaded into an EPILOG datastore. This procedure is explained
in the IBM Tivoli OMEGAMON XE on z/OS: Planning and Configuration Guide.
2. Verify the name of each sample EPILOG datastore in your installation.
3. At any point in a reporter session, enter the command
DATASTOR USE EDS(edsname1,edsname2)
where edsname is the name of the VSAM data set containing the sample EPILOG datastore.
This procedure renders the sample datastores accessible for the remainder of the reporter session. If you
want to access other EPILOG datastores during the same reporter session, you must reallocate the active
datastores. For further information about allocating and de-allocating EPILOG datastores during a reporter
session, see “The DATASTOR command” on page 107.
Chapter 3. Getting started
27
28
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 4. Displaying workload information
The DISPLAY command allows you to display information about the performance of a particular workload.
Workloads are defined as:
v Performance groups
v Batch jobs
v Started tasks
v TSO users
A sample workload display panel is shown in Figure 17.
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 14:00 To 14:15 On 07/31/99
Elap = 14:51 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.10 S
5.8|--> .
.
.
.
.
.
.
.
.
.|
|Long Wait
0.26 S 15.9|------> .
.
.
.
.
.
.
.
.|
|STIMER Wait
0.24 S 14.6|-----> .
.
.
.
.
.
.
.
.|
|Waiting For CPU
0.23 S 14.1|-----> .
.
.
.
.
.
.
.
.|
|Detected Wait
0.22 S 13.5|-----> .
.
.
.
.
.
.
.
.|
|ECB Wait
0.14 S
8.3|--->.
.
.
.
.
.
.
.
.
.|
|Average Trans Time
1.64 S
2142 MVS Transactions Ended
|
|Productivity Index
25%
|
+=============================================================================+
Figure 17. Sample workload display
This display was invoked by entering the command
DISPLAY PGN(2) YDAY STIME(9) ETIME(17)
which generates one display panel for each 15-minute SMF interval. For a description of the DISPLAY
command syntax, see “Using the DISPLAY command” on page 26.
The workload display consists of three parts.
v The screen header provides such information as:
– The workload name and symbolic name for performance groups
– The time period for which the data is displayed
– The length of the reporting interval
– The SMF ID of the system
v The body of the display shows you the execution states and wait reasons that contributed to the
response time. The first line is always Using CPU. Although this is not a wait reason, it provides a useful
figure for comparison. Following the first line are the main wait reasons. (These wait reasons are
described in Table 6 on page 49.)
v The last two lines sum up the execution states that are shown on the display. Of these two, the first line
differs slightly for the different types of workloads. For performance groups, it shows you the average
transaction time and the number of transactions that ended during the period. For batch jobs, started
tasks, and TSO users, it shows you the elapsed time.
The last line of the display shows you the productivity index, which is an indication of how severely the
workload is being degraded. (For more information on how the productivity index is derived, see
Table 39 on page 281.) This figure is a general indicator of how much time a workload spends doing
productive work (such as using CPU or active I/O), as opposed to time spent waiting for resources
(such as waiting for CPU or queued I/O). The productivity index may vary widely from installation to
installation, and at various times of the day. It is, however, characteristic of a workload at a given level
of performance. You should take note of this value during times of acceptable performance, and
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
29
compare it to values during poor performance periods. By default, the productivity index is not included
in any workload displays. To get it to display, you must add the PRODIDX keyword to the display
command, or issue the command SET PRODIDX.
In addition, there are a number of default options in effect for this display:
v The sample workload display in Figure 17 on page 29 shows performance degradation for a
performance group. You can also show similar data for batch jobs, started tasks, and TSO user
sessions. The keywords that select these types of workloads are described in “Batch job, started task,
and TSO user degradation” on page 35.
v Performance groups may be defined in the IPS with up to eight performance periods. By default, the
degradation data for all periods is averaged together into the display. You can, however, display data for
individual periods with the PGPERIOD keyword, as described in “Performance period analysis” on page
33.
v By default, performance group displays show average times per transaction for each wait reason, and
an average transaction time at the bottom of the display. You can display total times in these places
with the TOTAL keyword, as described in “Average and total performance group figures” on page 54.
v By default, a separate display panel is generated for each 15-minute interval. (This period is based
upon the RMF interval in effect when the data was collected.) You can combine data into larger
reporting intervals with the COMBINE keyword, as described in “Combining time intervals” on page 97.
v By default, workload degradation displays show a list of detail wait reasons that comprise the response
time (for performance groups) or the elapsed time (for batch jobs, started tasks, and TSO sessions).
You can alter the wait reason list in two ways:
– You can show only the main wait reason for each interval with the SUMMARY keyword, as described
in “Detail and summary displays (DETAIL, SUMMARY)” on page 46.
– You can display degradation by summary wait categories instead of detail wait reasons with the
SUMWAIT keyword, as described in “Summary wait categories (SUMWAIT)” on page 51.
v Only wait reasons that account for more than 5% of the total wait time are included in the display. You
can adjust this percentage with the PLOTMIN keyword, as described in “Limiting the display by wait
reason severity (PLOTMIN)” on page 54.
v By default, terminal input waits are not included on the display, since they are not generally considered
to be related to system degradation. You can, however, show them on a display with the TWAITON
keyword, as described in “Terminal wait swaps (TWAITON)” on page 60.
v The scale of the right side of the display is determined by the most severe wait reason on the display.
You can, however, set this scale yourself with the MAXSCALE keyword, as described in “Setting the
display scale (MAXSCALE)” on page 55.
The following sections describe the keywords used to display workload degradation. Following the
descriptions of the various types of degradation displays, there is a section on how to set various workload
display options.
System-wide degradation
To display workload degradation across the entire system, simply specify the SYSTEM keyword.
DISPLAY SYSTEM YDAY STIME(9) ETIME(17)
This command results in a set of panels for the reporting period. One of the resulting display panels might
look like this:
30
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+=============================================================================+
| System Wide
|
| From 14:00 To 14:15 On 07/31/99
Elap = 14:51 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.04 S
1.4|.
.
.
.
.
.
.
.
.
.
.|
|ECB Wait (W/ STIMER)
1.37 S 46.1|------------======> .
.
.
.
.
.|
|ECB Wait
0.55 S 18.5|------->.
.
.
.
.
.
.
.
.|
|STIMER Wait
0.27 S
8.9|--->.
.
.
.
.
.
.
.
.
.|
|Detected Wait
0.19 S
6.3|--> .
.
.
.
.
.
.
.
.
.|
|Waiting For CPU
0.17 S
5.6|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
2.98 S
2699 MVS Transactions Ended
|
|Productivity Index
14%
|
+=============================================================================+
Figure 18. System-wide degradation display
The SYSTEM display shows you a general picture of transactions running under MVS. The degradation
figures and percentages are derived by dividing total counts by the average number of address spaces
running during the interval.
For example, the Using CPU figure of 1.4% means that the average or typical MVS transaction spent 1.4%
of its time using CPU.
Notice the notation in the display header indicating that systemwide degradation is being displayed.
Performance group degradation
The PERFGROUP (or PGN) keyword allows you to select one or more performance groups for
degradation analysis. In addition, there are several options for refining your performance group selection:
v If you have defined symbolic names for certain performance groups during collector startup, you can
refer to those performance groups by their symbolic names.
v You can combine multiple control performance groups into a single report group by specifying more
than one control performance group number with the PERFGROUP keyword.
v For performance groups defined with multiple periods, you can display degradation data for one or more
performance periods. (By default, all performance periods are combined together into the display.)
v You can display system-wide degradation across performance groups using the XPG keyword.
These keywords are summarized in Table 3.
Table 3. Performance group keywords
Keyword
Short
Description
Operand
PERFGROUP
PGN
Displays data for performance groups as
indicated by one or more performance group
numbers.
0 to 9999
PGPERIOD
PGP
Displays data for one or more performance
periods. Always used with the PGN keyword;
invalid with the SYM keyword.
One or more performance periods.
SYMBOLIC
SYM
Displays data for a performance group by
user-defined symbolic name.
1 to 8 characters
Note: Asterisks (*) may be used as wildcard characters for this operand. The use of asterisks as wildcard characters
is described in “Using masks for operands” on page 25.
XPG
Displays systemwide degradation across
performance groups
Chapter 4. Displaying workload information
31
To select a specific performance group, include the performance group number as an operand with the
PERFGROUP keyword. For example, the command
DISPLAY PGN(2) YDAY STIME(15) ETIME(17)
might result in a display that includes a panel similar to Figure 19.
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 15:16 To 15:30 On 10/01/99
Elap = 13:44 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.10 S
3.6|-> .
.
.
.
.
.
.
.
.
.|
|Waiting For CPU
0.95 S 36.0|------------==> .
.
.
.
.
.
.|
|Terminal Output Wait
0.41 S 15.3|------> .
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.20 S
7.5|--->.
.
.
.
.
.
.
.
.
.|
|SRM Delay (MPL)
0.18 S
6.8|--> .
.
.
.
.
.
.
.
.
.|
|Private Page-In Wait
0.17 S
6.5|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
2.65 S
2347 MVS Transactions Ended
|
|Productivity Index
6%
|
+=============================================================================+
Figure 19. Performance group degradation display
Multiple performance groups
At times you may want to combine data from several performance groups and treat them as a single
workload. (For example, your installation may have several TSO control performance groups defined to
separate out various TSO users, but for certain purposes you may want to see TSO reported as a single
group.)
You can do so by combining several control performance groups in the DISPLAY command to be treated
as one report performance group. For example, suppose you have three TSO performance groups defined
at your installation (2, 4 and 12). These three groups may be combined by specifying them all as operands
to the PGN keyword. For example, the command
DISPLAY PGN (2,4,12) YDAY STIME(15) ETIME(17)
might result in the display shown in Figure 20:
+=============================================================================+
| Performance Groups = (2,4,12)
|
| From 15:16 To 15:30 On 10/01/99
Elap = 13:44 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.04 S
1.4|.
.
.
.
.
.
.
.
.
.
.|
|ECB Wait
0.89 S 33.4|------------=> .
.
.
.
.
.
.|
|Waiting For CPU
0.45 S 16.8|------> .
.
.
.
.
.
.
.
.|
|ECB Wait (W/ STIMER)
0.40 S 15.1|------> .
.
.
.
.
.
.
.
.|
|STIMER Wait
0.28 S 10.4|---->
.
.
.
.
.
.
.
.
.|
|Terminal Output Wait
0.16 S
5.9|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
2.65 S
2348 MVS Transactions Ended
|
|Productivity Index
15%
|
+=============================================================================+
Figure 20. Multiple performance group degradation display
The command produces the same results as if a report performance group containing these three control
groups had been defined in the ICS and specified in the DISPLAY command.
32
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Symbolic performance groups
If you assigned symbolic names to performance groups during collector startup, you can use the
SYMBOLIC (or SYM) keyword to substitute the symbolic name. Symbolic performance group names are
assigned during customization. In the following example, the user has assigned the symbolic names
PRODTSO and REGBATCH to two different performance groups.
DISPLAY SYMBOLIC(PRODTSO)
DISPLAY SYM(REGBATCH)
Note that symbolic performance group names cannot be used with the PGPERIOD keyword.
Performance period analysis
Performance groups are often defined with multiple periods to help manage workloads in the same
performance group that have different processing requirements. In particular, first period TSO is often of
interest because it is generally used as a critical measure of overall TSO performance.
You request performance period information by adding the PGPERIOD keyword (PGP for short) to the
performance group identifier (PGN). (PGPERIOD cannot be used with symbolic performance group
names; that is, with the SYMBOLIC keyword.)
For example, to display period 1 information about performance group 2, enter:
DISPLAY PGN(2) PGP(1)
The resulting display might look something like Figure 21.
+=============================================================================+
| Performance Group = (2)
Period = (1)
|
| From 09:00 To 09:15 On 08/15/99
Elap = 13:44 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.02 S
3.9|-> .
.
.
.
.
.
.
.
.
.|
|Terminal Output Wait
0.19 S 30.0|------------>
.
.
.
.
.
.
.|
|Waiting For CPU
0.12 S 18.5|------->.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.09 S 14.1|-----> .
.
.
.
.
.
.
.
.|
|Private Page-In Wait
0.04 S
5.9|--> .
.
.
.
.
.
.
.
.
.|
|Average Response Time 0.67 S
25230 MVS Transactions Ended
|
|Productivity Index
5%
|
+=============================================================================+
Figure 21. Performance period degradation display
Notice the notation in the screen header indicating that period 1 information is being displayed. The period
operand used with the PGPERIOD keyword allows you to specify more than one period. If you enter
multiple periods, the data is combined together into a single display panel.
For example, to specify periods 1 and 2 of performance group 2, enter:
DISPLAY PGN(2) PGP(1,2)
Note that the PGPERIOD keyword is also valid with the RPGN keyword, which displays resource
utilization information for performance groups.
Cross-performance group analysis
The cross-performance group display shows how various execution states are distributed across
performance groups. To get the display, simply use the XPG keyword as the workload type.
Cross-performance group displays are available for single intervals only; the COMBINE, SUMMARY, and
DETAIL keywords are invalid when used with XPG.
Chapter 4. Displaying workload information
33
For example, the command
DISPLAY XPG YDAY STIME(11) ETIME(17)
might result in the display shown in Figure 22.
+=============================================================================+
| Period: 11:45 to 12:00 on 08/23/99
Elap = 15:00 M A083 |
+-----------------------------------------------------------------------------+
|XPG
CPU DET CPW TIN LON 140 736 A9A EXC QA9A 157 Q4A3 Q736 4A3 |
|*SYSTEM* 1.4 43.4 5.0
2.0
|
|REGBATCH 65.5 4.7 44.6
100. 6.2 13.6 100. 0.9 100. 99.5 100. 70. 100.|
|TSO
26.3 51.9 6.3 100.
.2
99.1
.5
|
|PERF5
3.9
33.3
86.2
20.
|
|PERF9
2.9
10.8
91.6
.2
10.
|
+=============================================================================+
Figure 22. Cross-performance group degradation display
The format of the display is different from the degradation displays described thus far. Across the top, the
display shows the 14 top execution states, ranked by the number of times an address space was found in
that state. The abbreviations for each execution state (except for I/O and enqueues) are summarized in
Table 8 on page 58. Active I/O on a device is represented by just the device address (for example, 736).
Queued and DASD RESERVE execution states are shown by prefixing the device address with a Q or an
R (for example, Q736 or R736). Enqueue execution states are indicated by abbreviations for the enqueue
major names, which are described in the EPILOG for MVS Command Summary.
On the left side, the display lists performance groups active during the interval in ascending performance
group number order. (If a symbolic name has been defined for a performance group, the symbolic name is
displayed. In this example, SYSTEM is the symbolic name for performance group 0, REGBATCH for
performance group 1, and TSO for performance group 2.)
Each column shows a system-wide view of a given execution state, broken down by performance group.
For example, column 1 always shows a breakdown of the execution state CPU Utilization. In Figure 22,
this column shows that when EPILOG found the CPU being utilized, 65.5 percent of the time it was being
used by the REGBATCH performance group, 26.3 percent of the time it was being used by the TSO
performance group, and so on. Each column adds to 100%, except perhaps for rounding errors.
This format gives you insight into inter-performance group contention, that is, how the activity of one
performance group affects another performance group. To get a feel for this, let’s look at the two columns
that describe activity on device 736. The column labeled 736 shows active I/Os to that device, and the
column labeled Q736 shows queued I/O. You can see that when EPILOG for MVS found the device active,
86.2 percent of the time it was being used by performance group 5. When the device was found waiting,
70 percent of the time it was REGBATCH that was waiting. At this point, you might want to look at a
degradation display for REGBATCH. If waiting for device 736 was a major degradation factor for
REGBATCH, you could conclude that performance group 5 was causing a good deal of that.
Your installation’s performance group definitions will influence the way this information is interpreted. For
the purpose of the example, say that performance group 5 consists of CICS users, and REGBATCH
consists of overnight batch jobs that were submitted that day. This changes the interpretation radically.
Now, we will see performance group 5 as the preferred group, and will view any contention from
REGBATCH as undesirable. In the Q736 column, you can see that when EPILOG for MVS found I/O
queued for device 736, 20 percent of the time it was performance group 5, and that when the device was
found active, 13.6 percent of the time it was being used by REGBATCH. If queued I/O to device 736 was
a significant source of degradation to performance group 5, you can see that the REGBATCH performance
group is causing some of that. Most of the activity on device 736, however, is being done by performance
group 5 itself, which means that address spaces within performance group 5 are competing for that
device. This is a good example of self contention within a performance group.
34
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
By adding the SUMWAIT keyword to the command, you can display summary wait categories instead of
detailed wait reasons. A typical SUMWAIT XPG display is shown in Figure 23.
+-----------------------------------------------------------------------------+
| From:
15:26 To 15:30 On 07/07/99
Elap =
3:59 M SYSA |
+-----------------------------------------------------------------------------+
|XPG
MISC I/O SWAP OPER CPU PAGE SRM
|
|*SYSTEM*
2.5 29.3 91.4
19.8
|
|REGBATCH
3.5 30.9
97.6 42.2 27.9
|
|DEV TSO
.0 2.4
2.2
|
|STC TASK
22.6 14.5
2.6
|
|IMS CNTL
3.8
|
|VTAM
3.7
14.1
|
|JES2
3.1 7.7
|
|PERF15
15.1
.3
|
|PERF16
7.1
.3
27.0
|
|PERF17
.5 10.2
.6
17.0 12.8 72.1
|
|AUX A.S.
15.3
|
|Q BATCH
1.0
|
|PERF26
3.2
32.3
|
|CICS TST
10.5
7.3
14.2
|
|PERF28
3.8
|
|PERF30
3.7
.3
|
+-----------------------------------------------------------------------------+
Figure 23. Cross-performance group display with summary wait categories
This display presents a more simplified picture of cross-performance group degradation.
Batch job, started task, and TSO user degradation
You can display degradation information about batch jobs, started tasks, or TSO user sessions using any
of the following selection criteria:
v Job name, started task name, or TSO user ID
v Account code
v Program name
v JES job class
If you enter multiple criteria, they are logically ANDed together, so that a batch job, started task, or TSO
session must meet all the criteria to be included in the display.
In addition, you can display data at three different levels:
1. By default, degradation data is displayed at the level of the job, started task, or TSO user session. If
you are collecting data by intervals, then a single display panel is generated for each batch job, started
task, or TSO session meeting the selection criteria that ran during the time period specified with the
time/date keywords on the command. If you are not collecting data by intervals, a separate display
panel is created for each batch job, started task, or TSO session that ended during the specified time
period.
2. The STEP keyword allows you to break down a multi-step job or started task into its component steps.
(A separate display panel is generated for each step.)
3. The INTERVAL keyword allows you to display long-running jobs, started tasks, or TSO sessions by
RMF-based intervals. (A separate display panel is generated for each interval.)
The keywords that control these options are summarized in Table 4 on page 36, and described in the
following pages.
Chapter 4. Displaying workload information
35
Table 4. Batch workload keywords
Keyword
Short
Description
Operand
Wildcard (*)
ACCOUNT
ACCT
Selects batch job or TSO session degradation
data by installation account code.
1–12 chars
Yes
If the JOB or TSO keyword is included with the
command, a single display panel is generated for
each job run or TSO session that meets the
ACCOUNT criteria.
If neither of these keywords is included, a
separate display panel is generated for each job
or session step that meets the ACCOUNT criteria.
Note: The EPILOG collector requires a certain amount of SQA to collect account, class, and program information for
various workloads. Under normal conditions, this amount is automatically calculated and allocated during collector
startup. However, one of the collector options (SQAMAX) limits the maximum amount of SQA that can be allocated
by the collector, and if this limit is set too low, collection of information for the ACCOUNT, CLASS, and PROGRAM
keywords is affected. If you use these keywords and suspect that some records are missing, you should check to
see if the SQAMAX keyword has been used in the collector options. If it has, you may want to increase the maximum
SQA available to the collector.
CLASS
CLS
Selects batch job degradation data by JES job
class.
1 char
Yes
1–8 chars
Yes
If the JOB keyword is included with the
command, a separate display panel is generated
for each job the meets the JOB and CLASS
criteria.
If JOB is not included, a separate display panel is
generated for each job step that meets the
CLASS criteria.
INTERVAL
INT
Specifies that the data should be displayed at
RMF-based intervals rather than over the time
period specified with the time/date keywords on
the command. (That is, a separate display panel
is generated for each interval of data.) If RMF is
not active on your system, data is displayed at
intervals specified with the INTERVAL keyword,
as described in theIBM Tivoli OMEGAMON XE on
z/OS: Planning and Configuration Guide.
The INTERVAL keyword helps you to monitor
degradation data for long-running batch job steps,
started tasks, and TSO sessions.
This keyword can only be used if data has been
collected at intervals. The BATCHINT, STCINT,
and TSOINT keywords in KEPOPTN control
interval-driven data collection for batch jobs,
started tasks, and TSO users, respectively. These
keywords are also described in theIBM Tivoli
OMEGAMON XE on z/OS: Planning and
Configuration Guide.
JOBNAME
36
JOB
Selects batch job degradation data by job name.
The job name can be further qualified by JES job
number.
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 4. Batch workload keywords (continued)
Keyword
Short
Description
Operand
Wildcard (*)
PROGRAM
PGM
Selects batch job, started task, or TSO session
degradation data by program name.
1–8 chars
Yes
1–8 chars
Yes
1–8 chars
Selects TSO session degradation data by TSO
user ID. The TSO user ID can be further qualified
by JES number to specify a particular session.
Yes
If the JOB, STC, or TSO keyword is included with
the command, one display panel is generated for
each job run, started task, or TSO session that
met the criteria.
If none of these keywords are included, a
separate display panel is generated for each job,
task, or session step that executed the specified
program.
STARTTSK
STC
STEP
Selects started task degradation data by task
name. The task name can be further qualified by
JES number.
Specifies that the data should be displayed for
individual steps rather than for the complete job
or task. (That is, a separate display panel is
generated for each job or task step that meets
the criteria.)
The STEP keyword may not be used for
combined displays or for workload profile
displays; that is, with COMBINE or PNAME.
TSOUSER
TSO
Note: The use of asterisks as wildcard (variable) characters is described in“Using masks for operands” on page 25.
Examples
The following examples illustrate how these keywords are used to control the various reporting options.
DISPLAY JOB(PRINT01) YESTERDAY
Displays degradation data for job PRINT01 yesterday. A separate display panel is generated for each run
of the job. If the job consists of multiple steps, all steps are combined into a single display panel.
Let’s suppose, though, that the job PRINT01 was started at 11:30 PM yesterday and didn’t finish until 1:15
AM today. Would that run of the job be included in the displayed data?
The answer to this question depends on how batch job data was being collected, and when records were
written to the EPILOG datastore. Remember, batch job data can be collected in two ways:
v If the BATCHINT keyword is enabled, batch job data is written to the EDS at RMF-based intervals.
v If the BATCHINT keyword is disabled, job data is written to the EDS only at end-of-step or at end-of-job.
The BATCHINT keyword and batch job interval collection is fully described in the IBM Tivoli OMEGAMON
XE on z/OS: Planning and Configuration Guide.
If job data is being collected at intervals and an interval record was written before midnight yesterday, then
a display panel would be generated for this run of the job.
Chapter 4. Displaying workload information
37
If job data was not being collected at intervals, then a record would not be written to the EDS until
end-of-job or end-of-step. Thus, if a step in the job ended before midnight, then a record would be written
to the EDS, and a display panel would be generated for this run of the job. However, if no records were
written to the EDS until after midnight, then the job would not be displayed, since the DISPLAY command
specified YESTERDAY.
DISPLAY STC(IMSCNTL) LASTWEEK
Displays degradation data for started task IMSCNTL last week. A separate display panel is generated for
each time the task was started and stopped. If the task consists of multiple steps, all steps are combined
into a single display panel.
The question posed in the previous example about batch jobs that extend beyond the specified time period
also applies to started tasks. Since started tasks are processed in the same way as batch jobs, the
answer is essentially the same. If started task data is being collected at intervals and one or more intervals
ended within the time period, then a display panel is generated for the task. If the task was not being
collected at intervals, then a record would not be written to the EDS until end-of-task or end-of-step.
Unless a step record was written before midnight on Sunday of last week, no display panel would be
generated for the task.
The STCINT keyword, which controls started task data collection, is described in the IBM Tivoli
OMEGAMON XE on z/OS: Planning and Configuration Guide.
DISPLAY TSO(DEV*) TODAY STIME(9) ETIME (12)
Displays degradation data for TSO users with user IDs that begin with DEV. A separate display panel is
generated for each TSO user session.
The question posed in the first example about batch jobs that extend beyond the specified time period also
applies to TSO user sessions. Since TSO sessions are processed in the same way as batch jobs, the
answer is essentially the same. If TSO session data is being collected at intervals and one or more
intervals ended within the time period, then a display panel is generated for the session. If TSO sessions
are not being collected at intervals, then a record would not be written to the EDS until end-of-session.
Thus, no display panel would be generated for a session that extended beyond the time period of the
DISPLAY command.
The TSOINT keyword, which controls TSO session data collection, is described in theIBM Tivoli
OMEGAMON XE on z/OS: Planning and Configuration Guide.
DISPLAY STC(*) PGM(TEST01) INTERVAL
Displays degradation data for all started tasks that executed program TEST01. The INTERVAL keyword
specifies that a separate display panel is generated for each interval.
DISPLAY TSO(*) ACCT(4301) INTERVAL
Displays degradation data for all TSO user sessions that ran in account group 4301. The INTERVAL
keyword specifies that a separate display panel is generated for each interval.
DISPLAY JOB(PRINT01(J2544))
Displays a single display panel for the run of job PRINT01 with JES job number 2544. Notice that the job
number follows the job name and is prefixed with a J.
DISPLAY TSO(USER01(T4806)) INTERVAL
Displays degradation data for the TSO user session of USER01 with JES number 4806. Notice that the
JES number follows the TSO user ID and is prefixed with a T. The INTERVAL keyword specifies that a
separate display panel is generated for each interval.
DISPLAY STC(IMSCNTL(S3156)) STEP
38
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Displays degradation data for the run of started task IMSCNTL with JES number 3156. Notice that the JES
number follows the task name and is prefixed with an S.
The STEP keyword specifies that a separate display panel is generated for each step, assuming that
IMSCNTL is a multi-step started task. If started task data was collected at intervals, all interval records
within the step are combined into a single display panel.
DISPLAY ACCT(4708)
Displays degradation data for all workloads that ran with account code 4708. A separate display panel is
generated for each step that meets the ACCT criteria. If data was collected at intervals, all interval records
within the step are combined into a single display panel.
DISPLAY PGM(IEB*)
Displays degradation data for all workloads that executed programs beginning with IEB. A separate display
panel is generated for each step that meets the PGM criteria. If data was collected at intervals, all interval
records within the step are combined into a single display panel.
DISPLAY CLASS(A)
Displays degradation data for all batch job steps that ran in JES job class A. A separate display panel is
generated for each step that meets the CLASS criteria. If batch job data was collected at intervals, all
interval records within the step are combined into a single display panel.
The following sections describe each of the display types.
Job-, task-, and session-level displays
This section describes the display panels that are used when you display job, task, and session level
degradation displays.
The job-level display
The command
DISPLAY JOB(PAYROLL) YDAY STIME(9) ETIME(11)
generates a display similar to the following sample panel.
+=============================================================================+
| Job = PAYROLL
JES Number = 2544
Job Steps = 3 / 6 |
| Job Class = T
Account Number = 035G
Input Queue = 48.10 S |
| From 09:12 To 09:14 On 08/01/99
Elap =
1:33 M SYSA |
+-----------------------------------------------------------------------------+
|WAIT_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.09 S
.1|.
.
.
.
.
.
.
.
.
.
.|
|Swapped With WTOR
52.65 S 59.9|------------===========>.
.
.
.
.|
|Tape Mount Pending
25.05 S 28.5|----------->.
.
.
.
.
.
.
.|
|ECB Wait
5.01 S
5.7|--> .
.
.
.
.
.
.
.
.
.|
|Tape IEFU84 522 Que 5.01 S
5.7|--> .
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
1:27 M
|
|Productivity Index
1%
|
+=============================================================================+
Figure 24. Sample job-level display
The job-level display header contains the following information:
JOB
Job name.
JES NUMBER
JES job number. If the value for this field is missing from the display, it
may be due to the fact that not enough SQA is allocated for the EPILOG
collector. (One of the collection options (SQAMAX) limits the amount of
Chapter 4. Displaying workload information
39
SQA that can be allocated during collector startup. If this value is set too
low, missing data on batch job, started task, and TSO session displays
can result. If you are consistently missing data for this field, check to see if
the SQAMAX collection option was used, and, if it was, try increasing the
maximum amount of SQA available to the collector.)
JOB STEPS
The number of job steps included in the display and the total number of
steps in the job. In this example, the expression “3 / 6” means that there
are six steps in the job, but only three of them are included in the display.
Steps can be excluded from the display for two reasons:
v The step executed so quickly that it completely escaped data collection,
or
v The step finished executing before ten samples had been collected.
(The default sampling interval is 2.3 seconds; therefore, the elapsed
time of a step must be greater than 23 seconds in order for it to be
included in the display. However, the default sampling time can be reset
during installation, and may have been modified at your installation.
Generally, a step that executes so quickly as to be excluded from the
display does not contribute any significant delay to the job.
The total number of steps in the job (in this case, 6) is dependent upon
the last step included in the display. The reporter counts the last step
included in the display as the last step in the job. Thus, if any trailing steps
are excluded for either of the above reasons, the total number of steps in
the display will be less than the actual number of steps in the job.
JOB CLASS
Execution class. (See note about SQA allocation in the description of JES
NUMBER above.)
ACCOUNT NUMBER
Account code. (See note about SQA allocation in the description of JES
NUMBER above.)
INPUT QUEUE
The time that the job spent waiting on the input queue.
date/time
The start time, end time, and date range of the job. These figures are
based upon the first and last step included in the display.
ELAP
The elapsed time of the job. Again, this figure is based on the first and last
steps included in the display.
You will notice that the last line of the display body also shows the job
elapsed time, and that the elapsed time in the body (1:27 M) is less than
the elapsed time in the header (1:33 M). The elapsed time in the display
body is determined by adding up the elapsed times of all steps included in
the display. The discrepancy in the figures indicates that either one or
more intermediate steps in the job were excluded from the display, or
some time elapsed between steps that were included in the display. The
expression “3 / 6” in the display header confirms that three intermediate
steps were excluded.
System ID
The SMF ID of the system is displayed in the lower-right corner of the
header (in this example, the ID is SYSA).
Started task-level display
The command
DISPLAY STC(SMFDUMP) YDAY STIME(7) ETIME(13) CMB(6H)
generates a display similar to the following sample panel.
40
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+=============================================================================+
| Started Task = SMFDUMP
JES Number = 3147
STC Steps = 1 / 1 |
| From 0700 to 1300 on 6/24/99
Elap =
5:49 H SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
23:02 M
6.4|--> .
.
.
.
.
.
.
.
.
.|
|Disk SMF003 360 Act
1:48 H 29.5|------------>
.
.
.
.
.
.
.|
|Disk SMF005 320 Act
2:02 H 30.7|------------>
.
.
.
.
.
.
.|
|Tape IEFU84 522 Que
15:01 M
4.8|-> .
.
.
.
.
.
.
.
.
.|
|STC Elapsed Time
5:49 H
|
|Productivity Index
70%
|
+=============================================================================+
Figure 25. Started task degradation display
You can see from the example that the started task display is very similar to the job display. Except for the
display fields that pertain only to batch jobs (job class, account number, and input queue), the fields on the
started task display are the same as those described for the batch job display in Figure 24 on page 39.
The number of STC steps in the display header indicates that SMFDUMP was a single-step started task.
The TSO session-level display
You can generate the TSO session level display with a command in the following format:
DISPLAY TSO(USER01) YDAY STIME(12) ETIME(14)
+=============================================================================+
| TSO User = USER01
JES Number = 2828
TSO Steps = 1
|
|
Acct No = 112233
|
| From 12:19 to 12:24 on 6/13/99
Elap =
5:33 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
15.07 S 4.5|-> .
.
.
.
.
.
.
.
.
.|
|ECB Wait
1:23 M 24.9|---------> .
.
.
.
.
.
.
.|
|Swap Page-In Wait
55.37 S 16.6|------> .
.
.
.
.
.
.
.
.|
|SRM Delay (MPL)
41.53 S 12.4|---->
.
.
.
.
.
.
.
.
.|
|Terminal Output Wait
27.68 S 8.3|--->.
.
.
.
.
.
.
.
.
.|
|Private Page-In Wait
27.68 S 8.3|--->.
.
.
.
.
.
.
.
.
.|
|Disk TSOLA1 D9F Act 27.68 S 8.3|--->.
.
.
.
.
.
.
.
.
.|
|Disk TSO023 331 Act 27.68 S 8.3|--->.
.
.
.
.
.
.
.
.
.|
|TSO Trans Time
10:27 M
|
|Productivity Index
22%
|
+=============================================================================+
Figure 26. TSO session degradation display
The TSO session display is very similar to the batch job and started task displays. With the exception of
job class, which pertains to batch jobs only, the display fields on the TSO session display are the same as
those described for the batch job display in Figure 24 on page 39.
Step-level displays
The step level display breaks down a job or started task into its constituent steps. A typical step level
display is shown in Figure 27 on page 42. You can generate this type of display with a command in the
format
DISPLAY TSO(USER01) YDAY STIME(13) ETIME(14) STEP
Chapter 4. Displaying workload information
41
+=============================================================================+
| Job = USER01
JES Number = 3273
Program = IMASPZAP
|
| Job Class = F
Acct No = Operations
|
| Step = S14X
Procstep = ZAPSTEP
Step Number =
2
|
| From 13:54 To 13:54 On 08/01/99
Elap = 25.73 S SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.54 S
2.1|.
.
.
.
.
.
.
.
.
.
.|
|Waiting For CPU
12.61 S 49.0|------------=======>.
.
.
.
.
.|
|Disk SYSA24 322 Act 7.56 S 29.4|----------->.
.
.
.
.
.
.
.|
|Waiting For MVS Lock
2.52 S
9.8|--->.
.
.
.
.
.
.
.
.
.|
|ECB Wait
2.52 S
9.8|--->.
.
.
.
.
.
.
.
.
.|
|Job Step Elapsed Time 25.73 S
|
|Productivity Index
30%
|
+=============================================================================+
Figure 27. Batch job step level display
In addition to the job name, JES number, class, and account code, step level displays also show the
following information:
PROGRAM
The name of the program invoked by the step or procstep. If the value for
this field is missing from the display, it may be due to the fact that not
enough SQA is allocated for the EPILOG collector. (One of the collection
options (SQAMAX) limits the amount of SQA that can be allocated during
collector startup. If this value is set too low, missing data on batch job,
started task, and TSO session displays can result. If you are consistently
missing data for this field, check to see if the SQAMAX collection option
was used, and, if it was, try increasing the maximum amount of SQA
available to the collector.)
STEP
The job or task step name.
PROCSTEP
This field is displayed only if the step is part of a PROC, and shows the
name of the PROC step.
STEP NUMBER
The step number. All EXEC statements that invoke programs are counted
as job steps, including those found within PROCs. EXEC statements that
call PROCs are not counted.
date/time
The start time, end time, and date range of the step.
ELAP
The elapsed time of the step.
The step level display also shows the SMF ID of the system in the lower right-hand corner of the header.
The step level display for started tasks is very similar to the batch job step display.
Interval-level displays
You can display degradation data at RMF-based intervals by adding the INTERVAL keyword to the
command. If RMF is not active on your system, the reporter uses the interval specified with the INTERVAL
keyword in KEPOPTN during product installation.
You cannot use the INTERVAL reporting option unless you have been collecting data at the interval level.
The BATCHINT, STCINT, and TSOINT keywords in KEPOPTN control interval-driven data collection for
batch jobs, started tasks, and TSO users, respectively. These keywords are described in the IBM Tivoli
OMEGAMON XE on z/OS: Planning and Configuration Guide.
A typical interval level display is shown in Figure 28 on page 43. You can generate this type of display with
a command in the format
DISPLAY JOB(PRINT01) YDAY STIME(8) ETIME(13) INTERVAL
42
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+=============================================================================+
| Job = PRINT01
JES Number = 3147
Program = RATECALC
|
| Job Class = T
Acct No = 41277
BATCHINT = 2
|
| Step = S14X
Procstep = CALCULAT
Step Number =
2
|
| From 08:00 To 08:30 On 08/01/99
Elap = 30:24 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
8:26 M 27.6|----------> .
.
.
.
.
.
.
.|
|Waiting For CPU
4:58 M 16.3|------> .
.
.
.
.
.
.
.
.|
|Tape
521 Act 4:04 M 13.4|-----> .
.
.
.
.
.
.
.
.|
|Control Unit 52X Que 3:46 M 12.4|---->
.
.
.
.
.
.
.
.
.|
|Tape 001724 522 Act 2:36 M
8.6|--->.
.
.
.
.
.
.
.
.
.|
|Tape Mount Pending
2:28 M
8.1|--->.
.
.
.
.
.
.
.
.
.|
|Disk BLNG28 156 Act 1:53 M
6.2|--> .
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
30:24 M
|
|Productivity Index
60%
|
+=============================================================================+
Figure 28. Batch job interval level display
As you can see, the display is very similar to the step level display. There are, however, two important
exceptions:
v The Step Number field on interval displays shows the job step number during which the interval
occurred.
v The BATCHINT field indicates the number of RMF-intervals combined to produce the display. A value of
zero (0) indicates that the display covers less than one interval.
In this example, the BATCHINT value of 2 indicates that two RMF intervals were combined to produce
the display.
Figure 29 shows a typical started task interval display.
+=============================================================================+
| Started Task = IMSCNTL
JES Number = 3147
Program = DB2UTIL
|
|
STCINT = 2
|
| Step = S14X
Procstep = COMPRESS
Step Number =
3
|
| From 10:30 To 11:00 On 08/01/99
Elap = 30:04 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
10.91 S
.6|.
.
.
.
.
.
.
.
.
.
.|
|STIMER Wait
13:33 M 45.1|------------======> .
.
.
.
.
.|
|ECB Wait (w/STIMER)
12:36 M 41.9|------------=====> .
.
.
.
.
.|
|Waiting For CPU
3:09 M 10.5|---->
.
.
.
.
.
.
.
.
.|
|STC Elapsed Time
30:04 M
|
|Productivity Index
1%
|
+=============================================================================+
Figure 29. Started task interval display
As you can see, the display is very similar to the job interval display. Figure 30 on page 44 shows a typical
TSO session interval display.
Chapter 4. Displaying workload information
43
+=============================================================================+
| TSO User = TSO001
JES Number = 1321
Program = IKJEFT01
|
| Performance Group = 2 Acct No = 41289
TSOINT = 1
|
| Step = S14X
Procstep = $TSO
Interval = 5
|
| From 10:15 To 11:00 On 08/01/99
Elap =
15 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
2:14 M 14.8|------> .
.
.
.
.
.
.
.
.|
|Waiting For CPU
6:54 M 46.0|------------======> .
.
.
.
.
.|
|Waiting For MVS Lock
1:36 M 10.7|---->
.
.
.
.
.
.
.
.
.|
|SRM Delay (MPL)
2:14 M 14.8|------> .
.
.
.
.
.
.
.
.|
|Disk TSO025 320 Act 1:14 M
8.2|--->.
.
.
.
.
.
.
.
.
.|
|TSO Trans Time
13:32 M
|
|Productivity Index
24%
|
+=============================================================================+
Figure 30. TSO session interval display
Multiple operands
You can enter multiple operands with any of these workload selection keywords. Multiple operands are
logically ORed together. For example, to select batch workloads of class Q or class F, enter:
DISPLAY CLASS(Q,F)
Shared-DASD analysis
Using EPILOG, you can monitor the usage of a given DASD device in a multisystem environment. The
JDAS and PDAS keywords generate the shared-DASD degradation display, which shows the cross-system
impact of workload utilization against DASD. Utilization is broken down into active and queued I/O
components.
Keywords
There are two keywords that can be used with the DISPLAY command. JDAS displays shared-DASD
usage by workload, including batch jobs, started tasks, and TSO users. PDAS displays shared-DASD
usage by performance group.
Syntax
Use the following syntax for the JDAS and PDAS keywords.
DISPLAY
{JDAS|PDAS} {(volser)} [time period]
where:
volser
Volume to be analyzed.
time period
Time period to be analyzed, using the date and time keywords described in “Setting dates
and times” on page 91.
Examples
Suppose that you have determined that there may be cross-system contention on volume COM002. The
command
DISPLAY JDAS(COM002) YDAY STIME(14:15) ETIME(14:30)
44
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
generates displays similar to the sample report in Figure 31.
+==============================================================================+
| Volume = COM002
Period = 02:15 to 02:30 on 02/16/99
|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system ACTIVE I-O
|
+------------------------------------------------------------------------------+
| Workload_JES #_Sysid_Time_______%__|0___1___2___3___4___5___6___7___8___9___0|
| ASMPBKUP J4661 SYSA
5:32 M 82.1 |------------============>>>>>>>>>> . .|
| $AOASMP2 S4533 SYSF 45:98 S 11.4 |-----> .
.
.
.
.
.
.
.
.|
| TDNY30
T4014 SYSA 26:22 S
6.5 |--> .
.
.
.
.
.
.
.
.
.|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system QUEUED I-O
|
+------------------------------------------------------------------------------+
| Workload_JES #_Sysid_Time_______%__|0___1___2___3___4___5___6___7___8___9___0|
| $AOASMP1 S4243 SYSF 29:64 S 66.7 |------------============>>> .
.
.
.|
| ASMPBKUP J8812 SYSA 13:94 S 31.4 |------------==> .
.
.
.
.
.
.|
| $AOASMP2 S4533 SYSF
:84 S
1.9 |>
.
.
.
.
.
.
.
.
.
.|
+==============================================================================+
Figure 31. Shared-DASD degradation display (by workload)
The Shared-DASD Degradation Display shows cross-system usage by workload of the specified DASD
device.
In the sample display, the Active I/O and Queued I/O status percentages each add up to 100%. You can
see both the heavily active and the heavily inactive workloads accessing the Volume on a comparative
basis.
If you substitute the PDAS keyword for JDAS, the following type of display is generated.
+==============================================================================+
| Volume = COM002
Period = 02:15 to 02:30 on 02/16/99
|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system ACTIVE I-O
|
+------------------------------------------------------------------------------+
| Perf Grp______Sysid_Time________%__|0___1___2___3___4___5___6___7___8___9___0|
| HOTBATCH
SYSA
5:32 M 82.1 |------------============>>>>>>>>>> . .|
| REGBATCH
SYSF 45:98 S 11.4 |-----> .
.
.
.
.
.
.
.
.|
| PERF121
SYSA 26:22 S
6.5 |--> .
.
.
.
.
.
.
.
.
.|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system QUEUED I-O
|
+------------------------------------------------------------------------------+
| Perf Grp______Sysid_Time________%__|0___1___2___3___4___5___6___7___8___9___0|
| REGBATCH
SYSF 30:48 S 68.6 |------------============>>>>.
.
.
.|
| HOTBATCH
SYSA 13:94 S 31.4 |------------==> .
.
.
.
.
.
.|
+==============================================================================+
Figure 32. Shared-DASD degradation display (by performance group)
The display uses the symbolic names defined for each performance group. If these names have not been
defined, PERFnnn is used, where nnn is the performance group number. In this example, the jobs
$AOASMP1 and $AOASMP2 have been defined in the same performance group, REGBATCH. The
Queued I/O portion of the display contains a rolled up percentage for REGBATCH for these two jobs,
which appeared separately in the previous display.
Navigation
The JDAS and PDAS keywords require that you know the volser of the device to be analyzed. You can
navigate directly to the Shared-DASD Degradation Display from a workload degradation display. This kind
of navigation is described in “Navigating in a multisystem environment” on page 131.
Chapter 4. Displaying workload information
45
Workload display options
There are a number of options available specifically for workload displays. (These options are in addition
to those available for all display types, such as COMBINE and the color terminal options.) You can:
v Display degradation data at either the detail or summary level. The default is detail, and all displays
shown thus far have been detail displays.
v Summarize detailed wait reasons into larger wait categories.
v Control the size of the detail display by limiting it to wait reasons of a given severity. By default, only
wait reasons that account for more than 5% of the overall wait time are shown.
v Control the scale of the summary report graph.
v Display average or total performance group figures over a specified period of time. By default, the
performance group displays show you average figures for each interval.
v Exclude one or more wait reasons from the display.
v Add delays caused by terminal input waits to the performance group displays. Since terminal input wait
is not normally considered to be a system-related wait reason, by default it is not shown.
The keywords that control these options are summarized in the following table, and are described in the
following pages.
Table 5. Workload display options keywords
Keyword
Short
Description
Operand
AVERAGE
AVG
Displays average performance group figures, and
combines the data over the time range specified in the
command. (Default)
none
DETAIL
DET
Displays multiple wait reasons for each interval, job, or none
step. (Default)
EXWAIT
EXW
Excludes one or more wait reasons from the display.
One or more wait
reasons
MAXSCALE
MAX
Sets the scale for the graphic section of the summary
display.
nn seconds, minutes, or
hours
PLOTMIN
PMIN
Limits the display by excluding wait reasons that cause A percent
less than a certain percentage of the overall wait time.
(Default=5%)
SUMMARY
SUM
Displays only the most important wait reason for each
interval, job, or step.
none
SUMWAIT
SUMW
Summarizes the display by grouping similar wait
reasons together into summary wait categories.
none
TOTAL
TOT
Displays total performance group figures, and
combines the data over the time range specified in the
command.
none
TWAITOF
none
Excludes swaps due to terminal input waits in the
degradation display for performance groups. (Default)
none
TWAITON
none
Includes swaps due to terminal input waits in the
degradation display for performance groups.
none
Detail and summary displays (DETAIL, SUMMARY)
Workload displays can be shown at either the detail or summary level. DETAIL is the default; to show
summary information, simply add the SUMMARY keyword to the command.
46
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The DETAIL and SUMMARY keywords work very much the same for performance group and batch job
displays, with some minor differences.
For performance groups, the detail display shows multiple wait reasons for each interval; the summary
display shows only the most important wait reason. The following figures show examples of detail and
summary displays for performance group 2.
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 03:00 To 03:15 On 08/07/99
Elap = 14:54 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.04 S 11.8|---->
.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.11 S 29.4|----------->.
.
.
.
.
.
.
.|
|ECB Wait
0.04 S 11.8|---->
.
.
.
.
.
.
.
.
.|
|Waiting For MVS Lock
0.04 S 11.8|---->
.
.
.
.
.
.
.
.
.|
|Detected Wait
0.04 S 11.8|---->
.
.
.
.
.
.
.
.
.|
|Long Wait
0.02 S
5.9|--> .
.
.
.
.
.
.
.
.
.|
|STIMER Wait
0.02 S
5.9|--> .
.
.
.
.
.
.
.
.
.|
|ECB Wait (W/ STIMER)
0.02 S
5.9|--> .
.
.
.
.
.
.
.
.
.|
|Disk TSO022 320 Act 0.02 S
5.9|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
0.37 S
97 MVS Transactions Ended
|
|Productivity Index
22%
|
+=============================================================================+
Figure 33. DETAIL performance group degradation display
The scale of the detail display represents the percent of the total degradation attributable to that wait
reason, and is always zero through one hundred. The format of the graphic arrow changes if the wait
reason exceeds the 30% and 60% marks. These changes in the graphic arrow are illustrated in Figure 35
on page 48.
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 03:00 To 04:41 On 08/07/99
|
+-----------------------------------------------------------------------------+
|DATE__START_END___MAIN_REASON(*)_TIME(-)|0S____0.2____0.4____0.6____0.8_____1|
|08/07 03:00 03:15 Swap Page-In
0.37 S|****--------> .
.
.
.|
|
03:15 03:30 Using CPU
0.20 S|***--->.
.
.
.
.|
|
03:30 03:45 Wait For Lock
0.18 S|**---> .
.
.
.
.|
|
03:45 04:00 Term Out Wait
0.31 S|****------>
.
.
.
.|
|
04:00 04:15 Detected Wait
0.53 S|***--------------> .
.
.|
|
04:15 04:30 ECB Wait
0.31 S|***------->
.
.
.
.|
|
04:30 04:41 ECB Wait
1.23 S|************---------------------->+|
|---------------------------------------|
|Average For This Workload Display 0.45 S
|
|Productivity Index
20%
|
+=============================================================================+
Figure 34. SUMMARY performance group degradation display
The interval described by the detail display is summarized in the first line of the summary display. Only the
most important wait reason (Swap Page-In) is shown in the summary.
Notice that the arrows in the summary display consist of asterisks and hyphens. The total length of the line
represents the average transaction time for that interval. The segment of the line depicted with asterisks
indicates the average time spent on that wait reason for the interval.
For example, consider the interval from 3:00 to 3:15. (This interval is depicted in the sample detail display,
and in the first line of the summary display.) The total length of the arrow in the summary display is 0.37
Chapter 4. Displaying workload information
47
seconds, the average transaction time for that interval. The segment of the line represented with asterisks
is 0.11 seconds, meaning that transactions spent an average of 0.11 seconds waiting for swap page-ins
during that interval.
For batch jobs, the detail display shows multiple wait reasons for each job or step (depending on whether
it is a job level or step level display). The summary display shows only the most important wait reason for
each job or step. The rules for summary job and step level displays are the same as for detail displays. To
summarize:
v If you specify a JOBNAME, the summary display shows the main wait reason for each run of the
specified job. If you add the STEP keyword, it shows the main wait reason for each step of each run of
the job.
v If you do not specify JOBNAME and use the ACCOUNT, PROGRAM, or CLASS keyword, the summary
display shows the main wait reason for each job step that meets the selection criteria established by the
other keywords in the command.
Here are sample detail displays of the first two runs of batch job RPTGENER on 8/28/99:
+=============================================================================+
| Job = RPTGENER
JES Number = 4401
Job Steps = 1 / 1 |
| Job Class = F
Account Number =
Input Queue = 3.64 S |
| From 08:00 To 08:02 On 08/28/99
Elap =
1:10 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
6.46 S
9.2|--->.
.
.
.
.
.
.
.
.
.|
|Disk SYS024 140 Act 38.82 S 55.3|------------==========> .
.
.
.
.|
|Waiting For CPU
11.09 S 15.8|------> .
.
.
.
.
.
.
.
.|
|Disk SYS024 140 Que 5.55 S
7.9|--->.
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
1:10 M
|
|Productivity Index
65%
|
+=============================================================================+
+=============================================================================+
| Job = RPTGENER
JES Number = 4404
Job Steps = 1 / 1 |
| Job Class = F
Account Number =
Input Queue = 4.52 S |
| From 08:02 To 08:08 On 08/28/99
Elap =
5:36 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
6.05 S
1.8|.
.
.
.
.
.
.
.
.
.
.|
|SYSDSN
Enqueue
4:15 M 75.9|------------============>>>>>>> . . .|
|Waiting For CPU
31.95 S
9.5|--->.
.
.
.
.
.
.
.
.
.|
|Disk SYS024 140 Act 27.25 S
8.1|--->.
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
5:36 M
|
|Productivity Index
10%
|
+=============================================================================+
Figure 35. DETAIL batch job degradation display
This example illustrates the incremental changes in the graphic arrow as the degradation percentage
passes the 30% and 60% marks. The segment of the arrow below 30% is constructed with hyphens. If the
wait reason exceeds 30%, the segment of the arrow above 30 is constructed with equal signs instead of
hyphens. If the wait reason exceeds 60%, the segment above 60 is constructed with greater-than signs.
There are more panels in the detail display that are not shown here. A summary display of the same job
over a slightly longer time period is shown in Figure 36 on page 49.
48
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+=============================================================================+
| Job = RPTGENER
|
| From 08:00 On 08/28/99 To 16:36 On 08/29/99
|
+-----------------------------------------------------------------------------+
|DATE__START_END___MAIN_REASON(*)_TIME(-)|0M_____5_____10_____15_____20_____25|
|08/28 08:00 08:02 Da 140 SYS024
1:10 M|*>
.
.
.
.
.|
|
08:02 08:08 Enq SYSDSN
5:36 M|*****-->
.
.
.
.|
|
15:51 15:56 Enq SYSDSN
3:56 M|***--> .
.
.
.
.|
|
16:01 16:02 Da 140 SYS024 54.39 S|>
.
.
.
.
.|
|08/29 07:35 07:35 Tape Mnt Pend 34.79 S|>
.
.
.
.
.|
|
07:38 08:03 Ta 522 001050 23:05 M|*********----------------------> .|
|
16:32 16:33 CPU Wait
48.88 S|>
.
.
.
.
.|
|
16:35 16:36 Da 140 SYS024 44.55 S|>
.
.
.
.
.|
|---------------------------------------|
|Average For This Workload Display 4:36 M
|
|Productivity Index
20%
|
+=============================================================================+
Figure 36. SUMMARY batch job degradation display
As you can see from the example, there were eight runs of the job during the specified time period. Only
the main wait reason from each run is shown in the summary display.
It is possible to go directly from a summary display to a detail display without entering a new command. To
see a detail display for a line item in a summary display, simply place a D on that line and press Enter.
This feature is described more fully in Chapter 8, “Fast-path navigation,” on page 125.
Detail wait reasons
Table 6 shows the detail wait reasons that show up on the workload display. For a list of the wait reason
keywords that can be used with EPILOG commands (such as EXWAIT), see Table 8 on page 58.
Table 6. Detail wait reasons
Wait Reason
Description
Active I/O
Delay due to an active I/O in progress.
APPC Wait
Delay due to voluntary swap-out of an idle APPC address space. MVS/SP 4.2 introduced
the Advanced Program-to-Program Communication (APPC) transaction processor
service. APPC transaction processors are treated as started tasks by the EPILOG
collector and reporter.
Auxiliary Storage
Delay due to swaps caused by a shortage of auxiliary space. Auxiliary space shortage is
due to a high allocation of local page slots.
Central Storage
Delay caused by swap-out to improve central storage availability. (MVS/SP 4.2 and
above)
Common Page-In Wait
Delay due to address spaces waiting for a PLPA or common page-in.
Detected Wait
Delay due to swaps caused by a detected wait. A detected wait swap occurs when a
user address space has not executed in more than 8 SRM seconds or 2 seconds,
whichever is less (in current versions of MVS), and has not issued a long WAIT
command.
Disk Mount Pending
Delay due to address spaces waiting for an operator to mount a disk.
ECB Wait
The address space has issued a voluntary wait for some reason. Examples are an IMS™
message region waiting for work, a CICS region waiting for work, or even OMEGAMON
for MVS waiting between screen refreshes. High percentages of ECB waits may indicate
a program error, normal voluntary waits, or, if the Candle Subsystem is not running, a
system I/O reconfiguration.
ECB Wait (with Stimer)
Delay due to ECB waits with an STIMER TASK or REAL pending (see ECB Wait above).
Chapter 4. Displaying workload information
49
Table 6. Detail wait reasons (continued)
Wait Reason
Description
Enqueue Exchange Swap
Delay due to swaps performed to make room for users that are enqueued on a resource
required by other users.
Exchange Swap
Delay due to exchange swaps. Exchange swaps are performed when one user in a
given domain must be swapped out to make room for another user in the same domain
that has a higher recommendation value.
Enqueue Wait
Delay due to waiting for an enqueue.
HSM Delay (Backup)
Delay due to waiting for HSM to execute a data set backup request.
HSM Delay (Delete)
Delay due to waiting for HSM to delete a data set.
HSM Delay (JES C/I)
Delay due to waiting for HSM to execute a locate request for a control interval.
HSM Delay (Migrate)
Delay due to waiting for HSM to migrate a data set.
HSM Delay (Read CTL)
Delay due to waiting for HSM to read a control data set record.
HSM Delay (TSO HLIST)
Delay due to waiting for HSM to execute an HLIST command.
HSM Delay (Recall)
Delay due to waiting for HSM to recall a data set.
HSM Delay (Recover)
Delay due to waiting for HSM to recover a backup data set.
JES Cancel
Delay due to a job cancel request issued to JES.
JES Delay (Delete)
Delay due to a job delete request issued to JES.
JES Delay (Requeue)
Delay due to a job requeue request to JES.
JES Delay (Status)
Delay due to a job status request issued to JES that has been queued.
JES Delay (Sysout)
Delay due to a process SYSOUT request issued to JES that has been queued.
Long Wait Swap
Delay due to swaps caused by user-requested long waits.
Out Too Long
Delay caused by swap-out to enable the swap-in of an address space that had been
swapped out too long. (MVS/SP 4.2 and above)
Private Page-In Wait
Delay due to address spaces waiting for a private page-in operation.
Queued I/O
Delay due to waiting for queued I/O.
Real Storage Swap
Delay due to swaps caused by a shortage of real pageable frames.
Request Swap
Delay due to requested swaps.
Reserve I/O
Delay due to wait for RESERVEd I/O (MVS/370 only).
SRM Delay (MPL)
SRM Delay (MPL); this can occur when a domain is at the maximum MPL and a
unilateral or exchange swap must occur to get the ready work swapped in. The READY
delay occurs while waiting for SRM to allow the swap. When the ASM has scheduled a
swap-in, any subsequent delay will then show up in the Swap Page-In (SWAPIN)
category.
SRM Delay (RTO)
Delay imposed by SRM to meet the response time objective set in the IPS.
Stimer Wait
The address space has issued an STIMER and is voluntarily waiting for it to end. This is
typically not considered degradation since it is a voluntary wait.
Swap Page-In Wait
The address space has been given to the ASM queue and is waiting for MVS to swap it
into storage. A high value for this wait can be caused by I/O problems such as a slow I/O
device or DASD contention.
Swapped with WTOR
Delay due to a long or detected wait swap with WTOR pending.
System Paging
Delay caused by swap-out to reduce the system page fault rate. (MVS/SP 4.2 and
above)
50
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 6. Detail wait reasons (continued)
Wait Reason
Description
Tape Mount Pending
Delay due to tape mounts. The percentage indicated in the degradation data reflects the
percentage of time in which the tape mount pending condition was in effect. Note that
this may not always represent degradation.
For example, an application may issue a tape mount request some time before it actually
requires I/O to the data on that tape. (This is often the case with such applications as
CICS, which can perform asynchronous I/O.) Under these conditions, the job continues
running while the tape mount pending condition is in effect, with no degradation to that
job.
Remember also that the tape mount pending condition is in effect from the time that the
tape mount request is issued to the operator console to the time that the tape is actually
mounted and the tape drive is made READY. For those installations that use them, this
time may actually be spent waiting for a tape management system, not a system
operator.
Terminal Input Wait
Delay due to swaps caused by waiting for terminal input.
Terminal Output Wait
Delay due to swaps caused by a TSO session that is waiting for an output buffer or
waiting for the user to press Enter after “***” is displayed in line output mode.
Transition Swap
Delay due to swap-outs that occur because an address space was made
non-swappable. (When an address space is made non-swappable, it has to be swapped
out and then swapped back in with non-swappable status.)
Unilateral Swap
Delay due to swaps that occur because the MPL for a domain exceeded the target MPL.
Using CPU
Time spent using CPU.
Waiting for CPU
Delay due to address spaces waiting on the active CPU dispatching queue. High
percentages here may mean that the address space is low priority or that there is a
shortage of CPU cycles.
Waiting for MVS Lock
Delay due to address spaces waiting to acquire a local or global MVS lock.
Waiting For Staging
Waiting for a mass storage volume to be staged.
Summary wait categories (SUMWAIT)
By default, workload degradation displays show you detail wait reasons, such as being queued or active
on a particular I/O device, swapped due to an auxiliary storage shortage, and so on. The SUMWAIT
keyword groups these detail wait reasons into summary categories, such as I/O delay and Swap Delay,
and displays the degradation by summary wait category. This feature considerably simplifies the display,
and allows you to focus in quickly on the main sources of the problem without having to read through a
large number of wait reasons.
A typical degradation display showing detail wait reasons is shown in Figure 37 on page 52. The command
used to produce this display was:
DISPLAY JOB(APL17PTF) YESTERDAY
Chapter 4. Displaying workload information
51
+-----------------------------------------------------------------------------+
| Job = APL17PTF
JES Number = 2990
Job Steps = 1 / 1 |
| Job Class = A
Account Number = DCSP07
Input Queue = 14.36 S |
| From 11:36 To 11:38 On 07/01/99
Elap =
2:11 M SYSA |
+-----------------------------------------------------------------------------+
|Wait Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
5.11 S
3.9|-> .
.
.
.
.
.
.
.
.
.|
|Waiting For CPU
1:01 M 47.2|------------======> .
.
.
.
.
.|
|Private Page-In Wait 13.77 S 10.5|---->
.
.
.
.
.
.
.
.
.|
|SRM Delay (MPL)
13.77 S 10.5|---->
.
.
.
.
.
.
.
.
.|
|Disk APPL17 9E6 Act 13.77 S 10.5|---->
.
.
.
.
.
.
.
.
.|
|ECB Wait
6.82 S
5.2|--> .
.
.
.
.
.
.
.
.
.|
|Disk APL031 330 Act 6.82 S
5.2|--> .
.
.
.
.
.
.
.
.
.|
|Waiting For MVS Lock
6.82 S
5.2|--> .
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
2:11 M
|
|Productivity Index
21%
|
+-----------------------------------------------------------------------------+
Figure 37. Degradation display with detailed wait reasons
By adding the SUMWAIT keyword to the command, we can display the same degradation by summary
wait category:
DISPLAY JOB(APL17PTF) YESTERDAY SUMWAIT
+-----------------------------------------------------------------------------+
| Job = APL17PTF
JES Number = 2990
Job Steps = 1 / 1 |
| Job Class = A
Account Number = DCSP07
Input Queue = 14.36 S |
| From 11:36 To 11:38 On 07/01/99
Elap =
2:11 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|CPU Used
5.11 S
3.9|-> .
.
.
.
.
.
.
.
.
.|
|CPU Wait
1:01 M 47.2|------------======> .
.
.
.
.
.|
|I/O
12.80 S 17.3|------> .
.
.
.
.
.
.
.
.|
|Page + Swap
13.77 S 10.5|---->
.
.
.
.
.
.
.
.
.|
|SRM Delay(MPL)
13.77 S 10.5|---->
.
.
.
.
.
.
.
.
.|
|Misc Event
13.64 S 10.3|---->
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
2:11 M
|
|Productivity Index
21%
|
+-----------------------------------------------------------------------------+
Figure 38. Degradation display with summary wait reasons
As you can see, the detailed wait reasons have been grouped into summary wait reasons in the
SUMWAIT display. For example, the two I/O waits (Active on disk devices) have been grouped into a
summary category called I/O. The degradation shown next to I/O is the combined degradation of the
detailed wait reasons that comprise it.
The productivity index shown on the SUMWAIT display is based upon detailed wait reasons, not summary
categories. (This is because summary categories are groups of detailed wait reasons, and each category
is potentially a mixture or productive, unproductive, and idle execution states.)
The grouping of detailed wait reasons into summary categories is controlled by a list of summary wait
definitions in a member of the rhilev.REDPARM partitioned data set. For further information about this data
set, see “Reporter option defaults” on page 20.
When you invoke the reporter, the default member (named DEFAULTS) is used to provide summary
category definitions. Table 7 on page 53 shows the default summary wait categories and the detail wait
reasons included in each category. Detail wait reasons are described in Table 6 on page 49.
52
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 7. Default summary wait categories
Summary wait categories
Detail wait reasons
CPU USED
Using CPU
CPU WAIT
Waiting for CPU
ENQUEUE
Waiting for an enqueue
HSM
Waiting
Waiting
Waiting
Waiting
Waiting
Waiting
Waiting
Waiting
I/O
Active on an I/O device
Waiting for a logical channel
Queued on an I/O device
Waiting for a reserved I/O device
JES
Waiting
Waiting
Waiting
Waiting
Waiting
MISC EVENT
ECB waits with STIMER TASK or REAL pending
Voluntary waits
Waiting to acquire an MVS lock
SRM delay to meet the response time objective
Waiting for a mass storage volume to be staged
Waiting for a STIMER to end
OPER ACT
Waiting for a disk mount operation
Waiting for a tape mount operation
Swapped with WTOR pending
PAGE + SWAP
Waiting for a PLPA or common page-in
Waiting for a private page-in
Waiting to be swapped in
SRM DELAY
(MPL)
Waiting for SRM to allow a swap
SWAP
Auxiliary storage shortage swaps
Detected wait swaps
Enqueue exchange swaps
Exchange swaps
User-requested long wait swaps
Real storage shortage swaps
Requested swaps
Terminal output wait swaps
Transition swaps
Unilateral swaps
Out Too Long swaps
Central Storage swaps
System Paging swaps
APPC Wait swaps
TERM IN
Waiting for terminal input
for
for
for
for
for
for
for
for
for
for
for
for
for
a data set backup
a C/I locate
a data set deletion
an HLIST user command
a data set migration
a control data set record
a data set recall
backup data set recovery
a
a
a
a
a
job cancellation
job deletion
queued job status request
job requeue request
queued SYSOUT request
Chapter 4. Displaying workload information
53
You can override this definition by creating a member in rhilev.REDPARM with your own summary wait
definitions, and then invoking it with the SET command.
For example, if you had set up your own summary wait definitions in a member of rhilev.REDPARM called
MYSUMS, you would invoke it by entering:
SET SUMDEF(MYSUMS)
From that point on, anytime you requested a SUMWAIT display, the definitions contained in MYSUMS
would be used to format the display, instead of those in DEFAULTS.
You can also set the SUMWAIT option on with the SET command. To do so, you would enter:
SET SUMWAIT
After entering this command, all subsequent degradation displays show summary wait categories instead
of detailed wait reasons.
For a detailed discussion of the SET command, see “Setting reporter options” on page 137.
You can navigate from a SUMWAIT display to one showing detailed wait reasons by entering a W on any
of the summary wait lines. The resulting degradation display shows you the detailed wait reasons that
correspond to that summary wait category. This feature is described more fully in Chapter 8, “Fast-path
navigation,” on page 125.
Limiting the display by wait reason severity (PLOTMIN)
The PLOTMIN keyword excludes wait reasons from the display that account for less than a certain
percentage of the overall wait time. For example:
DISPLAY PGN(2) PLOTMIN(10)
If you specify PLOTMIN(10), only wait reasons that account for 10% or more of the overall wait time are
included in the display.
By default, the reporter excludes wait reasons that account for less than 5% of the wait time. Adjusting this
figure downwards shows more wait reasons; setting the figure at zero (0) shows all wait reasons.
By including the PLOTMIN (or PMIN) keyword in a DISPLAY command, you set the minimum wait reason
percentage for that command only. Alternatively, you can set a PLOTMIN default for all successive
DISPLAY commands by using the SET keyword. Setting defaults such as PLOTMIN is described in
“Setting display defaults” on page 95.
Average and total performance group figures
By default, performance group information is displayed as average figures for each interval. These figures
are calculated by dividing the total transaction elapsed time by the number of transactions that ended
during the interval. The AVERAGE and TOTAL keywords allow you to display average or total transaction
time over a specified period of time.
The AVERAGE keyword differs from the default in that it shows average figures over the specified time
period instead of the individual RMF interval. This has exactly the same effect as adding the COMBINE
keyword to the command without an operand. For example, if you enter:
DISPLAY PGN(2) AVERAGE YESTERDAY
The resulting display looks like the one in Figure 39 on page 55.
54
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 00:16 To 24:00 On 08/29/99
SYSA |
| Average Of 94 Intervals
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.06 S
4.6|-> .
.
.
.
.
.
.
.
.
.|
|Long Wait
0.60 S 43.4|------------=====> .
.
.
.
.
.|
|Waiting For CPU
0.17 S 12.2|---->
.
.
.
.
.
.
.
.
.|
|Average Trans Time
1.39 S
80948 MVS Transactions Ended
|
|Productivity Index
7%
|
+=============================================================================+
Figure 39. AVERAGE performance group degradation display
Notice that the AVERAGE display combines all intervals for the specified time period (YESTERDAY), and
shows you the following information:
v the start and end time/dates of the display
v the number of intervals combined in the display
v the average transaction time and the number of transactions that ended during the period
The TOTAL display shows you the total wait time of all transactions for each wait reason. Like the
AVERAGE display, it also combines intervals into the specified time period. You get total figures by adding
the TOTAL keyword to the command:
DISPLAY PGN(2) YESTERDAY TOTAL
The resulting display looks like the one in Figure 40.
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 00:16 TO 24:00 ON 08/29/99
SYSA |
| Total Of 94 Intervals
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
1:26 H
4.6|-> .
.
.
.
.
.
.
.
.
.|
|Long Wait
13:34 H 43.4|------------=====> .
.
.
.
.
.|
|Waiting For CPU
3:49 H 12.2|---->
.
.
.
.
.
.
.
.
.|
|Total Trans Time
31:17 H
80948 MVS Transactions Ended
|
|Productivity Index
7%
|
+=============================================================================+
Figure 40. TOTAL performance group degradation display
If you are using AVERAGE or TOTAL along with exception analysis keywords (SIF or RIF), see “Workload
exception filters” on page 143.
Setting the display scale (MAXSCALE)
You can use the MAXSCALE keyword to adjust the scale of summary displays. For example, if one
interval on a display has a much larger wait reason than other intervals, the scale of the display is
automatically set to show that interval. This setting, however, may make the other intervals difficult to read.
For example, consider the following summary display.
Chapter 4. Displaying workload information
55
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 18:02 To 24:00 On 06/02/99
SYSA |
+-----------------------------------------------------------------------------+
|DATE__START_END___MAIN_REASON(*)_TIME(-)|0S_____1______2______3______4______5|
|06/02 18:02 18:15 Using CPU
0.62 S|*-->
.
.
.
.
.|
|
18:15 18:30 Swap Page-In
0.50 S|*-->
.
.
.
.
.|
|
20:06 20:15 Using CPU
0.50 S|*-->
.
.
.
.
.|
|
21:30 21:45 Da 146 SPOOL1
0.56 S|*-->
.
.
.
.
.|
|
21:45 22:00 Swap Page-In
0.07 S|>
.
.
.
.
.|
|
22:00 22:15 Swap Page-In
0.46 S|-->
.
.
.
.
.|
|
22:15 22:30 Long Wait Swp
3.28 S|************---------->
.
.|
|
22:30 22:45 Long Wait Swp
3.38 S|*****************------>
.
.|
|
22:45 23:00 Detected Wait
0.51 S|*-->
.
.
.
.
.|
|
23:00 23:15 Using CPU
1.01 S|*----->.
.
.
.
.|
|
23:15 23:30 Long Wait Swp
0.83 S|***--> .
.
.
.
.|
|
23:30 23:45 Long Wait Swp
2.75 S|***************---> .
.
.|
|
23:45 24:00 Long Wait Swp
0.48 S|*->
.
.
.
.
.|
|---------------------------------------|
|Average For This Workload Display 1.15 S
|
|Productivity Index
22%
|
+=============================================================================+
Three intervals are much longer than the rest, and have extended the graphic section of the display. As a
result, the resolution of the other intervals on the scale is poor, and it is difficult to evaluate them
accurately. If we issue the following command, the maximum scale is set to one second:
DISPLAY PGN(2) SDATE(6/22) EDATE(6/22) SUM MAXSCALE(1S)
The resulting display is shown in Figure 41.
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 18:02 To 24:00 On 06/02/99
SYSA |
+-----------------------------------------------------------------------------+
|DATE__START_END___MAIN_REASON(*)_TIME(-)|0S____0.2____0.4____0.6____0.8_____1|
|06/02 18:02 18:15 Using CPU
0.62 S|*****--------------->.
.
.|
|
18:15 18:30 Swap Page-In
0.50 S|***------------->
.
.
.|
|
20:06 20:15 Using CPU
0.50 S|****-------------> .
.
.|
|
21:30 21:45 Da 146 SPOOL1
0.56 S|***---------------> .
.
.|
|
21:45 22:00 Swap Page-In
0.07 S|*->
.
.
.
.
.|
|
22:00 22:15 Swap Page-In
0.46 S|***------------>
.
.
.|
|
22:15 22:30 Long Wait Swp
3.28 S|*******************--------------->+|
|
22:30 22:45 Long Wait Swp
3.38 S|**************************-------->+|
|
22:45 23:00 Detected Wait
0.51 S|******-----------> .
.
.|
|
23:00 23:15 Using CPU
1.01 S|*****----------------------------->+|
|
23:15 23:30 Long Wait Swp
0.83 S|***************------------->
.|
|
23:30 23:45 Long Wait Swp
2.75 S|****************************------>+|
|
23:45 24:00 Long Wait Swp
0.48 S|**********------>
.
.
.|
|---------------------------------------|
|Average For This Workload Display 1.15 S
|
|Productivity Index
22%
|
+=============================================================================+
Figure 41. Degradation display scale adjustment
By reducing the scale, the resolution of the other intervals on the display is much greater. Notice how the
arrows for the four intervals that now extend beyond the graph end with a plus sign.
By default, the scale range is based on the highest interval in the display. The units of the display are set
to seconds, minutes, or hours, and the range of the display is set by rounding up to the next highest
multiple of five. For example, if the highest value in the display is 2.2 seconds, the scale range is rounded
up to five seconds. The only exception to this is when the highest value is less than one. In this case, the
scale range is set to one.
56
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
You override this setting in a very similar way with the MAXSCALE keyword. You set the units of the
display by including an s, m, or h suffix with the operand. The numeric part of the operand is processed
the same way as the highest interval value. If it is greater than one, it is rounded up to the nearest even
multiple of five. If it is less than or equal to one, it is set to one.
For example, if you set MAXSCALE(2M), the scale range is rounded up to five minutes. If you specify
MAXSCALE(1S), the range is set to one second.
The MAXSCALE keyword is valid with summary displays only; if you enter it for a detail display, it is
ignored.
Excluding wait reasons from the display
You can exclude one or more detail wait reasons from the display with the EXWAIT keyword. This is often
desirable when one wait reason dominates the degradation display. For example, when displaying a CICS
performance group, you will sometimes find very large ECB waits with STIMER:
+=============================================================================+
| Performance Group =
6
Symbolic Name = CICSP
|
| From 14:00 to 14:15 on 03/29/99
Elap = 14:44 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
4.7|-> .
.
.
.
.
.
.
.
.
.|
|ECB Wait (w/ STIMER)
56.5|------------==========> .
.
.
.
.|
|Disk SYSA21 D87 Act
14.2|-----> .
.
.
.
.
.
.
.
.|
|Waiting For CPU
6.3|--> .
.
.
.
.
.
.
.
.
.|
|Productivity Index
22%
|
+=============================================================================+
(Notice that since no CICS address space terminated during this interval, there are no average transaction
time figures, nor does the display show any values in the Time field.)
Since this wait reason does not really represent degradation, you may want to exclude it from the display.
To do so, include the EXWAIT(ccc) keyword in the command, where ccc = a detail wait reason. (Valid wait
reason names and short forms are shown in Table 8 on page 58.) For example, to exclude ECB waits from
the above display, enter:
DISPLAY PGN(6) EXWAIT(ECBSTIMER)
The resulting display looks like the one in Figure 42.
+=============================================================================+
| Performance Group =
6
Symbolic Name = CICSP
|
| From 14:00 to 14:15 on 03/29/99
Elap = 14:44 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
10.5|---->
.
.
.
.
.
.
.
.
.|
|Disk SYSA21
D87 Act
32.7|------------=> .
.
.
.
.
.
.|
|Waiting for CPU
14.6|-----> .
.
.
.
.
.
.
.
.|
|ECB Wait
7.7|--->.
.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
5.6|--> .
.
.
.
.
.
.
.
.
.|
|Productivity Index
45%
|
+=============================================================================+
Figure 42. Degradation display with excluded waits
The ECBSTIMER waits have been ignored and the remaining wait reason percentages have been
normalized to 100 percent. Note that when EXWAIT is used, the Time field is suppressed. By excluding
segments of the transaction with EXWAIT and normalizing the percentages, we have made the
Chapter 4. Displaying workload information
57
percentages irrelevant to the total time. Also, because a major wait was excluded, new wait reasons may
appear on the display because the percentages were normalized and their new values exceed the
PLOTMIN threshold.
The productivity index is recalculated with the normalized values for each remaining wait reason.
The EXWAIT keyword can accept multiple detail wait reasons as operands, but cannot be used with
summary wait categories.
Wait reason codes
The following table lists the wait reason codes that can be used with the EXWAIT keyword. For each
keyword, the table shows the short name, the summary wait reason code, and a description of the wait
reason or execution state. (The summary wait codes can be used as keywords in exception filter reporting,
which is described in Chapter 9, “Advanced reporting functions,” on page 137. Each summary wait code
represents a summary wait category, as shown in Table 7 on page 53.)
Table 8. Detail wait reason codes
Keyword
Short
Summary
Description
ACTIVEIO
AIO
I/O
Delay due to an active I/O in progress.
APPCWAIT
APS
SWAP
Delay due to voluntary swap-out of an idle APPC address
space.
AUXSTOR
AUX
SWAP
Delay due to swaps caused by a shortage of auxiliary space.
Auxiliary space shortage is due to a high allocation of local page
slots.
BACKUP
BKP
HSM
Delay due to waiting for HSM to execute a data set backup
request.
CANCEL
CAN
JES
Delay due to a job cancel request issued to JES.
CENTSTOR
CSS
SWAP
Delay caused by swap-out to improve central storage availability.
CILOCATE
CIL
HSM
Delay due to waiting for HSM to execute a locate request for a
control interval.
COMMON
COM
PGSW
Delay due to address spaces waiting for a PLPA or common
page-in.
CPUTIL
CPU
CPU
Time spent using CPU.
CPUWAIT
CPW
CPW
Delay due to address spaces waiting on the active CPU
dispatching queue. High percentages here may mean that the
address space is low priority or that there is a shortage of CPU
cycles.
DELETEDS
DLD
HSM
Delay due to waiting for HSM to delete a data set.
DELETEJOB
DEL
JES
Delay due to a job delete request issued to JES.
DETECTED
DET
SWAP
Delay due to swaps caused by a detected wait. A detected wait
swap occurs when a user address space has not executed in
more than 8 SRM seconds or 2 seconds, whichever is less (in
current versions of MVS), and has not issued a long WAIT
command.
DISKMOUNT
DMP
OPER
Delay due to address spaces waiting for an operator to mount a
disk.
ECBSTIMER
ECS
MISC
Delay due to ECB waits with an STIMER TASK or REAL
pending (see ECBWAIT below).
58
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 8. Detail wait reason codes (continued)
Keyword
Short
Summary
Description
ECBWAIT
ECB
MISC
The address space has issued a voluntary wait for some reason.
Examples are an IMS message region waiting for work, a CICS
region waiting for work, or even OMEGAMON for MVS waiting
between screen refreshes. High percentages of ECB waits may
indicate a program error, normal voluntary waits, or, if the
Candle Subsystem is not running, a system I/O reconfiguration.
ENQEXC
EEX
SWAP
Delay due to swaps performed to make room for users that are
enqueued on a resource required by other users.
ENQUEUE
ENQ
ENQU
Delay due to waiting for an enqueue.
EXCHANGE
EXG
SWAP
Delay due to exchange swaps. Exchange swaps are performed
when one user in a given domain must be swapped out to make
room for another user in the same domain that has a higher
recommendation value.
HLIST
HLS
HSM
Delay due to waiting for HSM to execute an HLIST command.
JOBSTATUS
JST
JES
Delay due to a job status request issued to JES that has been
queued.
LONGWAIT
LON
SWAP
Delay due to swaps caused by user-requested long waits.
MIGRATE
MIG
HSM
Delay due to waiting for HSM to migrate a data set.
MVSLOCK
LCK
MISC
Delay due to address spaces waiting to acquire a local or global
MVS lock.
OUTLONG
OLS
SWAP
Delay caused by swap-out to enable the swap-in of an address
space that had been swapped out too long.
PAGEIN
PAG
PGSW
Delay due to address spaces waiting for a private page-in
operation.
QUEUEDIO
QIO
I/O
Delay due to waiting for queued I/O.
READCTL
RCR
HSM
Delay due to waiting for HSM to read a control data set record.
READY
RDY
SRMD
SRM Delay (MPL); this can occur when a domain is at the
maximum MPL and a unilateral or exchange swap must occur to
get the ready work swapped in. The READY delay occurs while
waiting for SRM to allow the swap. When the ASM has
scheduled a swap-in, any subsequent delay will then show up in
the Swap Page-In (SWAPIN) category.
REALSTOR
RST
SWAP
Delay due to swaps caused by a shortage of real pageable
frames.
RECALL
RCL
HSM
Delay due to waiting for HSM to recall a data set.
RECOVER
RCV
HSM
Delay due to waiting for HSM to recover a backup data set.
REQUEST
REQ
SWAP
Delay due to requested swaps.
REQUEUE
RQU
JES
Delay due to a job requeue request to JES.
RESRVEIO
RIO
I/O
Delay due to wait for RESERVEd I/O (MVS/370 only).
SRMRTO
RTO
MISC
Delay imposed by SRM to meet the response time objective set
in the IPS.
STAGING
MSS
MISC
Waiting for a mass storage volume to be staged.
STIMER
STI
MISC
The address space has issued an STIMER and is voluntarily
waiting for it to end. This is typically not considered degradation
since it is a voluntary wait.
Chapter 4. Displaying workload information
59
Table 8. Detail wait reason codes (continued)
Keyword
Short
Summary
Description
SWAPIN
SWI
PGSW
The address space has been given to the ASM queue and is
waiting for MVS to swap it into storage. A high value for this wait
can be caused by I/O problems such as a slow I/O device or
DASD contention.
SYSOUT
PSO
JES
Delay due to a process SYSOUT request issued to JES that has
been queued.
SYSPAGE
SPS
SWAP
Delay caused by swap-out to reduce the system page fault rate.
TAPEMOUNT
TMP
OPER ACT
OPER
Delay due to tape mounts. The percentage indicated in the
degradation data reflects the percentage of time in which the
tape mount pending condition was in effect. Note that this may
not always represent degradation.
For example, an application may issue a tape mount request
some time before it actually requires I/O to the data on that
tape. (This is often the case with such applications as CICS,
which can perform asynchronous I/O.) Under these conditions,
the job continues running while the tape mount pending
condition is in effect, with no degradation to that job.
Remember also that the tape mount pending condition is in
effect from the time that the tape mount request is issued to the
operator console to the time that the tape is actually mounted
and the tape drive is made READY. For those installations that
use them, this time may actually be spent waiting for a tape
management system, not a system operator.
TERMIN
TIN
TERMI
Delay due to swaps caused by waiting for terminal input.
TERMOUT
TOU
SWAP
Delay due to swaps caused by a TSO session that is waiting for
an output buffer or waiting for the user to press Enter after “***”
is displayed in line output mode.
TRANSWAP
TSW
SWAP
Delay due to swap-outs that occur because an address space
was made non-swappable. (When an address space is made
non-swappable, it has to be swapped out and then swapped
back in with non-swappable status.)
UNILATERAL
UNI
SWAP
Delay due to swaps that occur because the MPL for a domain
exceeded the target MPL.
WTORWAIT
WTO
OPER
Delay due to a long or detected wait swap with WTOR pending.
For a list of wait reasons sorted by wait reason name, see Table 6 on page 49.
Terminal wait swaps (TWAITON)
When the system is waiting for a TSO user to enter data on the terminal, the user is not actually within an
MVS transaction; such a user is generally shown by degradation analysis to be swapped out for a terminal
input wait. Because users in this state are not within a transaction, this does not represent degradation
and is not ordinarily reported as such.
However, if you add the TWAITON keyword to a DISPLAY request, you can display these states. The
resulting display looks like the one in Figure 43 on page 61.
60
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 21:30 To 21:45 On 06/02/99
Elap = 14:57 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.06 S
.9|.
.
.
.
.
.
.
.
.
.
.|
|Terminal Input Wait
5.79 S 72.3|------------============>>>>>
.
.
.|
|Trans And Think Time
7.10 S
142 MVS Transactions Ended
|
|Productivity Index
22%
|
+=============================================================================+
Figure 43. Degradation Display with Terminal Input Waits
Instead of response time, the display now shows response plus think time.
By including the TWAITON keyword in a DISPLAY command, you display terminal input waits for that
command only. Alternatively, you can set a TWAITON default for all successive DISPLAY commands by
using the SET keyword. Setting defaults such as TWAITON is described in “Setting display defaults” on
page 95.
Once you have set the default to TWAITON, the TWAITOFF keyword allows you to suppress the display of
terminal input waits for a given display and still leave the default set to TWAITON.
Chapter 4. Displaying workload information
61
62
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 5. Displaying resource information
You can use the EPILOG reporter to display utilization information about system resources. This data has
been either extracted directly from RMF or generated by the EPILOG collector. This gives you online
access to the important data that RMF collects, plus additional useful information that is not supplied by
RMF.
Resource display keywords
You invoke the resource panels explicitly by using the appropriate keyword with the DISPLAY command.
For example, to use the RCHN keyword, enter a command in the following format:
DISPLAY RCHN [time period]
This command generates a set of Channel Resource Display panels for the specified time period. A
sample panel is shown in Figure 45 on page 66.
The resource keywords are shown in Table 9. For a description of the DISPLAY command syntax for these
keywords, see “Using the DISPLAY command” on page 26. Each sample resource panel in this chapter
shows a sample DISPLAY command above the sample report panel.
Table 9. Resource display keywords
Keyword
Description
Operand
RALL
All resource types
none
RCCH
Cache subsystem statistics
none
RCHN
Channel activity
none
RCPU
Hardware and address space CPU activity
none
RDAS
DASD device information
VOLSER or device address (optional).
RDOM
SRM domain statistics
List or range of domain numbers (optional).
RINF
General system information (CPU model,
MVS level, and so on)
none
RLCU (MVS/XA and
MVS/ESA only)
I/O queuing data
LCU number (optional).
RPAG
Real and virtual storage usage with paging
and swapping activity
none
RPDS
Page data set statistics
none
RPGN
SRM performance group statistics
List or range of performance group numbers
(optional).
RSDS
Swap data set statistics
none
RSRM
SRM MPL settings
none
RSWA
Swap activity statistics
none
RSWR
Swap reason statistics
none
RVLF
VLF statistics
none
You can also invoke resource panels automatically using the AUTOMATIC keyword on a DETAIL display.
This feature is known as Automatic Analysis and is discussed in “Automatic analysis” on page 161.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
63
Sample resource displays
The following pages show sample displays and field descriptions for the various resource types. (Many
fields correspond directly to fields which appear in RMF Post Processor reports.)
Note that some of the panels (RCHN, RCPU, RDAS, and RPAG) have slightly different formats when
displaying data from different MVS systems, and that some of the panels contain fields that relate to
specific environments, such as 3090 processors, expanded storage, and PR/SM™ machines.
Cache subsystem statistics (RCCH)
The cache subsystem resource panel provides detailed statistics and basic information for devices
attached to IBM 3880-13/23 and 3990-3 cache control units. This display applies to MVS/XA and
MVS/ESA systems only.
This display will help you tune your IBM cache control unit subsystems to help ensure that only priority
tasks are using cache, and that they are using it efficiently.
The following table defines the display fields on the sample RCCH panel.
Field
Description
Volume Serial
The volume serial identifier.
Dev #
The cache device address or number.
CU Type
The controller model (3880-13, 3880-23, or 3990-3)
SD/SSID
The cache device storage director (SD) for 3880-13 or 3880-23 controllers. The cache
device subsystem ID (SSID) for 3990 controllers only.
Cache State
The caching status. Possible values for this field are described below:
Activ - Device caching is active.
Inact - Device caching is inactive.
PenAc - Activate device cache is pending.
PenDa - Deactivate device cache is pending.
TimeO - The Cache API did not return in a timely manner.
Read Hit %
The total read hit percentage for the cache device, expressed in tenths of a percent.
Write Hit %
The total write hit percentage for the cache device, expressed in tenths of a percent.
DFW Hit %
The total DASD fast write write hit percentage for the cache device, expressed in tenths of a
percent. (3990 subsystem)
R/W %
The total read write hit percentage for the cache device, expressed in tenths of a percent.
(3990 subsystems only)
Inhibit Cache%
The inhibit cache percentage for the cache device, expressed in tenths of a percent.
Bypss Cache%
The bypass cache percentage for the cache device, expressed in tenths of a percent.
DFW Bypss%
The DASD fast write retry percentage for the cache device, expressed in tenths of a
percent. (3990 subsystem only)
The command
DISPLAY RCCH RIF ((Volser = CAN*) (Rdhittot < 70%))
generates a display similar to the following sample panel.
64
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
*******************************************************************************
+=========================== DASD Cache Activity =============================+
| From:
00:14 to 00:29 on 08/30/99
Elap = 14:57 M
SYSH |
+-----------------------------------------------------------------------------+
|Volume Dev
CU
SD/ Cache
Read Write DFW R/W Inhibit Bypss DFW |
|Serial #
Type
SSID State
Hit % Hit % Hit % % Cache% Cache% Bypss%|
|------ ---- ------- ---- ------ ----- ----- ----- ----- ----- ----- ----- |
|CAN001 02A0 3990-3 0010 activ
68.1 82.4 92.2 12.2
0.0 12.8
2.4 |
|CAN003 02A2 3990-3 0010 activ
72.7 88.3 48.1
4.4 32.8
0.4 |
|CAN004 02A3 3990-3 0010 inact
66.1 84.1 77.4 33.3
1.1 82.8
0.0 |
|CAN006 02A5 3990-3 0010 PenDa
45.9 72.4 82.2 1.1 13.9 1.1
0.1 |
|CAN011 03A0 3880-23 FAFB activ
69.2
n/a n/a 7.5 38.1 13.2
n/a |
|CAN012 03A0 3880-23 FAFB activ
69.2
n/a n/a 7.5 38.1 13.2
n/a |
|CAN014 03A3 3880-23 FAFB activ
n/a n/a
n/a |
|CAN016 03A3 3880-23 FAFB activ
67.4
n/a n/a 4.5
1.1 0.0
n/a |
|CAN022 0382 3880-13 7879 inact
59.2 64.2 n/a 48.5
8.9 n/a
n/a |
|CAN032 0382 3880-13 7879 activ
59.2 64.2 n/a 48.5
8.9 n/a
n/a |
+=============================================================================+
Figure 44. RCCH resource display (MVS/ESA)
Channel activity (RCHN)
The channel resource panel shows you information about channel activity during the interval. You will
notice that the display panel is slightly different for MVS/XA and MVS/ESA systems than for MVS/370
systems. The table below defines the display fields on the sample panels shown in Figure 45 on page 66
and Figure 46 on page 66.
Field
Description
CPU/CSID
Channel set ID to which the channel is attached. (MVS/370 only)
CHPID
Channel path ID in hex. (MVS/XA and MVS/ESA only)
Chn
Channel number. (MVS/370 only)
Typ
Channel type, such as byte multiplexor, selector, block multiplexor, ESCON® channel, ESCON
converter, or ESCON director port.
%Busy
Percentage of the time during the interval that the channel was busy.
I/Os /sec
Average number of successful START and RESUME I/O instructions issued to the channel per
second. (MVS/370 only)
Avg Srv.
Time(ms)
Average time (in milliseconds) required to complete each I/O operation to the channel. (MVS/370
only)
The command
DISPLAY RCHN STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 45 on page 66.
Chapter 5. Displaying resource information
65
+============================ Channel Activity ===============================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
| CPU/
I/Os Avg Srv. | CPU/
I/Os Avg Srv. |
| CSID Chn Typ %Busy /sec Time(ms) | CSID Chn Typ %Busy /sec Time(ms) |
| ---- --- --- ----- ---- -------- | ---- --- --- ----- ---- -------- |
|
0
0 Byt
.2
0
|
|
|
0
1 Blk
16.4 30.4
5
|
|
|
0
2 Blk
19.5 25.2
7
|
|
|
0
3 Blk
23.1
0
|
|
|
0
4 Blk
22.1
0
|
|
|
0
5 Blk
.4
0
|
|
|
0
6 Blk
2.1
0
|
|
|
0
7 Blk
1.2
0
|
|
+=============================================================================+
Figure 45. Sample channel resource display (MVS/370)
The command
DISPLAY RCHN STIME(2300) ETIME(2315) YDAY
generates a display similar to Figure 46.
+============================ Channel Activity ===============================+
| Period: 23:00 to 23:15 on 09/06/99
Elap =15:00 M
SYSX |
+-----------------------------------------------------------------------------+
| CHPID
Type
%Busy
| CHPID
Type
%Busy | CHPID Type
%Busy |
| -----------| ------------ | ----- -------- |
| 00
Byt
|
|
|
| 01
Blk
1.5
|
|
|
| 02
Blk
1.2
|
|
|
| 03
Blk
.6
|
|
|
| 04
Blk
6.9
|
|
|
| 05
Blk
|
|
|
| 06
Blk
.1
|
|
|
| 07
Blk
|
|
|
+=============================================================================+
Figure 46. Sample channel resource display (MVS/XA and MVS/ESA)
CPU activity (RCPU)
The CPU resource panel shows you information about CPU activity on your system. Sample display
panels are described in the following pages. The fields on the panel are described in the table below.
Field
Description
Batch Min
Batch Max
Batch Avg
The minimum, maximum, and average number of batch jobs running during the time
period.
STC Min
STC Max
STC Avg
The minimum, maximum, and average number of started tasks running during the
time period.
TSO Min
TSO Max
TSO Avg
The minimum, maximum, and average number of TSO users during the time period.
Ready
Users In
Average number of ready users swapped in.
Ready
Users Out
Average number of ready users swapped out.
66
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Field
Description
CPU Utilization
SRB
TCB
MVS
Percent CPU used by SRBs, TCBs, and MVS for all processors. (In processors that
support logical partitioning, these values pertain to the logical CPU in the partition
where EPILOG is running. If MVS is running as a guest under VM, these values
pertain only to the CPU percentage provided by VM.)
CPUn
n = the CPU ID.
nn%
Total percent CPU utilization for CPUn.
Interrupt Rate
Number of I/O interrupts per second handled by the processor. (MVS/XA and
MVS/ESA only)
%TPI
Percent of the total I/O interrupts that were circumvented by the TPI (Test Pending
Interrupt) instruction. This number is calculated by dividing the number of interrupts
handled by the TPI instruction by the total number of interrupts for the interval. (The
total number of interrupts for the interval is interrupts handled by the TPI instruction
plus the number of interrupts handled by the interrupt handler.) (MVS/XA and
MVS/ESA only)
Average Logical
Processor
Utilization
The Average percentage of total CPU cycles assigned to this partition that this
partition used during the time period (logical partitioning machines only).
For example, if a processor complex has four processors, and only two of them are
assigned to this partition, the Average Logical Processor Utilization is calculated
from the two processors that are assigned to this partition.
Physical CPU
Utilization
The average percentage of complex-wide CPU cycles this partition used during the
time period (logical partitioning machines only).
The Physical CPU Utilization is calculated from all the processors in the processor
complex whether or not the processors are assigned to this partition. For example, if
a processor complex has four processors, and only two of them are assigned to this
partition, the Physical CPU Utilization is calculated from all four processors.
LPAR Management
Overhead
for this Partition
The percentage of logical partitioning management overhead for the partition in
which the EPILOG collector is running (logical partitioning machines only).
In addition to the fields described in the preceding table, a display will contain warning messages if one or
more intervals had missing, invalid, or changed data, such as a missing CPU speed factor or a processor
configuration change.
The command
DISPLAY RCPU STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 47.
+============================== CPU Activity =================================+
| From 11:45 to 12:00 on 08/23/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
|
Min
Max
Avg
Ready Users
CPU Utilization
|
|
------------------ -----------------------------|
| Batch
2
4
2.9
In =
1.5 SRB = 5.3 % CPU2 = 62.1 %
|
| STC
17
20
18.3
Out =
.1 TCB = 37.7 %
|
| TSO
13
20
17.2
MVS = 19.1 %
|
+=============================================================================+
Figure 47. Sample CPU resource display (MVS/370)
The command
Chapter 5. Displaying resource information
67
DISPLAY RCPU STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 48.
+============================== CPU Activity =================================+
| From 11:45 to 12:00 on 08/23/99
Elap =15:00 M
SYSX |
+-----------------------------------------------------------------------------+
|
Interrupt
|
|
Min Max
Avg Ready Users
CPU Utilization
Rate
%TPI |
|
--- --- ---- ----------- ------------------------- --------- ---- |
| Batch 17
22 19.3 In =
1.9 SRB=
6.4% CPU1 = 59.5%
284.3 1.9 |
| STC
39
44 41.6 Out =
.0 TCB= 96.6% CPU2 = 64.7%
20.5 2.6 |
| TSO
62
75 69.1
MVS= 21.1%
|
+=============================================================================+
Figure 48. Sample CPU resource display (MVS/XA and MVS/ESA)
The command
DISPLAY RCPU STIME(0614) ETIME(0629) YDAY
generates a display similar to Figure 49.
+============================== CPU Activity =================================+
| From:
06:14 to 06:29 on 12/20/99
Elap = 15:00 M
SYSA |
+-----------------------------------------------------------------------------+
|
Interrupt
|
|
Min Max
Avg Ready Users Logical CPU Utilization
Rate
%TPI |
|
--- --- ---- ----------- ------------------------ --------- ----- |
| Batch
0
1
.2 In =
1.2 SRB= 10.6% CPU1= 40.3%
246.1 .75 |
| STC
103 105 103.7 Out =
.0 TCB= 74.3% CPU2= 29.5%
97.5 1.75 |
| TSO
69
85 79.3
MVS= 29.9% CPU3= 24.0%
37.7 3.79 |
|
CPU4= 20.9%
76.1 6.49 |
|
|
|
Average Logical Processor Utilization= 28.7%
|
|
Physical CPU Utilization for this Partition= 28.7%
|
|
LPAR Management Overhead for this Partition= 3.2%
|
+=============================================================================+
Figure 49. Sample CPU resource display (with logical partitioning)
DASD device activity (RDAS)
The DASD resource panel shows you information about DASD device activity during the interval. By
default, the RDAS keyword displays data for all online DASD devices. You can, however, select specific
devices by entering the volume serial number or unit address as an operand. The volser and address
operand may contain asterisks, which serve as wildcard characters. (The use of asterisks as variable
characters is described in “Using masks for operands” on page 25.)
For example, to display all devices whose volsers begins with VSAM you would enter:
DISPLAY RDAS(VSAM*)
To select all the devices on the 14x string you would enter:
DISPLAY RDAS(014*)
The RDAS keyword matches the supplied pattern against both the volser and unit address; if either
matches, data for that device is displayed.
If this causes an ambiguity, the command can be restricted further by using the V and U suboperands. For
example, if you enter:
DISPLAY RDAS(CA*)
68
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
any device whose volser began with CA or any device on the CAx string is selected. To select only one or
the other, enter the command as:
DISPLAY RDAS(V(CA*))
for volser selection, or
DISPLAY RDAS(U(CA*))
for unit address selection.
There are four types of DASD resource displays:
v MVS/370
v MVS/XA (non-3090)
v MVS/XA (3090)
v MVS/ESA
The type of display you get depends on the system that the collector ran on. These displays are described
in the following pages. The display in Figure 50 on page 70 shows a DASD resource panel for an
MVS/370 system. The fields in this display are described below.
Field
Description
Volume Serial
Volume serial identifier.
SMF id
SMF ID of the system on which the data was collected.
Dev Adr
Device address.
I/O Rate
Rate per second at which I/O requests were issued to the device.
Total
The average end-to-end service time per I/O to the device (in milliseconds).
Path Q
The average time that an I/O spent queued on the path (in milliseconds).
Dev Q
The average time that an I/O spent queued on the device (in milliseconds).
Srv
The average time that an I/O spent transferring data to the device (in milliseconds).
% Dev Busy
Percentage of time during the time period that the device was found busy.
% RSV Delay
Percentage of time during the time period that a shared device was reserved by another
processor or a control unit was busy with another path.
Open Dsns
Average number of DCBs and ACBs concurrently open on a volume.
The command
DISPLAY RDAS STIME 0900 ETIME 0915 YDAY
generates a display similar to the following sample panel.
Chapter 5. Displaying resource information
69
+========================== DASD Device Activity =============================+
| From:
09:00 to 09:15 on 03/19/99
Elap = 14:59 M
SYSA |
+-----------------------------------------------------------------------------+
|Volume SMF Dev I/O
----Time in Milliseconds ---- % Dev
% RSV Open |
|Serial id
Adr Rate Total = Path Q + Dev Q + Srv
Busy
Delay Dsns |
|------ ---- --- ------ -------------- ---- --------- ---- |
|CAN031 SYSA 330
.7
12
0
0
12
.9
2.0|
|CAN032 SYSA 7A6
.1
21
0
0
21
.2
|
|CICS17 SYSA 9E2
0
0
0
0
2.0|
|COM001 SYSA 146
4.8
13
0
1
12
5.9
.1
25.8|
|COM002 SYSA 152
3.5
7
1
0
6
2.1
.2
81.5|
|COM003 SYSA 157
2.6
3
0
0
3
.8
.1
48.0|
|COM004 SYSA D90
1.4
17
0
1
16
2.2
|
|COM005 SYSA D85
0
0
0
0
2.0|
|HSM001 SYSA 323
.2
27
0
5
22
.4
1.0|
|IMS120 SYSA 9E5
0
0
0
0
4.0|
|IMS130 SYSA 9E3
0
0
0
0
21.0|
|MP134A SYSA 145
2.8
11
0
2
9
2.6
5.3|
|MVSA21 SYSA 150
4.1
3
1
0
2
.9
.3
|
|OMON21 SYSA 334
.3
7
0
0
7
.2
2.8|
|OMON22 SYSA 153
2.0
28
1
3
24
4.9
.1
2.1|
|OMON23 SYSA 336
0
0
0
0
4.0|
|OMON24 SYSA 322
0
0
0
0
3.0|
|OMON25 SYSA 324
0
0
0
0
12.0|
|OMON26 SYSA D88
0
0
0
0
4.7|
|OMON27 SYSA 147
.3
24
0
0
24
.7
12.5|
|PPSMPE SYSA D91
.1
18
0
0
18
.1
|
|QM0001 SYSA D96
.1
31
0
0
31
.2
2.7|
+=============================================================================+
Figure 50. DASD resource display (MVS/370)
Figure 51 on page 71 and Figure 52 on page 72 show DASD resource panels for an MVS/XA system
running on non-3090 and 3090 CPUs. Figure 53 on page 73 shows a DASD resource panel for an
MVS/ESA system. The fields in these displays are described below.
Field
Description
Volume Serial
Volume serial identifier.
SMF id
SMF ID of the system on which the data was collected.
Dev #
Device number.
LCU
Logical control unit associated with the DASD device.
I/O Rate
Average number of I/Os to the device per second.
Total
The average end-to-end service time per I/O to the device (in milliseconds).
IOSQ
The average time that an I/O spent on the IOS queue for the device (in milliseconds).
Pend
The average time that an I/O spent in the pending condition (in milliseconds).
Conn
The average connect time for an I/O to the device (in milliseconds).
Disc
The average disconnect time for an I/O to the device (in milliseconds).
CUB
Average delay to an I/O request caused by the control unit busy condition (in milliseconds).
(3090 only)
DB
Average delay to an I/O request caused by the device busy condition (in milliseconds). (3090
only)
% Dev Util
The percentage of the time during the interval that the device was in use.
Open Dsns
Average number of DCBs and ACBs concurrently open on a volume. (RMF versions prior to
4.1)
Avg Allcs
Average number of allocations issued against a device. (MVS/ESA only)
The command
70
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
DISPLAY RDAS STIME 2100 ETIME 2115 YDAY
generates a display similar to Figure 51.
+========================== DASD Device Activity =============================+
| From:
21:00 to 21:15 on 06/13/99
Elap = 15:00 M
SYSX |
+-----------------------------------------------------------------------------+
|Volume SMF Dev
I/O
------Time in Milliseconds------% Dev
Open|
|Serial id
# LCU Rate Total = IOSQ + Pend + Conn + Disc
Util
Dsns|
|------ ---- --- --- ---- ----------- ---- -----------|
|CAN031 SYSX 330
.4
11
0
0
4
7
.4
.1|
|CAN032 SYSX 7A6
19
0
0
2
17
.1|
|COM001 SYSX 146
.4
5
0
1
3
1
.2
12.0|
|COM002 SYSX 152
.8
3
0
0
2
1
.3
30.0|
|COM003 SYSX 157
.5
6
0
1
3
2
.3
20.0|
|COM006 SYSX 337
.4
33
11
0
3
19
.9
6.0|
|MP215X SYSX 332
.2
18
0
0
10
8
.4
7.0|
|MVSA21 SYSX 150
.5
4
0
1
2
1
.1
7.0|
|OMON21 SYSX D9A
.4
201
11
0
177
13
36.1
.7|
|OMON26 SYSX D88
14
0
0
2
12
|
|OMON27 SYSX 147
10
0
0
2
8
|
|QM0001 SYSX D96
2.5
65
1
0
50
14
56.7
1.7|
|SPOLA1 SYSX D9F
1.8
27
5
0
5
17
22.2
3.5|
|SYSA22 SYSX 155
.1
96
0
0
46
50
.5
|
|SYSA23 SYSX D98
.1
55
0
0
46
9
.3
|
|SYSA24 SYSX 7A3
.1
51
0
2
40
9
.3
|
|SYSA25 SYSX 333
.1
49
0
1
36
12
.3
|
|TSO021 SYSX 141
.1
7
0
1
3
3
.1
2.2|
|TSO022 SYSX 320
.1
13
0
1
2
10
.1
1.0|
|TSO023 SYSX 331
12
0
0
2
10
|
|TSO024 SYSX 7A1
18
0
1
3
14
2.5|
|TSO025 SYSX 335
.1
16
0
1
3
12
.1
.5|
+=============================================================================+
Figure 51. DASD resource display (MVS/XA non-3090)
The command
DISPLAY RDAS STIME 10 ETIME 11 SYSID(MERGE) YDAY
generates a display similar to Figure 52 on page 72.
Chapter 5. Displaying resource information
71
+========================== DASD Device Activity =============================+
| From:
10:00 to 10:15 on 10/18/99
Elap = 14:59 M
|
| Sysid:
SYSD SYSG SYSA SYSB
|
| Merge of 4 Intervals
|
+-----------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev Open|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Dsns|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- -----|
|CHKPT1 SYSA 150 00C
1.2
43
0
22
7 14
21.9 17.7 1.0|
|
SYSB 150 00A
1.0
29
0
10
6 13
9.9 17.4 1.0|
|CHKPT2 SYSD 300 00E
1
0
0
1
0
|
|
SYSA 300 00F
.7
33
0
0
10 23
2.3 1.0|
|
SYSB 300 010
.7
32
0
0
10 22
2.1 1.0|
|CIMS01 SYSD 334 00E
2
0
1
1
0
|
|
SYSB 334 010
.9
19
9
0
3
7
.9 4.0|
|CIMS02 SYSD 317 00E
2
0
1
1
0
|
|
SYSB 317 010
.7
17
6
0
3
8
.8 7.0|
|CIMS03 SYSD 333 00E
2
0
1
1
0
|
|
SYSB 333 010
3.0
6
0
0
2
4
1.7 1.0|
|COM001 SYSD D89 054
.2
16
0
4
9
3
2.2 1.6
.2 24.0|
|
SYSG D89 057
.3
14
0
5
4
5
3.0 1.9
.4 133.3|
|
SYSA D89 055
6.5
10
1
2
5
2
.2 1.1
5.0 339.5|
|
SYSB D89 056
3.2
10
1
2
4
3
.1 1.3
2.6 267.7|
|COM002 SYSD D8A 054
27
0
15
2 10 2.0 12.6
17.5|
|
SYSG D8A 057
51
0
39
8
4 1.3 37.7
145.4|
|
SYSA D8A 055
1.3
17
1
2
6
8
.2 1.7
2.7 483.1|
|
SYSB D8A 056
.6
15
0
3
6
6
.4 2.0
2.0 238.1|
+=============================================================================+
Figure 52. DASD resource display (MVS/XA 3090)
In this example, the SYSID and MERGE keywords have been used to show shared-DASD usage across
multiple systems in a multi-system environment. For further information about this feature, see “Reporting
in a multiple datastore environment” on page 106.
The command
DISPLAY RDAS STIME 10 ETIME 11 YDAY COMBINE(1H)
generates a display similar to Figure 53 on page 73.
72
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+========================== DASD Device Activity ==============================+
| From:
10:00 to 11:00 on 10/18/99
Elap = 1:00 H
SYSA |
| Combination of 4 Intervals
|
+------------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev
Avg|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Allcs|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- ------|
|CHKPT1 SYSA 150 00C
1.1
69
0
48
7 14
47.8 17.2
1.0|
|CHKPT2 SYSA 300 00F
.7
34
0
0 10 24
2.5
1.0|
|COM001 SYSA D89 055
6.8
16
4
3
6
3
.2 2.0
5.7 234.2|
|COM002 SYSA D8A 055
1.3
19
1
3
6
9
.3 2.6
3.7 368.1|
|COM003 SYSA D8B 055
2.6
12
2
2
3
5
.3 1.0
1.9 103.1|
|DB2002 SYSA 14D 00C 12.1
35
12
0
6 17
29.6 144.6|
|EPIL01 SYSA 321 00F
2.5
15
0
5
3
7
4.8
2.3
1.7|
|HSM001 SYSA 307 00F
.3
23
0
0 12 11
.8
1.2|
|HSM002 SYSA 322 00F
1.6
27
3
5
5 14
4.3 5.4
8.1|
|HSM003 SYSA 9C2 046
.2
29
0
0 15 14
.6
.1|
|IMS100 SYSA 14B 00C
6.3
26
1
1
6 18
1.0 15.2 27.4|
|IMS210 SYSA 161 00C
27
0
1
1 25
35.0|
|MIS001 SYSA 313 00F
.3
18
0
1
4 13
1.3
.6
8.3|
|MIS002 SYSA 312 00F
7.1
28
5
0
8 15
16.7 71.8|
|MIS003 SYSA 306 00F
.2
6
0
0
4
2
.1 22.6|
|MP310A SYSA D8C 055
7.0
21
5
1 11
4
.3 .6 10.0 351.5|
|MTLIB1 SYSA D80 055
17
0
1
3 13
.6
|
|MTLIB2 SYSA 15B 00C
.4
15
0
1
3 11
.7
.8 19.0|
|MTLIB3 SYSA 14C 00C
1.6
17
0
1
6 10
1.1 2.6 25.0|
+==============================================================================+
Figure 53. DASD resource display (MVS/ESA)
In this example, the COMBINE keyword has been used to combine the RDAS data from four RMF
intervals into a single one-hour interval. For a description of the COMBINE keyword, see “Combining time
intervals” on page 97.
Navigation capabilities
You can navigate from a detail line in an RDAS display to a degradation analysis panel showing workload
utilization against a DASD device. For example, suppose that you are evaluating the RDAS panel shown
in Figure 52 on page 72. You notice that device COM002 is active on four systems. (The line displaying
this information has been flagged with an arrow in the left margin.) You want to display further information
about workload utilization by system for this device. By placing the letter J or P on the first column of the
detail line for device COM002, you can navigate directly to a Shared-DASD Degradation Display for the
device. For further information, see “Shared-DASD analysis” on page 44 and “Navigating in a multisystem
environment” on page 131.
SRM domain statistics (RDOM)
The SRM domain resource panel shows you MPL and activity information for each SRM domain. A sample
panel is shown in Figure 55 on page 109. The dispaly fields on the panel are described in the table below.
By default, the RDOM keyword displays data for all domains in which there was any activity during the
interval.
Field
Description
Dmn
Domain number.
IN TAR MPL
OUT TAR MPL
Average in and out Target MPL values over the interval, as calculated by SRM. (MVS/SP
4.2 and above)
Min MPL
Max MPL
Minimum and maximum MPL values specified for the domain in the IPS.
Tar MPL
Average target MPL computed by SRM. (pre-MVS/SP 4.2)
Cur MPL
Actual MPL for the interval.
Contn Index
Contention index computed by SRM for the interval.
Chapter 5. Displaying resource information
73
Field
Description
Avg U Ready
Average number of ready users.
Users In
Average number of swappable users from the domain that were in-storage during the
interval.
Users Out
Average number of swapped-out users from the domain during the interval.
Going Out
Average number of users from the domain that were marked for swap-out.
Avg NSwap
Average number of non-swappable in-storage users during the interval.
Intvl
Srvce
Average number of SRM service units used by all members of the domain during the
interval.
The command
DISPLAY RDOM STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 54.
+=============================== SRM Domains =================================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
| Dmn
Min Max Tar Cur Contn Avg U Users Users Going
Avg Intvl |
|
MPL MPL MPL MPL Index Ready
In
Out
Out NSwap Srvce |
|
--- --- --- --- ----- ----- ----- ----- ----- ----- ----- |
|
0
255 255 255
.10
1
40 |
|
1
1 10
1
4.0
.11
.01
22 |
|
2
2 10
3
2
.8
2.6
2.71
2226 |
|
3
10 255
10
.7
.41
229 |
|
4
4
7
4
.1
.20
86 |
|
5
2
5
2
.4
.1
.17
125 |
|
6
10 12
10
3
2.9
3.00
4 1991 |
|
7
1 255
1
5
3 |
+=============================================================================+
Figure 54. Sample SRM domain resource display (Pre-MVS/SP 4.2)
To specify a subset of domains, you can optionally include a list of domains and one domain range. For
example, if you only wanted to see domains 1 through 3 and domains 5 and 7, enter the DISPLAY
command as follows (the spaces on either side of the parentheses are optional):
DISPLAY RDOM( 1:3 5 7 )
General system information (RINF)
The general information panel shows you information about your system resources. Sample panels are
shown in Figure 55 on page 75 and Figure 56 on page 76. The display fields on the panel are described in
the table below.
Field
Description
Operating System and Level
Operating system and level (for example, MVS/SP 3.1.0).
Model
CPU model.
Serial
CPU serial number.
Mode
Processor mode during the time period:
Native = Native MVS, VM = MVS under VM, VM/PMA = MVS under VM/PMA, or
Partitioned = MVS in logical partitioning mode
74
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Field
Description
IPS
The current IPS suffix.
ICS
The current ICS suffix.
OPT
The current OPT suffix.
RMF
RMF version number during the time period.
Note: The RINF report only shows the following fields on systems with logical partitioning.
Partition Name
The name assigned to the partition in which EPILOG is running.
Partition Number
The partition number in which EPILOG is running.
Physical Processors
The number of processors available to the logical partitioning processor complex.
Processor Weight
Relative importance of the logical partition with respect to the other partitions that
share the same physical CPUs. The system uses this value when more than one
logical partition is ready to use CPU resources simultaneously. The allocation of
available processing resources is influenced by the processor weight specified for
the partition.
A value of MIX in this field indicates a logical partition with both shared and
dedicated processors.
Processor
Complex
Utilization
Percent of total CPU cycles used by all of the partitions in the processor complex,
based on 100 percent per processor. For example, the base of a dyadic machine is
200.
CPU Capping
Displays whether a CPU cap is in effect for the partition in which the EPILOG
collector is running. A value of MIX indicates a logical partition with both capped and
uncapped processors.
Total LPAR
Management
Overhead
Total logical partitioning management overhead for the processor complex. This
includes overhead attributable to the partitions, as well as hypervisor overhead.
The command
DISPLAY RINF STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 55.
+======================= General System Information ==========================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
| MVS SP1.3.4 Model=3083
Serial=226543 Mode= Native
|
| IPS=83
ICS=83
OPT=83
RMF 2.4 Enhanced
|
+=============================================================================+
Figure 55. Typical system resource display
The command
DISPLAY RINF STIME(0614) ETIME(0629) YDAY
generates a display similar to Figure 56 on page 76.
Chapter 5. Displaying resource information
75
+======================= General System Information ==========================+
| From:
06:14 to 06:29 on 12/20/99
Elap = 15:00 M
SYSA |
+-----------------------------------------------------------------------------+
| MVS SP4.2.0 Model=3090
Serial=403237
Mode= Partitioned
|
| IPS=AA
ICS=AA
OPT=AA
RMF 4.2.1
|
| Partition Name=MVSA
Partition Num=1 Physical Processors=4
|
| Processor Weight=600
Processor Complex Utilization=255.1 %
|
| CPU Capping=On
Total LPAR Management Overhead= 5.3 %
|
+=============================================================================+
Figure 56. Typical system resource display (with logical partitioning)
I/O queuing (RLCU)
The RLCU resource panel shows you information about I/O queuing organized around the logical control
unit (LCU), analyzing the activity between channel paths, control units, and devices. This panel applies to
MVS/XA and MVS/ESA systems only. By default, RLCU displays data for all logical control units. You can,
however, specify selected LCUs by including an LCU name as an operand. For example, to display I/O
queuing data for logical control unit 00A, enter:
DISPLAY RLCU(00A)
You can use a generic format to specify multiple LCUs. For example, to specify all LCUs beginning with
01, enter:
DISPLAY RLCU(01*)
Only one operand is allowed with the RLCU keyword.
The RLCU display differs for non-3090 and 3090 processors. The display for non-3090 processors is
shown in Figure 57 on page 77. The display for 3090 processors is shown in Figure 58 on page 78.
The display fields on the non-3090 processor panel are described in the following table.
Field
Description
LCU
The logical control unit number representing one or more physical control units and
their associated devices.
I/Os/Sec
Shows the rate at which I/O requests were successfully selected for initiation by the
associated LCU.
Avg # I/O Q
Indicates the average number of I/O requests queued for this LCU.
% Requests Deferred:
Total
Dev Busy
CU Busy
Shows the percentage of I/O requests deferred, broken down by the reason (device
busy or control unit busy).
Total percent of I/O requests on the LCU that were deferred.
Percent of requests deferred due to a DEVICE BUSY condition.
Percent of requests deferred due to a CONTROL UNIT BUSY condition.
% All Ch Path Busy
The percentage of the time that all channel paths for this LCU were busy at the
same time.
CU
Identifies each control unit in the logical control unit (LCU). If there is more than one
channel path per control unit, a separate line is generated for each channel path.
(The CU field is filled in once for a list of channel paths.)
Chn
The channel path ID associated with this control unit.
Comment
Indicates any change to the status of the I/O configuration that may have affected
RLCU values.
The command
76
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
DISPLAY RLCU STIME(1830) ETIME(1845) YDAY
generates a display similar to Figure 57.
+====================== Logical Control Unit Activity ========================+
| From:
18:30 To 18:45 On 07/07/99
Elap =15:00 M
SYSX |
+-----------------------------------------------------------------------------+
|
% Requests Deferred
% All
|
|
I/Os Avg #
Dev
CU
Ch Path
|
| LCU /Sec I/O Q Total = Busy + Busy
Busy CU Chn
Comment
|
| --- ----- ----- ----------- ------- --- --- --------------------- |
| 00A
15.0
.00
6.87
4.38
2.49
.00 011 01
|
|
012 02
|
| 00D
11.3
.03
9.69
5.51
4.18
3.27 032 03
|
|
04
|
| 021
3.0
.00
1.44
1.07
.37
.00 072 07
|
|
073 10
|
| 023
6.0
.00
.24
.24
.00
7.89 093 11
|
| 024
.6
.00
3.88
.39
3.49
.11 0A1 12
|
| 028
9.3
.00
3.59
.84
2.75
.16 0D1 15
|
|
0D2 16
|
+=============================================================================+
Figure 57. Sample I/O queuing resource display (non-3090 processor)
The display fields on the 3090 processor panel are described in the following table.
Field
Description
I/O processor
The I/O processor identifier.
I/Os/sec
The rate per second at which I/O requests are placed on the IOP initiation queue.
Avg # I/O Q
Indicates the average number of I/O requests queued for the I/O processor.
LCU
The logical control unit number representing one or more physical control units and
their associated devices.
I/Os/sec Deferred
The rate at which the IOP placed deferred I/O requests on the control unit header
queue.
Avg #I/O Q
Indicates the average number of I/O requests queued for this LCU.
% All Ch Path Busy
The percentage of the time that all channel paths for this LCU were busy at the
same time.
% CU Busy
The percentage of I/O requests deferred for that control unit.
This value is reported even if the channel path status was changed during the
interval. (You can check the Comment field to see if any changes were made to the
channel path during the interval.) In the event of such a change, this data should be
interpreted with caution.
CU
Identifies each control unit in the logical control unit (LCU). If there is more than one
channel path per control unit, a separate line is generated for each channel path.
(The CU field is filled in once for a list of channel paths.)
CHPID
The channel path ID(s) associated with this control unit.
CHPID Ints/sec
The rate at which I/O requests for devices attached to this control unit are
completed by this channel path.
If the channel path was varied online or offline during the interval, this field is left
blank.
Comment
Indicates any change to the status of the I/O configuration that may have affected
RLCU values.
Chapter 5. Displaying resource information
77
The command
DISPLAY RLCU STIME(915) ETIME(930) YDAY
generates a display similar to Figure 58.
+====================== Logical Control Unit Activity ========================+
| From:
09:15 To 09:30 On 08/05/99
Elap =15:00 M
SYS9 |
+-----------------------------------------------------------------------------+
|
I/O
I/Os
Avg #
|
|
Processor
/sec
I/O Q
|
|
-----------------|
|
00
172.23
.01
|
|
|
|
I/Os
% All
CHPID
|
|
/sec
Avg # Ch Path
% CU
Ints
|
| LCU Deferred I/O Q
Busy
Busy CU CHPID /sec
Comment
|
| --- -------- ----- ------- ----- --- ----- ------- --------------------- |
| 005
.000
.00
.00
1.20 100
01
.09
|
|
1.00
19
.11
|
|
2.86 130
09
.11
|
|
.00
21
.09
|
| 006
.000
.00
.00
.17 101
01
.64
|
|
1.35
19
.65
|
|
.00 131
09
.76
|
|
1.66
21
.53
|
| 007
.000
.00
.00
.00 102
04
.21
|
|
2.15
19
.20
|
|
1.52 132
09
.22
|
|
83.47
24
26.19 Path varied
|
| 008
.199
.01
.00 13.54 200
02
4.79
|
|
20.33
1A
4.76
|
|
15.00 230
12
4.86
|
|
19.90
2A
4.58
|
|
|
+=============================================================================+
Figure 58. Sample I/O queuing resource display (3090 processor)
Paging and storage (RPAG)
The paging and storage resource panel shows you the swapping and paging activity for each area of
storage in your system. The display panels for different systems differ slightly. The MVS/370 display is
shown in Figure 59 on page 79. The display for MVS/XA and MVS/ESA is shown in Figure 60 on page 80.
The display fields on the MVS/370 panel are described in the following table.
Field
Description
storage areas
SQA
CSA
LPA
LSQA
Pvt
Free
Nuc
Total
On MVS/XA and MVS/ESA systems, the designation Nuc refers to the MVS nucleus
code, both above and below the 16 MB line.
Virtual
Storage
Virtual storage (in kilobytes) for each storage area shown in the display.
Real
Average real frame counts (in 4K frames).
Fixed
Average fixed frame counts (in 4K frames).
78
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Field
Description
Non-Fix
Average non-fixed frame counts (in 4K frames).
Page/Sec In
Average page-ins per second.
Page/Sec Out
Average page-outs per second.
Swap paging
/sec In
Average number of pages swapped in per second.
Swap paging
/sec Out
Average number of pages swapped out per second.
The command
DISPLAY RPAG STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 59.
+=========================== Paging and Storage ==============================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
|
|Virtual |
Average Frame Counts
|
Page/sec
| Swap paging/sec |
|
|Storage | Real = Fixed + Non-Fix| In
| Out | In
Out |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| SQA |
512K |
178 |
178 |
|
|
|
|
|
| CSA | 3032K |
151 |
51 |
100 |
.1 |
.1 |
|
|
| LPA | 4380K |
507 |
20 |
487 |
.9 |
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| LSQA|
|
197 |
197 |
|
|
|
|
|
| Pvt | 7552K | 2167 |
50 | 2117 | 5.6 | 4.3 | 42.7 | 42.8 |
| Free|
|
278 |
|
278 |
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| Nuc |
896K |
224 |
224 |
|
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|Total| 16384K | 3702 |
720 | 2982 | 6.7 | 4.4 | 42.7 | 42.8 |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|
Total Pages / second
|
11.1
|
85.6
|
+=============================================================================+
Figure 59. Paging and storage resource display (MVS/370)
The display fields on the MVS/XA and MVS/ESA panels are described in the following table.
Field
Description
storage areas
Note that there are four storage areas mapped only for MVS/XA and MVS/ESA:
EPvt
ECSA
ELPA
ESQA
These areas do not show any frame count, paging, or swapping statistics; this is
because MVS itself does not report them separately. Real storage use and paging
numbers for above-the-line virtual storage areas are reported in the corresponding
below-the-line area’s rows. For example, real storage usage for both above and
below-the-line SQA is reported in the row labeled SQA.
Installed Frames
The total amount of 4096-byte frames of expanded storage installed on the entire
processor complex (expanded storage only).
Online Frames
The amount of expanded storage online during the interval (in 4096-byte frames;
expanded storage only).
Chapter 5. Displaying resource information
79
Field
Description
Pages to ES per Sec
The number of pages per second sent to expanded storage from real storage for
paging and swapping requests (expanded storage only).
Pages Migrated per Sec
The number of pages per second migrated from expanded storage to auxiliary
storage (expanded storage only).
Available Frames:
Min
Max
Avg
The number of expanded storage frames available during the interval (expanded
storage only).
Minimum number of available frames.
Maximum number of available frames.
Average number of available frames.
The command
DISPLAY RPAG STIME(0015) ETIME(0030) YDAY
generates a display similar to Figure 60.
+=========================== Paging and Storage ==============================+
| Period: 00:15 to 00:30 on 09/09/99
Elap =15:00 M
SYSX |
+-----------------------------------------------------------------------------+
|
|Virtual |
Average Frame Counts
|
Page/sec
| Swap paging/sec |
|
|Storage | Real = Fixed + Non-Fix| In
| Out | In
Out |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|EPvt | 2021M |
|
|
|
|
|
|
|
|ECSA |
452K |
|
|
|
|
|
|
|
|ELPA | 1220K |
|
|
|
|
|
|
|
|ESQA | 8448K |
|
|
|
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| Nuc | 1340K |
338 |
338 |
|
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| SQA |
384K |
133 |
133 |
|
|
|
|
|
| LPA | 3468K |
613 |
47 |
566 |
.0 |
|
|
|
| CSA | 3116K |
92 |
9 |
83 |
.0 |
.0 |
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| LSQA|
|
211 |
211 |
|
|
|
|
|
| Pvt | 9212K | 1070 |
17 | 1053 |
.0 |
.3 | 2.5 | 2.6 |
| Free|
| 1637 |
| 1637 |
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|Total| 2048M | 4094 |
755 | 3339 |
.0 |
.3 | 2.5 | 2.6 |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|
Total Pages / second
|
.3
|
5.2
|
+=============================================================================+
|
Expanded Storage Usage
|
| Installed
Online
Pages to ES
Pages Migrated
Available Frames
|
| Frames
Frames
per Sec
per Sec
Min Avg Max
|
| --------------------------------------- --- --|
| 32K
2K
1.03
0.00
50 250
1K
|
+=============================================================================+
Figure 60. Paging and storage resource display (MVS/XA and MVS/ESA)
The expanded storage section is displayed only on expanded storage systems. It shows the following
information:
Installed Frames
The amount of installed storage (physical), which is manipulated in 4K
blocks.
Online Frames
The amount of storage defined to the system (available). This amount is
set at IPL time, and can be modified by an operator CONFIG command or
a hardware failure.
Available Frames
The amount of storage available to applications running on the system.
This amount varies with the system load.
80
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
In Figure 60 on page 80, the total amount of expanded storage installed for the processor complex is
shown in the Installed Frames column as 32K, which is equivalent to 32,768 frames, or 128 Megabytes
(32K is equivalent to 32,768; 32,768 frames multiplied by 4096 bytes equals 134,217,728, or 128
Megabytes). The amount of expanded storage online is shown in the Online Frames column as 2K, which
is equivalent to 2,048 frames, or 8 Megabytes, of storage (2K is equivalent to 2,048; 2,048 frames
multiplied by 4096 bytes equals 8,388,608, or 8 Megabytes). Out of 32,768 installed frames, 2,048 frames
are online to this partition during the interval. These figures depend on the configuration of your system.
Conversely, the available frames represent the number of frames unassigned to the partition (physical or
logical) during the interval, and will vary over time, depending on system demand and usage patterns. In
Figure 60 on page 80, between 50 and 1,024 frames were available during the interval, with an average of
250 frames.
Page data set activity (RPDS)
The page data set resource panel shows you information about page data set activity during the interval. A
sample panel is shown in Figure 61 on page 82. The display fields on the panel are described in the table
below.
Field
Description
Space
Type
Page space type. Valid types are PLPA, common, duplex, local, or swap.
Dev Adr
Device address.
VOLSER
Volume serial number.
Dev Type
Device type.
% Full
The average percent of slots used on the data set.
% In Use
Percent of the time during the interval when the data set was considered busy by
the ASM.
I/Os per sec
Average number of I/O requests for the data set per second during the interval.
Pages
per IO
Average number of pages per I/O request.
Trnsfr
Time
Average time required to complete a page transfer (in seconds).
The command
DISPLAY RPDS STIME(1145) ETIME(1200) YDAY
generates a display similar to Figure 61 on page 82.
Chapter 5. Displaying resource information
81
+========================== Page Dataset Activity ============================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
| Space
Dev
Dev
%
% In
I/O's
Pages
Trnsfr |
| Type
Adr
VOLSER
Type
Full
Use
per sec
per IO
Time |
| ------------------------ ------------ ------ |
| PLPA
150
MVSA21
3380
57.3
2.7
.9
1.1
.028 |
| Common
140
SYS021
3380
26.2
.9
.2
1.5
.030 |
| Local
143
TYPE21
3380
27.2
22.9
2.2
10.3
.010 |
| Local
153
TYPE22
3380
27.6
22.0
2.2
11.5
.009 |
| Local
141
TSO021
3380
24.7
23.2
2.3
10.6
.010 |
| Local
151
TSO022
3380
28.3
21.2
2.2
10.9
.009 |
+=============================================================================+
Figure 61. Sample page data set resource display
Performance group statistics (RPGN)
The performance group resource panel shows you the activity and status of one or more performance
groups during the interval. A sample panel is shown in Figure 62 on page 83.
The fields in this display are described in the following table. By default, the RPGN keyword displays data
for all performance groups that were active during the interval. To specify only certain performance groups,
you can optionally include a list of performance group numbers and one performance group range. For
example, if you only wanted to display data for performance groups 0 through 11 and performance groups
200 through 202, enter the command as follows:
DISPLAY RPGN(0:11,200,201,202)
The resulting display is shown in Figure 62 on page 83.
Field
Description
PGN
Performance group number.
PGP
Performance group period. If the performance group is not defined in the IPS with multiple
periods, then all activity is considered to be period 1.
SU’s/Sec
Average SRM service units used per second.
Avg Resp
Average response time for a transaction.
Avg Trx
Average number of transactions concurrently active during the time period.
Swaps/Trx
Average number of swaps per transaction during the interval.
%Res
Percent of the time that at least one transaction was in storage.
Avg WKST
Average working set size for an address space in the performance group (in kilobytes or
megabytes).
Avg In
Average number of resident address spaces from the performance group.
Total Stg
Average amount of working set storage used by the performance group over the time period (in
kilobytes or megabytes).
SRB%
Percent of CPU used by SRBs in the performance group over the time period.
TCB%
Percent of CPU used by TCBs in the performance group over the time period.
EXCPs /Sec
Average EXCPs per second for the performance group (as counted by SMF).
The command
DISPLAY RPGN STIME(1245) ETIME(1300) YDAY
82
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
generates a display similar to Figure 62.
+------------------------- SRM Performance Groups ----------------------------+
| From:
12:45 To 13:00 On 08/06/99
Elap =15:00 M
SYSX |
+-----------------------------------------------------------------------------+
|
SU's
Avg
Avg Swaps
Avg Avg Total
EXCPs|
| PGN PGP
/Sec
Resp
Trx /Trx %Res WKST * In = Stg
SRB% TCB% /Sec|
|-----------------------------------------------------------------------------|
|
0
1
69
2.0
100.0 193 2.0
386 1.6
.3
.1|
|
1
1
109
4.90S
.1
1.0 89.4 210
.1
21
.1
.8 2.9|
|
2 2734
1:47M
1.7
.3 99.0 945 1.7 1606
.9 12.3 30.4|
|
SUM
2843
1:36M
1.9
.3 98.3 904 1.8 1627
.9 13.1 33.3|
|
2
1
40
0.23S
.9 78.4
10
.4
.6|
|
2
25
1.56S
.9 100.0
5
.2
.4|
|
4
13.10S
1.0 100.0
|
|
SUM
65
0.54S
.1
.9 87.1 150
.1
15
.6 1.1|
|
5
1
342
3:40M
7.5
7.5 100.0 167 7.5 1254
.2 2.5 3.6|
|
7
1
938
3:45M
.2
100.0
.2
.2 3.5 5.6|
| 10
1
179
0.28S
.2
.9 81.9 1080
.1
108
.1 1.5 2.6|
|
2
125 13.99S
.2
.8 98.9 130
.1
13
.1 1.0 2.1|
|
4
533 39.73S
.1
1.0 100.0
.1
.1 2.4 2.1|
|
SUM
837
2.96S
.4
.8 92.5 305
.4
122
.3 4.9 7.0|
| 11
1
166
0.37S
.3
1.0 80.7 875
.2
175
.1 1.4 1.9|
|
2
182
3.23S
.3
1.0 99.5 3813
.3 1144
.1 1.2 2.2|
|
4
2 20.70S
2.0 100.0
47
|
|
SUM
349
0.66S
.5
1.0 90.8 2732
.5 1366
.2 2.6 4.2|
| 200* 1
201
1.13S
.2
.9 95.2
95
.2
19
.1 1.1 2.7|
| 201* 1
126
1.90S
.1
.4 96.9 240
.1
24
.7 1.8|
| 202* 1
36
0.30S
1.1 76.1
.3
.2|
|
|
| "*" = Report Performance Group
|
+-----------------------------------------------------------------------------+
Figure 62. Sample performance group resource display
All report performance groups are designated with an asterisk. As you can see in Figure 62, separate lines
are generated for each period of performance groups that are defined with multiple periods. The last line
for the performance group (with SUM in the PGP column) contains combined figures for all periods shown
in the display.
You can select individual performance periods by adding the PGPERIOD keyword to the command (or
PGP for short).For example, if you wanted to see data only for period 1 of performance group 2, enter the
DISPLAY command as follows:
DISPLAY RPGN(2) PGP(1)
The PGPERIOD keyword also accepts multiple operands to specify multiple performance periods. For
example, if you wanted to display data for periods 1 and 2 of performance groups 2, 10, and 11, enter the
following command:
DISPLAY RPGN(2,10,11) PGP(1,2)
The resulting display might look like Figure 63 on page 84.
Notice that the SUM line displayed for each performance group combines together data only for those
periods you have selected.
The command
DISPLAY RPGN(2,10,11) PGP(1,2) STIME(1245) ETIME(1300) YDAY
generates a display similar to Figure 63 on page 84.
Chapter 5. Displaying resource information
83
+------------------------- SRM Performance Groups ----------------------------+
| From:
12:45 To 13:00 On 08/06/99
Elap =15:00 M
SYSX |
+-----------------------------------------------------------------------------+
|
SU's
Avg
Avg Swaps
Avg Avg Total
EXCPS|
| PGN PGP
/Sec
Resp
Trx /Trx %Res WKST * In = Stg K SRB% TCB% /Sec|
|-----------------------------------------------------------------------------|
|
2
1
40
0.23S
.9 78.4
10
.4
.6|
|
2
25
1.56S
.9 100.0
5
.2
.4|
|
SUM
65
0.42S
.1
.9 87.1 150
.1
15
.5 1.1|
| 10
1
179
0.28S
.2
.9 81.9 1080
.1
108
.1 1.5 2.6|
|
2
125 13.99S
.2
.8 98.9 130
.1
13
.1 1.0 2.1|
|
SUM
304
2.64S
.3
.8 90.2 407
.3
122
.2 2.5 4.8|
| 11
1
166
0.37S
.3
1.0 80.7 875
.2
175
.1 1.4 1.9|
|
2
182
3.23S
.3
1.0 99.5 3813
.3 1144
.1 1.2 2.2|
|
SUM
347
0.62S
.5
1.0 90.7 2638
.5 1319
.2 2.6 4.2|
+-----------------------------------------------------------------------------+
Figure 63. RPGN display with performance periods
Swap data set activity (RSDS)
The swap data set resource panel shows you the activity of each swap data set during the interval. A
sample panel is shown in Figure 64. The display fields on the panel are described in the table below.
Field
Description
Dev Adr
Device address.
VOLSER
Volume serial number.
Dev Type
Device type.
% Full
Average percent of swap sets in use during the time period.
% In Use
Percent of the time during the interval when the data set was considered busy by the ASM.
IO per second
Average number of I/O requests for the data set per second during the interval.
Avg Srv Time
Average service time (in seconds per page).
The command
DISPLAY RSDS STIME(1130) ETIME(1145) YDAY
generates a display similar to Figure 65.
+========================== Swap Dataset Activity ============================+
| Period: 11:30 to 11:45 on 03/01/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
|
Dev
Dev
% In
IO per
Avg Srv
|
|
Adr
VOLSER
Type
%Full
Use
second
Time
|
|
-------------------------------|
|
331
SWAP01
3380
18.3
14.3
4.6
.031
|
|
153
SWAP02
3380
25.8
18.3
4.6
.040
|
|
332
SWAP03
3380
16.4
14.2
4.6
.031
|
|
144
SWAP04
3380
27.1
19.6
4.5
.043
|
+=============================================================================+
Figure 64. Sample swap data set resource display
SRM MPL adjustment values (RSRM)
The SRM resource panel shows you the indicators and parameters that SRM used to control the
multiprogramming level (MPL) during the interval. (The term “happy values” was formerly used as a
84
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
nickname for these indicators and their parameters.) A sample panel is shown in Figure 65. The display
fields on the panel are described in the table below.
Field
Description
SRM parameters and observed Possible values are:
values
HIGH THRESHOLD
OBSERVED MAX
OBSERVED AVG
OBSERVED MIN
LOW THRESHOLD
CPU
Average percent CPU utilization, as calculated by SRM.
UIC
Average unreferenced interval count, as calculated by SRM.
ASMQ
Average ASM queue length, as calculated by SRM.
Page Fault
Page fault rate (non-swap, non-VIO page-ins plus reclaims) in pages per second.
Dmand Pging
Demand paging rate (non-swap, non-VIO page-ins plus page-outs) in pages per
second.
Page Delay
Page delay time (excluding swap paging) in milliseconds, as calculated by SRM.
% of Frames Fixed Below 16M Percent of the first 16M of real storage that was fixed (that is, non-pageable).
% of Frames Fixed Total
Percent of total real storage that was fixed (that is, non-pageable).
The command
DISPLAY RSRM STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 65.
+========================= SRM MPL Adjustment Values =========================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
|
Page Dmand Page % of Frames Fixed |
| SRM Parameters
CPU UIC ASMQ Fault Pging Delay Below 16M
Total |
|
--- --- ---- ----- ----- ----- ---------- ------- |
|
High Threshold 101
3 1000
60
100
400
88
72
|
|
|
|
Observed Max
91 255
38
17
22
270
15
16
|
|
Observed Avg
62 255
14
8
10
155
13
14
|
|
Observed Min
46 255
0
1
1
96
13
13
|
|
|
|
Low Threshold
99
1 1000
45
75
250
82
66
|
+=============================================================================+
Figure 65. SRM resource display
System swap activity (RSWA)
The system swap resource panel shows you information about swap activity during the interval. The
displays for non-3090 and 3090 processors differ slightly. Figure 66 on page 86 shows the display for
non-3090 processors. Figure 67 on page 87 shows the display for 3090 processors.
The fields on these display panels are described in the following table.
Chapter 5. Displaying resource information
85
Field
Description
Logical
See Logical Swaps/Sec.
Physical
Rate of physical swap-outs per second (non-3090 only).
Percent logical that became
Physical
Percent of logical swaps that became physical swaps (non-3090 only).
Average Pages per Swap:
(non-expanded storage only)
Out
In
Average number of pages per swap-out and swap-in.
Average pages per swap-out.
Average pages per swap-in.
Physical Swaps/Sec: (3090
only)
Direct
Transition
Physical swaps are those made directly from central storage to auxiliary storage,
without the involvement of expanded storage. The difference between direct and
transition physical swaps depends on whether the address space was logically
swapped before being physically swapped:
Direct swaps indicate the number of swap-outs that were not marked as logically
swapped when the physical swap was made.
Transition swaps indicate the number of swap-outs that were already marked as
logically swapped when the physical swap was made.
Logical Swaps/Sec
The total number of address spaces that were marked as logically swapped during
the interval, divided by the number of seconds in the interval. (Some of these may
have subsequently transitioned to auxiliary or expanded storage during the interval.)
Expanded Storage Swaps/Sec: Shows the swap-out rate from central storage to expanded storage, and from
expanded storage to auxiliary storage (3090 only).
Direct
Trans
Direct swap-outs are those physically swapped from central storage to expanded
Migrated
storage without first being marked as logical swaps.
Transition swaps are those which were already marked as logical swaps when the
physical swap from central storage to expanded storage is done.
Migrated swaps are those in which address spaces in expanded storage are
migrated to auxiliary storage. (An address space may have arrived in expanded
storage either through a direct or transition swap.)
The command
DISPLAY RSWA STIME(1145) ETIME(12) YDAY
generates a display similar to Figure 66.
+========================== System Swap Activity =============================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
|
Swap Rate per second
Average Pages per Swap
|
| Logical= .59
Physical= .60
Out= 71.9
In= 71.7
|
| % Logical that became Physical= 4.2
|
+=============================================================================+
Figure 66. Swap activity resource display (Non-3090 processors)
In an MVS system with a 3090 processor, the command
DISPLAY RSWA STIME(7) ETIME(715) YDAY
generates a display similar to Figure 67 on page 87.
86
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+-------------------------- System Swap Activity -----------------------------+
| From:
07:00 To 07:15 On 07/15/99
Elap =15:00 M
SYS9 |
+-----------------------------------------------------------------------------+
|-----------------------Swap Rate Per Second------------------|Avg Pages/Swap |
| Physical_Swaps/Sec
| Logical | Expanded_Storage_Swaps/Sec |
|
| Direct
Transition |Swaps/Sec| Direct
Trans
Migrated | Out
In
|
| -------- ---------- |---------| ----------- -------- | ---- ---- |
|
.00
.00
|
.84 |
.00
.03
.00 | 3.5
2.4 |
+-----------------------------------------------------------------------------+
Figure 67. Swap activity resource display (3090 processor)
System swap reason (RSWR)
The swap reason resource panel breaks down the total system swap activity into swap reasons (swap
categories). Systems with expanded storage and systems without expanded storage will produce slightly
different display panels. Figure 68 on page 88 shows the display for processors without expanded storage.
Figure 69 on page 89 shows the display for processors with expanded storage.
The fields in these panels are described in the following table.
Field
Description
Swap Category
This field may include any of the swap reasons described below:
Terminal Wait - Terminal waiting for input or output buffers.
Detected Long Think - An address space that had been logically swapped out had
to be physically swapped out before it was ready to execute again. On systems
without expanded storage, “n/a” is displayed as the swap rate. On systems with
expanded storage, the category is not shown on the display panel.
Long Wait - User-requested long waits.
Detected Wait - User address spaces that have not executed for 8 RSM seconds or
2 seconds, whichever is less (in current versions of MVS), and have not issued a
long WAIT command.
Unilateral - The MPL of a domain exceeded the target MPL.
Exchange - A user in a domain that must be swapped out to allow another user in
that domain to be swapped in.
Requested - Because a user had to be swapped because of an operator command
or system action, such as VARY STORAGE OFFLINE.
Enq Exchange - To make room for users enqueued on a resource required by other
users.
Aux Storage Shortage - Because of a shortage of local page slots.
Real Storage Shortage - Because of a shortage of real pageable frames.
Transition to Nonswap - An address space has been made non-swappable.
Central Storage - Swapped out to improve central storage availability (MVS/SP 4.2
and above).
System Paging - Swapped out to reduce the system page fault rate (MVS/SP 4.2
and above).
Out Too Long - Swapped out to enable the swap-in of an address space that had
been swapped out too long (MVS/SP 4.2 and above).
APPC Wait - Swap out of an idle APPC address space (MVS/SP 4.2 and above).
Total - Total of all the above.
Chapter 5. Displaying resource information
87
Field
Description
----
On systems without expanded storage, all swap categories are shown, even if there
were no swaps of that type during the interval.
On systems with expanded storage, if there were no swaps for a particular category
during the interval, that category is suppressed on the display. If the display shows a
category with zero values for swap rates, this indicates that there were swaps of
that type during the interval, but that the rate was less than .01 when the data was
averaged over the interval.
Phys Swaps per sec:
Direct
Transition
For systems without expanded storage, this value indicates the number of physical
swaps caused by each of the above reasons.
For systems with expanded storage, physical swaps are those made directly from
real storage to auxiliary storage, without the involvement of expanded storage. The
expanded storage display breaks down physical swaps into two types: direct and
transition.
Direct swaps indicate the number of swap-outs that were not marked as logically
swapped when the physical swap was done.
Transition swaps indicate the number of swap-outs that were already marked as
logically swapped when the physical swap was made.
Logical Swaps/Second
The number of swap requests per second that at least begin as logical swaps;
expanded storage only. (Some of these requests may transition directly to auxiliary
storage, and others may transition to expanded storage.)
Expanded Storage Swaps/sec: Shows the rate of three types of swaps (in swaps per second):
Direct swaps are those which RSM sends directly from central storage to expanded
Direct
storage, without first being marked as logical swaps.
Trans
Transition swaps are those which RSM first made logical swaps, and then became
Migrated
physical swaps to expanded storage.
Migrated swaps are those which were directly sent to expanded storage, or were
transitioned to expanded storage, and later migrated to auxiliary storage.
The command
DISPLAY RSWR STIME(1145) ETIME(1200) YDAY
generates a display similar to Figure 68.
+====================== Swap Activity by Swap Reason =========================+
| Period: 11:45 to 12:00 on 08/23/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
|
Phys Swaps
Phys Swaps |
| Swap Category
per sec
Swap Category
per sec |
| ----------------------------------------- |
| Terminal Wait
1.72
Requested
.00 |
| Detected Long Think
.02
Enq Exchange
.00 |
| Long Wait
.01
Aux Storage Shortage
.00 |
| Detected Wait
.02
Real Storage Shortage
.00 |
| Unilateral
.00
Transition to Nonswap
.00 |
| Exchange
.00
----- |
|
Total
1.78 |
+=============================================================================+
Figure 68. Swap reason resource display (non-expanded storage)
In an MVS system with expanded storage, the command
DISPLAY RSWR STIME(7) ETIME(715) YDAY
88
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
generates a display similar to Figure 69.
+---------------------- Swap Activity By Swap Reason -------------------------+
| From:
07:00 To 07:15 On 07/15/99
Elap =15:00 M
SYS9 |
+-----------------------------------------------------------------------------+
|
| Physical_Swaps/Sec | Logical |Expanded_Storage_Swaps/Sec|
| Swap Category
| Direct
Transition |Swaps/Sec| Direct Trans Migrated |
| ------------| --------------- |---------| ------ ------ -------- |
| TERMINAL WAIT
|
.00
.00
| .74 | .00
.02
.00 |
| LONG WAIT
|
.00
.00
|
.03 | .00
.00
.00 |
| DETECTED WAIT
|
.00
.00
| .06 | .00
.01
.00 |
| UNILATERAL
|
.00
.00
|
.00 | .00
.00
.00 |
+-----------------------------------------------------------------------------+
Figure 69. Swap reason resource display (with expanded storage)
Virtual Lookaside Facility statistics (RVLF)
The Virtual Lookaside Facility (VLF) resource panel displays information about each VLF class. This
display applies to MVS/ESA systems only.
One of the ways to improve performance in MVS/ESA is I/O avoidance. It is faster to retrieve data from
virtual storage than from physical devices. VLF was introduced in MVS/SP 3.1 to provide the capability to
avoid I/O by caching user-defined data objects in a dataspace, where they are managed according to
class. SYS1.PARMLIB member COFVLxx defines which objects are in each class, and how the class is
managed.
You can tune VLF to maximum effectiveness by using the RVLF display to evaluate the performance of
each class definition and storage allocation. You must balance the performance gains through I/O
avoidance against the cost of virtual storage and increased paging.
The fields on the display panel are described in the following table.
Field
Description
VLF Class
The name of this class.
Virtual Storage Pages Used
Shows the number of pages of virtual storage used by this class.
Virtual Storage Pages
Maximum
Shows the maximum number of virtual storage pages that this class can use.
Virtual Storage Pages Object
The size of the largest object which VLF attempted to load into virtual storage during
the interval.
Caching Activity Add
The number of objects in this class added to virtual storage during the interval.
Caching Activity Delete
The number of objects in this class deleted from virtual storage during the interval.
Caching Activity Trim
The number of objects in this class trimmed from virtual storage during the interval.
Caching Activity Reads/sec
The number of times storage was searched for an object in this class during the
interval, divided by the number of seconds in the interval, expressed in tenths of a
second.
Caching Activity Hit %
The number of times an object in this class was found in cache divided by the
number of times storage was searched for objects in this class, expressed as a
percentage in tenths of a percent.
Note: *SYSTEM is a total for all classes (and an average for Hit % of all classes).
The command
DISPLAY RVLF RIF(Class = *)
Chapter 5. Displaying resource information
89
generates a display similar to Figure 70.
+==================== Virtual Lookaside Facility Classes =====================+
| From:
00:14 to 00:29 on 08/30/99
Elap = 14:57 M
SYSH |
+-----------------------------------------------------------------------------+
| VLF
- Virtual Storage Pages --------- Caching Activity ---------- |
| Class
Used
Maximum
Object
Add
Delete Trim
Reads/sec Hit % |
| --------------------------- ------ ------ ------ ----- |
| CSVLLA
1100
2048
850
1122
822 1112
4.4
84.7 |
| IGGCAS
14
4096
8
14
1
12
4.6
99.3 |
| IKJEXEC
7844
8192
244
6442
344
6
.3
72.1 |
| SPFLIBS
280
8192
44
93
0
0
14.2
98.1 |
| IRRGTS
458
2048
2
12
0
2
1.1
88.4 |
| --------------------------- ------ ---------- ----- |
| *SYSTEM
10638
24576
850
8105
1210 1126
142.1
90.9 |
+=============================================================================+
Figure 70. RVLF resource display (MVS/ESA)
90
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 6. Using the reporter
This chapter shows various ways that you can use the EPILOG reporter to customize the way that you
select, format, or view historical data.
Selecting data
You to select the data that will be shown in response to a DISPLAY command. You can set a date and
time period to be reported; you can select performance records to be shown; and you can set defaults for
data selection.
Setting dates and times
The reporter provides a number of date and time keywords to allow you to tailor the displays to a
particular period or periods. This feature reduces the amount of data on the screen and improves the
clarity of the display. If you omit date/time keywords from the DISPLAY command, the reporter displays the
requested data for every interval in the datastore, generally resulting in long and expensive displays that
are difficult to read.
You can specify the time period in a number of different ways:
v You can display data over a specified period of time using the STARTDATE, STARTTIME, ENDDATE,
ENDTIME, and EXDATE keywords. The RANGE and BAND keywords determine whether to treat the
specified period as a continuous period or as a series of periods, respectively. For example, if you
enter:
DISPLAY... STARTDATE(9/5/99) STARTTIME(0900) ENDDATE(9/10/99) ENDTIME(1700) RANGE
the data display will begin at 9:00 AM on 9/5/99 and end at 5:00 PM on 9/10/99. If you had specified
BAND instead of RANGE, the display would only contain data gathered between the hours of 9:00 AM 5:00 PM on the specified days. BAND is the default value.
v You can limit the data display to certain days of the week using the DAYOFWK keyword. Its operands
let you limit the data to one or more days (MONDAY, TUESDAY, WEDNESDAY, THURSDAY, FRIDAY,
SATURDAY, SUNDAY, or ALL), or to weekdays (WEEKDAYS) or weekends (WEEKEND).
v You can use any of the following calendar date keywords:
TODAY
THISWEEK
THISMONTH
THISYEAR
YESTERDAY
LASTWEEK
LASTMONTH
LASTYEAR
LASTWEEK and THISWEEK extend from Sunday to Saturday of the specified week; LASTMONTH and
THISMONTH extend from the first day to the last day of the specified month.
The THISYEAR and LASTYEAR keywords cover long time periods and therefore may take a long time
to process when used with the DISPLAY command.
v You can specify relative dates and times by including a plus sign (+) or minus sign (-) with the
STARTDATE, STARTTIME, ENDDATE, ENDTIME, and EXDATE keywords.
These keywords are summarized in Table 10 on page 92, and some examples of how they interact are
provided in the next section.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
91
Table 10. Date and time keywords
Keyword
Short
Description
Operand
BAND
BAND
Displays data between the start and end times on the
specified days (default value).
none
DAYOFWK
DAY
Displays data only for the specified days. The operands
used with this keyword can be abbreviated to an
unambiguous short form. For example, WEDNESDAY can
be abbreviated to WED or just W, but SATURDAY can
only be shortened to SA. (This keyword may not be used
when deleting records from the Profile datastore with the
EXCLUDE command.)
ALL
WEEKDAY or WKDAY
WEEKEND or WKEND
MONDAY
TUESDAY
WEDNESDAY
THURSDAY
FRIDAY
SATURDAY
SUNDAY
ENDDATE
EDATE
Specifies the end date of the display. If no ENDDATE is
specified, data is selected up to the most recent record in
the database.
Date in any of the formats
shown following the table.
ENDTIME
ETIME
Specifies the end time of the display. Defaults to 23:59:59
on the specified ENDDATE.
Time in any of the formats
shown following the table.
EXDATE
EXD
Excludes the specified date(s) from the display.
Date(s) in any of the
formats shown following
the table.
LASTMONTH
LMN
Displays data from the first day to the last day of the
previous month.
none
LASTWEEK
LWK
Displays data from Monday to Sunday of the previous
week.
none
LASTYEAR
LYR
Displays data from the first day to the last day of the
previous year.
none
RANGE
RANGE
Displays data from the start time and date to end time and none
date.
STARTDATE
SDATE
Specifies the start date of the display. If no STARTDATE is Date in any of the formats
specified, data is displayed from the first record in the
shown following the table.
database.
STARTTIME
STIME
Specifies the start time of the display. If no STARTTIME is Time in any of the formats
specified, data is displayed from the first record on the
shown following the table.
specified STARTDATE.
THISMONTH
TMN
Displays data from the first day of the current month to the none
most recent record in the database.
THISWEEK
TWK
Displays data from the first day of the current week to the
most recent record in the database.
none
THISYEAR
TYR
Displays data from the first day of the current year to the
most recent record in the database.
none
TODAY
TDAY
Displays data from the beginning of the current day to the
most recent record in the database.
none
YESTERDAY
YDAY
Displays data from the beginning of the previous day to
the last record of the previous day.
none
Date and time formats
Dates can be entered in either of two formats:
v Julian
92
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
yyddd (ddd = 001 through 366)
v Gregorian
mmddyy
mm/dd/yy
mm/dd
(assume the current year)
dd
(assume the current month and year)
Note: If slashes are used with Gregorian dates, leading zeros are not required (for example, 7/2/99 is an
acceptable date entry). If slashes are not used, leading zeros are required.
Times must be entered in 24-hour format (that is, 1300 instead of 1:00 PM), and can be abbreviated as
follows:
h
One-digit hour without a leading zero
hh
Two-digit hour
hmm or h:mm
Hours and minutes without a leading zero
hhmm or hh:mm
Hours and minutes
hhmmss or hh:mm:ss
Hours, minutes, and seconds
Sample time period statements
This section provides some examples of how the time period keywords interact.
DISPLAY... SDATE(9/13/99) DAY(MON,THUR) STIME(0900) ETIME(1300) BAND
Displays data collected on Mondays and Thursdays between the hours of 9:00 AM and 1:00 PM. The
display begins on September 13, 1999 and continues through the present.
DISPLAY... SDATE(99011) STIME(0900) EDATE(99018) ETIME(1700) RANGE
Displays data from 9:00 AM on January 11, 1999 until 5:00 PM on January 18, 1999.
DISPLAY... DAY(WKDAY) SDATE(7/1/99) STIME(0900) EDATE(7/10/99) ETIME(1700) EXDATE(7/4/99) BAND
Displays data collected on weekdays from 9:00 AM - 5:00 PM starting July 1, 1999 until July 10, 1999.
Data from July 4 is excluded from the display.
Relative dates and times
You can specify relative dates and times by including a plus sign (+) or a minus sign (-) before the
expression to be evaluated.For example, to enter a start date thirty days prior to the current date, enter:
STARTDATE(-30)
Each unit within a time or date expression can be independently assigned a relative value. For example:
STARTDATE(-1/+1)
specifies a start date one month prior to tomorrow. (The -1 in the month position means subtract 1 from
the current month; the +1 in the day position means add one to the current day.)
If no sign is entered before a unit, it is assumed to be absolute rather than relative. For example:
Chapter 6. Using the reporter
93
STARTDATE(-1/15)
specifies a start date of the 15th day of the previous month. An entry of +0 or -0 defaults to the current
value. For example:
STARTTIME(+0:+15)
specifies a start time of fifteen minutes after the current hour. If the sign is omitted before the zero, it is
interpreted as an absolute value. For example:
STARTTIME(-1:00)
specifies a start time of one hour (on the hour) before the current hour. If this command were entered at
10:45 AM, the start time would be 9:00 AM.
If a relative value in one unit causes a change in another unit, that change “wraps around” and increments
(or decrements) the second unit accordingly. For example, if you entered this start date on March 15:
STARTDATE(-1/-15)
the resulting start date would be January 31. The month is evaluated first and decremented to February.
The date is evaluated next and comes out to zero. Since zero is below the range of valid dates (1 - 31),
the month is decremented again (to January), and the date is set at the last day in January. If you had
entered:
STARTDATE(-1/-16)
the resulting start date would be January 30. This wrap-around feature only takes place between units
entered in the same keyword; an excessively high (or low) value in STARTTIME will not increment (or
decrement) any values in the STARTDATE keyword, nor vice versa.
The relative time and date values provide a great deal of flexibility. However, as you can see from some of
these examples, the evaluation of these values can become confusing when several relative values are
used in the same command. It is generally best to use them sparingly until you have become accustomed
to the way they interact. Here are some general guidelines:
v Relative date values are evaluated in the following order: year, month, day. Relative time values are
evaluated in hour, minute, second order.
v A value is considered absolute if it is not preceded by a plus or minus sign.
v A +0 or -0 entry defaults to the current value (current day, current hour, and so on). If you enter a 0 that
is not preceded by a plus or minus sign, it is interpreted as an absolute value.
v Values entered must be valid for the unit they represent. The valid values are:
Years
1 - 99
Months
1 - 12
Days
1 - 31
Hours
0 - 23
Minutes
0 - 59
Seconds
0 - 59
v Values entered in the same keyword will “wrap around” if they are found to be outside allowed
boundaries. The wrap-around feature does not apply to values entered in different keywords.
You can set defaults for the time/date keywords with the SET command. This is described in “Setting
display defaults” on page 95.
94
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Setting exception thresholds
It can be very tedious to have to search through reporter output looking for periods when response time
was bad or when a particular type of resource bottleneck was significant. To relieve this problem, the
DISPLAY command accepts exception filters which select only those records that meet certain selection
criteria.
In performance group analysis, for example, you may want to select only intervals where response time
was greater than two seconds, or where private page-in waits accounted for less than 20 percent of the
total transaction’s lifetime. Likewise, for batch jobs, you may only be interested in analyzing long-running
jobs (say, where the elapsed time exceeds 1 hour), or those executions of a particular program in which
the I/O waits accounted for more than 25% of the program’s run time.
You specify exception filters by adding the SELECTIF or REPORTIF keyword to the DISPLAY command,
along with a valid wait reason and appropriate selection criteria. (SELECTIF and REPORTIF can be
shortened to SIF and RIF, respectively.)
For further information concerning the use of exception thresholds, see Part 2, “Advanced reporting
features,” on page 135.
Setting display defaults
When using the DISPLAY command you will often find that you are entering the same set of options (such
as start/end date and time) over and over. To reduce the amount of typing required, the SET command
allows you to establish defaults for certain keywords. For example, if you intended to issue a series of
DISPLAY commands targeted at yesterday from 9:00 AM to 5:00 PM, you might want to issue the
following SET command first:
SET YESTERDAY STIME(9) ETIME(17)
From now on, anytime you issue a command, EPILOG will assume you have also entered the clause
YESTERDAY STIME(9) ETIME(17). If you enter another time or date frame with a given command, the
specified keywords override the default settings for that command.
The current default settings can be viewed at any time by entering the SET command without any
keywords. Adding the keyword CLEAR clears all default settings (as if the reporter session had been
stopped and another session started.)
Not all DISPLAY keywords can be SET; those that can fall into three categories:
v All date and time specification keywords (STIME, BAND, THISMONTH, and so on; see “Setting dates
and times” on page 91.)
v Exception analysis keywords (RESPONSE, AIO, PAG, and so on; see Table 8 on page 58.)
v Various display and batch reporting options. These options are outlined in Table 11 on page 96.
Note: When multiple SET commands are issued with certain keywords, such as SYSID or TITLE, the
operands are appended. For example, if you enter
SET SYSID (SYSA) SET SYSID (SYSB)
the result is:
SET SYSID (SYSA, SYSB)
Chapter 6. Using the reporter
95
Table 11. Keywords valid with SET
Keyword
Short
Description
Operand
AMATRIX
none
Specifies an Automatic Analysis matrix member of rhilev.REDPARM. (The 1–7 chars
A-matrix is used to associate resource panels with detail wait reasons, and
is invoked during system navigation, with the AUTO keyword, and with the
RESOURCE command.) By default, these displays use the @DEFAULT
member of rhilev.REDPARM; the SET command allows you to specify
another member.
AUTOMATIC
AUTO
Displays resource information specified by AMATRIX for highest wait
reason (only applies to DETAIL reports).
none
LIMIT
LIM
Limits the number of records matching selection criteria. Zero means “no
limit.” (Default = 10 for batch, 0 for full-screen.)
0–32767
LOGOFF
none
Turns off logging to EPLOG.
none
LOGON
none
Writes all command output to EPLOG.
none
PLOTMIN
PMIN
Sets the percentage threshold for wait reasons. (Default=5)
0–99
PRDXDEF
none
Specifies a productivity index definition member of rhilev.REDPARM. (The
productivity index definition is used to calculate the productivity index
shown on workload degradation displays, and the member is invoked
whenever a degradation display is requested.) By default, degradation
displays use the DEFAULTP member of rhilev.REDPARM; the SET
command allows you to specify another member.
1–7 chars
PRODIDX
none
Turns on the productivity index display.
none
NOPRODIDX
none
Turns off the productivity index display.
none
SHORT
none
Indicates that title block should be in condensed format.
none
SUMDEF
none
Specifies a summary wait definition member of rhilev.REDPARM. (The
summary wait definitions are used to group detailed wait reasons into
summary wait categories, and the member invoked when the SUMWAIT
keyword is used.) By default, SUMWAIT displays use the DEFAULTS
member of rhilev.REDPARM; the SET command allows you to specify
another member.
1–7 chars
SUMWAIT
none
Displays summary wait categories instead of detailed wait reasons on a
workload degradation display.
1–7 chars
SYSID
none
SMF System ID. (Default is to accept any SMF ID.)
1–4 chars
TITLE
none
Single spaced report title enclosed in quotes. More than one title may be
entered.
1–72 chars
TITLE2
none
Double spaced report title enclosed in quotes. More than one title may be
entered.
1–72 chars
TITLE3
none
Triple spaced report title enclosed in quotes. More than one title may be
entered.
1–72 chars
TWAITOFF
none
Excludes terminal input wait swaps (default).
none
TWAITON
none
Includes terminal input wait swaps (default is TWAITOFF).
none
The following keywords cannot be used with SET: AVERAGE, COMBINE, DETAIL, XPG, STEP,
SUMMARY, TOTAL, workload identifiers (PGN, JOB, and so on), and all resource panel names (RCPU,
RINF, and so on).
96
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Combining time intervals
Degradation and resource data that has been collected over multiple time intervals can be summarized
using the COMBINE keyword. Performance groups and system resources are monitored continuously, and
data for them is collected and stored in increments corresponding to the RMF interval. The COMBINE
keyword is used to summarize this data over longer periods of time. Since batch workloads do not run
continuously, however, this method of summarization does not apply to them. COMBINE summarizes
batch workload data according to the keywords you use in the command; that is, by job name, program,
account code, or class.
Combining performance group and resource data is described below. Combining batch workload data is
described in “Batch jobs, started tasks, and TSO user sessions” on page 100.
Performance groups and system resources
The fundamental unit of monitoring and reporting on an MVS system is usually the RMF interval. The
length of this interval varies from installation to installation, but generally ranges from 15 minutes to 1 hour.
Performance group and resource records are collected by EPILOG for MVS at RMF intervals (if RMF is
running), or at intervals set by the user at installation time (if RMF is not running). The fact that collection
is kept synchronized with RMF makes it easier to match and compare performance data.
Unfortunately, the RMF collection interval length is not always a convenient or desirable unit for reporting
performance data; displaying RMF data broken down to the RMF interval level may provide more detail
than is really desirable.
In light of this, the SINGLE and COMBINE keywords allow you to display performance group and resource
data for individual RMF intervals, or to combine RMF intervals into larger intervals, respectively. (The
SINGLE and COMBINE keywords are mutually exclusive; SINGLE is the default and has been in effect in
all the previous examples.) For example, you might want to generate a SUMMARY report for yesterday’s
TSO period 1 data. To do so, you would enter the following command:
DISPLAY SUMMARY PGN(2) PGP(1) YESTERDAY
The resulting display would look like the following.
+=============================================================================+
| Performance Group = 2
Symbolic Name = DEV TSO
Period 1
|
| From
07:17 to 23:57 on 08/22/99
|
+-----------------------------------------------------------------------------+
|DATE__START__END___MAIN_REASON(*)_TIME(-)|0S____0.2____0.4____0.6____0.8____1.
|08/22 07:17 07:30 Swap Page-in
0.18 S|***--->.
.
.
.
|
|
07:30 07:45 Using CPU
0.25 S|**------->
.
.
.
|
|
07:45 08:00 Using CPU
0.17 S|**---->.
.
.
.
|
|
08:00 08:15 Term Out Wait
0.23 S|***----->
.
.
.
|
|
08:15 08:30 ECB Wait
0.22 S|***----->
.
.
.
|
|
08:30 08:45 Term Out Wait
0.31 S|****-------> .
.
.
|
|
08:45 09:00 Term Out Wait
0.39 S|******------->.
.
.
|
|
09:00 09:15 Swap Page-in
0.40 S|***----------->
.
.
|
|
09:16 09:30 Swap Page-in
0.33 S|***---------> .
.
.
|
|
09:30 09:45 Swap Page-in
0.33 S|****-------> .
.
.
|
|
09:45 10:00 Swap Page-in
0.31 S|****-------> .
.
.
|
|
10:00 10:15 Swap Page-in
0.28 S|***------->
.
.
.
|
|
|
|
etc...
|
|
|
|
...
|
|
|
+=============================================================================+
Chapter 6. Using the reporter
97
In its default SINGLE interval form, this report would be 96 lines long; perhaps too detailed a display for
casual perusal. Instead you can ask EPILOG for MVS to combine this data into 1-hour combined intervals;
that is, EPILOG will present the data as if the RMF intervals had actually been one hour long.
You do so by including the COMBINE keyword with an operand specifying one hour:
DISPLAY PGN(2) PGP(1) SUMMARY YESTERDAY COMBINE(1H)
EPILOG responds with a summarized display, as shown in Figure 71.
+=============================================================================+
| Performance Group = 2
Period 1
|
| From 07:00 to 23:57 on 08/22/99
SYSA |
+-----------------------------------------------------------------------------+
|DATE__START__END___MAIN_REASON(*)_TIME(-)|0S____0.2____0.4____0.6____0.8____1.
|08/22 07:17 08:00 Swap Page-in
0.20 S|**----->
.
.
.
|
|
08:00 09:00 Term Out Wait
0.29 S|****------>
.
.
.
|
|
09:00 10:00 Swap Page-in
0.34 S|***---------> .
.
.
|
|
10:00 11:00 Long Wait Swp
0.38 S|*****-------->.
.
.
|
|
11:00 12:00 Swap Page-in
0.44 S|***------------>
.
.
|
|
12:00 13:00 Swap Page-in
0.31 S|***--------> .
.
.
|
|
13:00 14:00 Prvate Pagein
0.37 S|***---------->.
.
.
|
|
14:00 15:00 Prvate Pagein
0.79 S|******--------------------->.
|
|
15:00 16:00 Prvate Pagein
0.52 S|*****-------------> .
.
|
|
16:00 17:00 Long Wait Swp
0.48 S|******-----------> .
.
|
|
17:00 18:00 Swap Page-in
0.26 S|**------->
.
.
.
|
|
18:00 19:00 Using CPU
0.17 S|**---->.
.
.
.
|
|
19:00 20:00 Swap Page-in
0.15 S|***--> .
.
.
.
|
|
20:00 21:00 Swap Page-in
0.22 S|**------>
.
.
.
|
|
21:00 22:00 Swap Page-in
0.23 S|****---->
.
.
.
|
|
22:00 23:00 Swap Page-in
0.20 S|***---->
.
.
.
|
|
23:00 23:57 Using CPU
0.18 S|**---->.
.
.
.
|
|--------------------------------------|
|Average for this Workload Display 0.33 S
|
| Productivity Index
22%
|
+=============================================================================+
Figure 71. Combined performance group summary display
The START and END columns on this display reflect the start and end times of the actual RMF intervals at
the beginning and end of the combined interval. For example in Figure 71, the very first data line shows
07:17 to 08:00 rather than 07:00 to 08:00; this is to call your attention to the fact that the combined interval
is not necessarily completely filled with data.
If you do not supply a combined interval length with the COMBINE keyword, then all the intervals between
the start time and end time are combined into one display. For example, to see a COMBINEd resource
display of CPU activity from 9:00 AM to 5:00 PM, enter:
DISPLAY
RCPU
YESTERDAY STIME(9) ETIME(17) COMBINE
The result is a single display panel for the entire day, as shown in Figure 72.
+============================== CPU Activity =================================+
| From 09:10 to 05:05 on 08/22/99
Elap =16:30 H
SYSA |
| Combination of 65 Intervals
|
+-----------------------------------------------------------------------------+
|
Min Max
Avg
Ready Users
CPU Utilization
|
|
--- --------------------------------------------|
| Batch
0
8
2.3
In =
1.4
SRB = 3.3 % CPU2 = 40.6 %
|
| STC
13
25
22.8
Out =
.0
TCB = 27.3 %
|
| TSO
0
29
11.3
MVS = 10.0 %
|
+=============================================================================+
Figure 72. Combined resource display
98
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The header block of this display tells you how many single RMF intervals are summarized in the combined
display (65).
Not all the numbers shown are just a straight average of the component intervals. For example, in the
CPU activity panel shown here, the min and max active jobs are not the average of all the component
intervals but in fact are the lowest minimum and the highest maximum seen among the single RMF
intervals. Every attempt is made to generate a display which represents what you would have seen had
the actual RMF interval been (in this case) 16 hours and 30 minutes long.
The time interval length specified as an operand of COMBINE should generally be an even multiple of
your RMF collection interval length. Using an interval which is not an even multiple, while allowed, can
often (not surprisingly) cause displays which are irregular, confusing, or difficult to use. You should also not
use an interval shorter than your RMF collection interval. If you are specifying your interval in hours, we
recommend that you use a number that divides evenly into 24. For example: 1, 2, 3, 4, 6, 8, or 12.
Obviously, indiscriminate use of COMBINE can cause EPILOG to read and process a large amount of
data. When combining weeks or months worth of data together you might consider using batch mode
rather than doing it online.
Combined resource displays
Special considerations arise when a resource has been reconfigured during a combined interval. Consider
the following example as it relates to SRM domain data displayed with the RDOM keyword.
Suppose your installation has domain 15 defined with a minimum MPL of 10 and a maximum of 255. At
12:00 a systems programmer changes the min and max MPL to 1 using the SETDMN command. What
should EPILOG do when asked to combine intervals before and after 12:00? Averaging domain 15’s min
and max MPL is clearly the wrong thing to do.
To avoid this problem, EPILOG simply treats “domain 15 with min MPL=10 and max MPL=255” and
“domain 15 with min and max MPL=1” as two separate entities; each is given its own separate line in the
combined display, and they are not averaged together. (Note that the new domains will not necessarily be
in ascending sequence; the more recent domain 15 line will generally appear at the bottom of the RDOM
display, after all the older domain definitions.)
The same situation occurs with a number of other resource types, and the solution is similar. For example,
in the RDAS display, it is not unit address or volser alone that determines a unique device. Entries from
different intervals are only combined if both the unit address and volser match. The following table defines
the components of a unique entry for each resource panel.
Panel
Definition
RCCH
N/A
RCHN
Channel number
RCPU
CPU ID (If the CPU ID, the CPU speed factor, or the number of CPUs is changed, a message to
that effect is displayed with the panel.)
RDAS
Unit address and volser
RDOM
Domain number and min and max MPLs
RLCU
CPU model (309x and non-309x)
RINF
N/A
RLCU
N/A
RPAG
Unit address and volser
RPDS
Page data set name, volser, device type, device address, and page space type (PLPA, common,
etc.)
Chapter 6. Using the reporter
99
Panel
Definition
RPGN
Performance group number
RSDS
Swap data set name, volser, device type, and device address
RSRM
N/A
RSWA
N/A
RSWR
N/A
RVLF
N/A
Batch jobs, started tasks, and TSO user sessions
COMBINE can also be used to summarize batch jobs, started tasks, and TSO sessions. The principal
difference between combining these types of workloads and performance groups and resources is that
COMBINE doesn’t accept an operand when used with jobs, tasks, or TSO sessions.
Remember that these workloads can be displayed at different levels of detail. (Multi-step jobs, started
tasks, and TSO sessions can be displayed at the step level as well as at the job, task, or session level.
Also, if data was collected at RMF-based intervals, the INTERVAL keyword displays these workloads at
the interval level.)
The way in which these workloads are combined with the COMBINE keyword is determined by the level at
which they are being displayed.
Here are some general guidelines for using COMBINE with these workloads. Following the guidelines,
several examples are provided that illustrate COMBINE processing with batch jobs, started tasks, and
TSO user sessions.
v If the INTERVAL keyword is used to display data by RMF-based intervals, then the data is reported by
RMF-based intervals, and COMBINE is not allowed.
v If the INTERVAL keyword is not included on the DISPLAY command, but data was collected by
intervals, the data is automatically combined into a single display panel, according to the time/date
range specified on the command. The COMBINE keyword in this instance is superfluous.
v If the data was not collected by intervals and data is being displayed at the step level (by use of the
ACCOUNT, PROGRAM, or CLASS keyword) and the COMBINE keyword is used, then all steps that
meet the criteria are combined into a single display panel. (Note that steps can only be combined when
ACCOUNT, PROGRAM, or CLASS is used without JOB, STC, or TSO. This is because the STEP
keyword, which is used with JOB, STC, and TSO to display data at the step level, is not allowed with
the COMBINE keyword.)
v If the data was not collected by intervals and data is being displayed at the job, started task, or TSO
session level, then all jobs, tasks, or sessions that meet the criteria are combined into a single display
panel.
Examples
DISPLAY JOB(INV*) COMBINE LASTWEEK STIME(9) ETIME(17)
Displays degradation data for all batch jobs beginning with INV that ran last week between the hours of
9:00 AM and 5:00 PM. The data is combined into a single display panel.
DISPLAY STC(CICS) YDAY STIME(13) ETIME(17) INT
Displays data for started task CICS from 1:00 PM to 5:00 PM yesterday. Data is displayed in intervals.
(That is, a separate display panel is generated for each interval.)
DISPLAY PGM(POSTDB) THISWEEK COMBINE
100
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Generates one display panel that combines all batch job steps, started task steps, and TSO session steps
that executed program POSTDB this week.
DISPLAY CLASS(A) TODAY INTERVAL
Displays degradation data for all batch jobs that ran in class A today. The data is displayed in intervals.
(That is, a separate display panel is generated for each interval.)
A sample combined job level display is shown in Figure 73.
+=============================================================================+
| Job = USER04A
Average Input Queue =
4:23 M |
| From 05:32 On 09/09/99 To 13:28 On 09/13/99
SYSA |
| Average Of 162 Jobs
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
7.75 S
7.6|--->.
.
.
.
.
.
.
.
.
.|
|Waitine For CPU
23.26 S 22.8|---------> .
.
.
.
.
.
.
.|
|Swapped With WTOR
15.50 S 15.2|------> .
.
.
.
.
.
.
.
.|
|Disk PAY031 330 Act 12.65 S 12.4|---->
.
.
.
.
.
.
.
.
.|
|Disk PAY027 147 Act 6.94 S
6.8|--> .
.
.
.
.
.
.
.
.
.|
|SYSDSN
Enqueue
6.53 S
6.4|--> .
.
.
.
.
.
.
.
.
.|
|Waiting For MVS Lock
5.10 S
5.0|--> .
.
.
.
.
.
.
.
.
.|
|Average Job Time
1:41 M
|
|Productivity Index
30%
|
+=============================================================================+
Figure 73. Sample combined job level display
Notice that the screen header shows you the beginning and end of the combined time period, the number
of jobs summarized in the display, and the average time spent on the input queue.
A sample combined step level display is shown in Figure 74.
+=============================================================================+
| Program = IEBGENER
|
| From 07:34 To 17:33 On 09/16/99
SYSA |
| Average Of 26 Jobs
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
3.71 S
1.9|.
.
.
.
.
.
.
.
.
.
.|
|Tape Mount Pending
1:09 M 35.7|------------==> .
.
.
.
.
.
.|
|Waiting For CPU
42.55 S 21.8|-------->
.
.
.
.
.
.
.
.|
|Disk PAY023 336 Act 15.03 S
7.7|--->.
.
.
.
.
.
.
.
.
.|
|Disk SYS024 140 Act 14.25 S
7.3|--> .
.
.
.
.
.
.
.
.
.|
|Swapped With WTOR
10.15 S
5.2|--> .
.
.
.
.
.
.
.
.
.|
|Average Job Step Time 3:15 M
|
|Productivity Index
20%
|
+=============================================================================+
Figure 74. Sample combined step level display
In addition to the time period of the combined display, the screen header also shows how many runs of
the started task were used to construct the display. (A started task “run” occurs each time the task is
stopped and started.)
Screen display functions
You can tailor the way that performance data is displayed. This section provides instructions for scrolling
through display output, displaying the current command, logging display output, and setting up a color
terminal.
Chapter 6. Using the reporter
101
Scrolling through reporter output
The reporter presents output in the form of display panels. For resource and performance group displays,
these panels are based on the RMF interval, or on a different interval specified with the COMBINE
keyword. For batch workloads, display panels correspond to jobs or job steps.
A single DISPLAY command often generates more than one display panel. For example, if the RMF
interval is set to 15 minutes, the following command generates four panels:
DISPLAY PGN(2) TODAY STIME(9) ETIME(10)
By default, the reporter generates only as many display panels as will fit on the display screen of the
terminal. (If you are in split-screen mode, it generates only as many panels as will fit on the space being
used for the reporter.) When you press Enter, the reporter discards the currently displayed output and
generates the next set of panels. When you reach the last set of display panels, the reporter displays
LAST FRAME in the header.
This is done to reduce the processing time required by DISPLAY commands and to conserve storage. It is
also the reason you cannot ordinarily scroll back through reporter output. The only exception to this occurs
when a single display panel exceeds the length of the display screen. In this case, you can scroll up and
down within the display panel using the UP and DOWN PF keys (usually 7 and 19 for UP, and 8 and 20
for DOWN).
There are two kinds of display and scrolling options that give you flexibility in how you format and scroll
through the displays:
v The CONTROL MODE commands allow you to determine how panels are presented when you press
Enter.
v The COMPLETE keyword generates all display panels when the DISPLAY command is executed. This
keyword gives you the ability to scroll up and down through reporter output, but also has some
drawbacks with respect to processing time and storage utilization.
The CONTROL MODE commands and the COMPLETE keyword are described in the following sections.
Setting the display mode
When EPILOG is used in full-screen mode, the display header includes a Mode field, which is set to PAGE
by default. In PAGE mode, display panels are presented as described above; that is, the reporter
generates as many display panels as will fit on the screen. When you press Enter, the currently displayed
panels are discarded, and the next set of panels is displayed. In addition to PAGE, there are three other
settings for the display mode:
ONE
The screen shows only one display panel at a time, even though the next one may fit.
ROLL
Each time you press Enter, the topmost panel is discarded and the next panel beneath it
rolls to the top of the screen.
HOLD
The topmost panel is held on the screen and ROLL mode is invoked for the panels below.
Note that in HOLD mode you may only be able to view part of a bottom display panel
because the held panel on top is taking up most of the screen. When this is the case, you
will be able to scroll to the rest of the non-held panel.
You may change the display mode directly using the CONTROL command (abbreviated CNTL) with the
mode as an operand (CNTL PAGE, CNTL HOLD, and so on). Alternatively you could use PF2 (or PF14),
which by default are set to issue the CONTROL MODE command. CONTROL MODE advances the
display mode through all the possible settings; just keep pressing the PF key until it rolls around to the one
you want.
102
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The COMPLETE keyword
By adding the COMPLETE keyword to a DISPLAY command, you generate all display panels when the
command is executed. You can then scroll up and down through the panels using the UP and DOWN
ISPF commands, or their PF key equivalents.
You should keep in mind that the COMPLETE keyword adds considerably to the processing time and
storage requirements of the DISPLAY command. We recommend that you do not use COMPLETE with
DISPLAY commands that generate more than 20 or 30 panels, depending on the size of the panels.
You can use the LIMIT keyword with COMPLETE to reduce the processing time and storage
requirements.This keyword limits the display to a specified number of panels. For example, if you enter:
DISPLAY PGN(2) LASTWEEK RIF(RESP >2S) COMPLETE LIMIT(10)
only the first ten display panels that meet the exception criteria are generated by the command. (When the
reporter is invoked in batch mode, the default value for LIMIT is set to 10 panels. In online mode, however,
the only limit to the number of display panels that can be generated by a single DISPLAY command is the
amount of storage available.)
Displaying the current command
When a reporter command is executed, the command is normally not included in the output display. There
are two ways to display the command:
v Reshow the command on the command line by pressing PF4 or PF16. These keys are preset to issue
the CONTROL RECALL command.This command saves keystrokes because it allows you to modify
and reissue a command rather than type in the next command from scratch. After reshowing the
command, you must modify at least one character in the command to reissue it. Therefore, to reissue
the same command, overtype the first letter of the command and press Enter.
If you enter a command on multiple input lines (using the hyphen as a continuation character), only the
first line of the command is displayed.
v You can add a title block to the display by pressing PF5 or PF17.These keys are preset to issue the
CONTROL TITLE command. A sample title block is shown below.
------------------------------------------------------------------------------|
|
| Workloads:
Performance Group(2)
|
| Periods:
Starting On 9/18/99 and Ending on 9/18/99
|
| Report if :
Response Time is greater than 2 Seconds
|
| Misc Options: COMBINE(60M) AVERAGE PLOTMIN(10)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Performance Group =
2
Symbolic Name = DEV TSO
|
| From 17:00 To 18:00 On 09/18/99
SYSA |
| Average of 4 Intervals
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.08 S
3.6|-> .
.
.
.
.
.
.
.
.
.|
|Long Wait
0.88 S 38.1|------------===>.
.
.
.
.
.
.|
|ECB Wait (w/ STIMER)
0.64 S 27.5|----------->.
.
.
.
.
.
.
.|
|Average Trans Time
2.31 S
5548 MVS Transactions Ended
|
|Productivity Index
22%
|
+=============================================================================+
Figure 75. Sample display panel with a title block
The title block gives you a summary of the command. In this example, the input command was:
DISPLAY PGN(2) SDATE(9/19/99) RIF(RESP >2S) COMBINE(1H) PLOTMIN(10)
Chapter 6. Using the reporter
103
The AVERAGE keyword in the miscellaneous options indicates that average, not total, figures are being
displayed. If the COMBINE keyword had not been included in the command, the word SINGLE would have
been displayed as a miscellaneous option, indicating that single intervals were being displayed.
You can also obtain a shortened version of the title block by adding the SHORT keyword to the CONTROL
TITLE command.
Logging displays
The LOGON and LOGOFF keywords write out reporter displays to a log file. This file is defined by the
ddname EPLOG, and is automatically allocated to SYSOUT=A by the sample CLIST used to invoke the
reporter.
When a display is logged, all display panels generated by the command are written out, not just those
currently displayed on the screen. Also, a title block that describes the command is included with the
logged display.
You can log displays in three ways:
v You can log the output of a single command by entering LOGON as a keyword with the command.
v You can turn on default screen logging by entering the SET LOGON command. All reporter output
generated after this command is automatically logged. If you then want to display information that you
do not want logged, include the LOGOFF keyword in the command. This keyword suppresses logging
for that particular command, but leaves the default logging intact. (This feature is described in “Setting
display defaults” on page 95.)
v You can log a display that is already on the screen by pressing PF9 or PF21. These keys are preset to
issue the CNTL LOG command.
If your printer does not have lowercase characters, use the FOLDON keyword to log your displays in
uppercase characters only. The FOLDON and FOLDOFF keywords are described in “Controlling
uppercase output” on page 104.
Controlling uppercase output
The FOLDON and FOLDOFF keywords control output as it is written to a screen or log file. FOLDON
causes output to be written entirely in uppercase characters. FOLDOFF causes output to be written in
upper and lower case characters. The default is FOLDOFF.
Use FOLDON and FOLDOFF on reporter commands or on the CNTL command. For example, you can
specify DISPLAY FOLDON... or CNTL FOLDON....
Setting up a color terminal
In full-screen mode, EPILOG can use the extended color capabilities of 3279 displays which are so
equipped. Extended color support is turned on automatically; if your terminal is not configured to support it,
EPILOG will fall back on simple high and low intensity highlighting.
You control extended color support with the CONTROL COLOR command (abbreviated CNTL COLOR).
This command allows you to set colors and highlighting attributes for the following types of lines:
Headings and separator lines
Warnings and errors
Data entry lines
Odd display panels
Even display panels
Highlighted areas
Measure lines
Title blocks and graph lines
104
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Note: If you are running the reporter under ISPF, the graph lines default to the same color and
highlighting attributes as the rest of the display panel.
For each line type, you select a color and a highlighting attribute. The color selections are standard 3279
colors. The four available highlighting attributes are:
B
Causes the line type to blink.
U
Underscores the line type.
R
Reverses the line type. Normally, the display consists of colored characters against a dark
background. The REVERSE instruction displays dark characters against a colored background.
D
No highlighting (default).
You control extended color support and highlighting with the CONTROL COLOR command (abbreviated
CNTL COLOR) and its operands:
OFF
Turns off extended color mode.
None
Turns on color mode using the default color and highlighting definitions.
string
A 1–8 character string that sets the color areas for 3279 operation and turns color mode
on. Each character in the string is a color designation (or a period to indicate no change)
for the specific color areas.
Note: Display screen formatting errors may result if you turn extended color mode on (with CONTROL
COLOR) while using a non-color terminal. (PROG471 and IKT00405I SCREEN ERASURE
CAUSED BY ERROR RECOVERY PROCEDURE are two likely error messages.) If this happens,
use the PA1 key to reset EPILOG into non-color mode.
The default color assignments for the eight-character string are shown in Figure 76.
Figure 76. CONTROL COLOR command display
Figure 76 shows the display that appears when you enter the CONTROL COLOR command. To change
the color of a particular area, type the letter of the color you want the area to be on top of the letter which
points to that area, and press Enter.
Chapter 6. Using the reporter
105
You can also assign certain attributes to an area so that any text that appears in that area blinks, appears
in reverse video (dark letters on white background), or is underscored. To change the attributes of an area,
type the letter of the attribute you want for the area on top of the period which points to that area.
Control color
Control hilite
'G R W T Y g B W'
'. . r . . . . .'
│ │ │ │ │ │ │ │
│ │ │ │ │ │ │ Title blocks and graph lines
│ │ │ │ │ │ Measure line
│ │ │ │ │ Highlighted areas
│ │ │ │ Even display panels
│ │ │ Odd display panels
│ │ Data entry line
│ Warnings and errors
Headings and separator lines
In this example, highlighted areas are changed to green, and the data entry line is reversed.
Reporting in a multiple datastore environment
Multiple datastore reporting allows you to define multiple datastores across multiple systems. The reporter
can allocate and read multiple datastores during a session, and you can dynamically allocate and free
datastores without exiting a session. As a result, you can generate EPILOG reports based on data from
multiple processors or systems from a single reporter session.
You can limit the number of datastores that are included in display panels, as explained in “The SYSID
keyword” on page 107. You can generate a DASD Activity Report (RDAS) showing DASD device activity in
a shared-DASD environment, as explained in “Merging shared-DASD data” on page 108. And, if a
workload degradation display shows a performance bottleneck on a DASD device, you can navigate
directly to the Shared-DASD Degradation Analysis report, which shows cross-system usage of the device
in contention, as described in “Navigating in a multisystem environment” on page 131.
Environmental considerations
The multiple datastore reporting feature has been designed to take advantage of 31-bit addressability in
the MVS/XA and MVS/ESA environments. This advantage is not available in the MVS/370 environment. If
you use the multiple datastore reporting feature in the 370 environment, we recommend that you allocate
not more than three active datastores in a single reporter session.
The datastore list
The reporter uses two datastore lists: an initial datastore list, which is used for reporter initialization, and
an active datastore list, which is used during a reporter session. When a reporter session begins, EPILOG
for MVS builds the active datastore list by copying the initial datastore list. This process is described in
“Initializing the active datastore list” on page 111.
After initialization, you can add or delete datastores in the reporter list by using the DATASTOR command
and its keywords. When a command request is processed, each datastore currently in use is checked for
records that match the selection criteria.
Allocating datastores
You can control multiple-system reporting by allocating or deallocating EPILOG datastores during a
reporter session. When the reporter is invoked, it displays a datastore status message indicating the
datastores that are currently active, available, or unavailable. If you want to add or remove datastores from
the status list, use the DATASTOR command.
106
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The DATASTOR command
The DATASTOR command allows you to allocate and free datastores dynamically during a reporter
session. Since the DATASTOR command alters the active datastore list, and not the initial datastore list,
changes that you make with the DATASTOR command are effective for the life of the current reporter
session only.
The syntax for maintaining the active datastore list is:
DATASTOR {USE|ADD} EDS(dsname1, dsname2, ...)
or
DATASTOR {DROP} [EDS(n1, n2, ...)] [ALL]
where:
ADD
Allows you to add one or more datastores to the active datastore list. Datastores are
added to the end of the list.
DROP
Removes one or more datastores from the active datastore list. You can also DROP all
datastores from the active list by using the ALL operand.
USE
Defines a list of datastores that supersede the active datastore list. The USE operand is
similar to the ADD operand, except that the ADD operand adds datastores to the active
datastore list, while the USE operand replaces it by defining a new list of one or more
datastores.
dsname
Datastore name.
n
Sequence number of a datastore in the active datastore list.
ALL
Deletes all datastores from the active datastore list.
Usage notes
v Each time a DATASTOR command is processed, the INQUIRE SUMMARY display is automatically
invoked, showing the datastores that are currently active.
v The DATASTOR USE and ADD commands automatically verify that the newly allocated file is a valid
datastore.
v When a command request is issued during the session (such as DISPLAY), the active datastores are
searched and displayed in SYSID within time order, and not in the initial datastore list order.
v If multiple datastores are allocated, certain types of requests may be time-consuming to process (for
example, extensive COMBINE/MERGE operations). In these cases, a message informs the user of this
condition.
Displaying datastore status
The INQUIRE command helps you to determine the status of EPILOG datastores in a multi-datastore
configuration. See “Displaying datastore information” on page 111 for further information.
The SYSID keyword
The SYSID keyword allows you to control the way that multiple-system data is displayed. If your active
datastore list defines multiple data sets for multiple systems, and you issue a DISPLAY command, one
panel is generated for each system per recording interval. You can use the SYSID keyword to limit the
number of systems that will be included in the displays, thereby reducing the number of panels by interval.
Limiting the number of systems in a report also reduces the amount of search time required for a display,
assuming that your datastores are classified by SYSID. (If there are datastores in your installation
containing data that was written prior to Version 300, you can convert them to the Version 300 or later
format by using the KEBRSET utility to initialize the data sets with the SYSID field. For further information,
see the discussion of the KEBRSET utility in theEPILOG for MVS Customization Guide.)
Chapter 6. Using the reporter
107
For example, assume that you issue a DISPLAY RDAS command. Without the SYSID keyword, this
command displays one Device Activity panel for each system being monitored. If your active datastore list
defines five systems (SYSA, SYSB, SYSC, SYSD, SYSE), five panels are displayed for each interval. With
the SYSID keyword, you can restrict the display as follows:
Keyword
Description
SYSID (SYSA)
Generates one panel per reporting interval for SYSA; suppresses displays
for the other systems.
SYSID (SYSA, SYSB, SYSC)
Generates one panel each for systems A, B, and C; suppresses displays
for systems D and E.
The SYSID keyword can also be used in this way for the following commands: DISPLAY, COMPARE,
PROFILE, EXTRACT, COMPEXT, OBTAIN, SET, and SETP. If you use the SET SYSID command, it
remains in effect until it is overridden by a DISPLAY SYSID command or another SET SYSID command.
Merging shared-DASD data
In a multisystem configuration, performance problems can result from contention across multiple systems
for shared-DASD devices. To help you detect the cross-system impact of shared-DASD usage, you can
combine DASD activity information from multiple datastores into a single shared-DASD display.
You use the MERGE keyword to merge DASD historical data into a single display panel showing
shared-DASD activity across multiple systems. This feature is available when you use the DISPLAY
command with the RDAS keyword.
If you issue the DISPLAY RDAS command, the SYSID keyword allows the following syntax:
SYSID (MERGE)
Show one RDAS panel per reporting interval, with one detail line for each
system being monitored.
SYSID (MERGE SYSA SYSB SYSC)
Combine systems A, B, and C into one display panel; exclude the other
systems.
SYSID ((MERGE, SYSA, SYSB), (SYSC, SYSD, MERGE) SYSE)
Generate three RDAS panels; one combines SYSA and SYSB; the second
combines SYSC and SYSD; and the third shows SYSE.
The MERGE operand can be placed anywhere among a number of SYSIDs and will apply to all the
SYSIDs enclosed in the same pair of parentheses.
Examples
Assume that there is a shared-DASD environment between two systems, SYSA and SYSF. The SYSA
records are in PROD.EDS.SYSA.F1 and PROD.EDS.SYSA.F2, and the SYSF records are in
PROD.EDS.SYSF.F1 and PROD.EDS.SYSF.F2.
The reporter that you are using is normally used to report on SYSA. When you invoke the reporter, it
builds the active datastore list by copying the initial datastore list, which only contains datastores with the
SYSA qualifier. To investigate the possibility of cross-system impact on shared-DASD utilization, you follow
these steps.
1. Start the reporter session by allocating the datastores that contain SYSF records.
DATASTOR ADD EDS (PROD.EDS.SYSF.F1,PROD.EDS.SYSF.F2)
For the rest of this reporter session, if you request a display, all four datastores will be checked for
matching records.
2. Issue a DISPLAY command.
DISPLAY RDAS STIME(10) ETIME(11) YDAY
108
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The RDAS keyword generates a set of panels, one display panel for each system per reporting
interval. Figure 78 on page 164 shows the two panels that are generated for the first interval (10:00
AM to 10:15 AM).
+========================== DASD Device Activity =============================+
| From:
10:00 to 10:15 on 10/18/99
Elap = 14:59 M
SYSA |
+-----------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev
Avg|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Allcs|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- -----|
|CHKPT1 SYSA 150 00C 1.2
43
0
22
7 14
21.9 17.7
1.0|
|COM001 SYSA D89 055 6.5
10
1
2
5
2
.2 1.1
5.0 339.5|
|COM002 SYSA D8A 055 1.3
17
1
2
6
8
.2 1.7
2.7 483.1|
+=============================================================================+
+========================== DASD Device Activity =============================+
| From:
10:00 to 10:15 on 10/18/99
Elap = 14:59 M
SYSF |
+-----------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev
Avg|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Allcs|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- -----|
|CHKPT1 SYSF 150 00A 1.0
29
0
10
6 13
9.9 17.4
1.0|
|COM001 SYSF D89 056 3.2
10
1
2
4
3
.1 1.3
2.6 267.1|
|COM002 SYSF D8A 056
.6
15
0
3
6
6
.4 2.0
2.0 238.1|
+=============================================================================+
Figure 77. The RDAS display in a multi-system environment (without MERGE)
The panels from SYSA are mixed with the SYSF panels in chronological sequence. If two collectors
are active during the reporting time selection, and the RMF interval is 15 minutes, eight panels will be
generated. You can eliminate this redundancy by merging the RDAS displays.
3. To merge SYSA and SYSF records together, you enter the command
DISPLAY RDAS STIME(10) ETIME(1015) SYSID(MERGE) YDAY
A panel from the resulting display is shown in Figure 78.
+========================== DASD Device Activity =============================+
| From:
10:00 to 10:15 on 10/18/99
Elap = 14:59 M
|
| Sysid:
SYSA SYSF
|
| Merge of 2 Intervals
|
+-----------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev
Avg|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Allcs|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- -----|
|CHKPT1 SYSA 150 00C 1.2
43
0
22
7 14
21.9 17.7
1.0|
|
SYSF 150 00A 1.0
29
0
10
6 13
9.9 17.4
1.0|
|COM001 SYSA D89 055 6.5
10
1
2
5
2
.2 1.1
5.0 339.5|
|
SYSF D89 056 3.2
10
1
2
4
3
.1 1.3
2.6 267.1|
|COM002 SYSA D8A 055 1.3
17
1
2
6
8
.2 1.7
2.7 483.1|
|
SYSF D8A 056
.6
15
0
3
6
6
.4 2.0
2.0 238.1|
+=============================================================================+
Figure 78. The RDAS display in a multisystem environment (with MERGE)
This MERGE keyword has merged data from SYSA and SYSF into one panel per interval. This display
panel shows that volumes CHKPT1, COM001, and COM002 are shared between the two systems.
Merging across different MVS environments
The MERGE function does not merge across all MVS systems equally.
If you attempt to merge across MVS/XA and MVS/ESA environments, the merge is performed, but there is
a change in the RDAS screen. The column header for the last column in the RDAS display, which is Open
Dsns for MVS/XA systems and Avg Allocs for MVS/ESA systems, is changed to Open Allcs, indicating a
merge between the two kinds of information. The detail lines in the RDAS display show either open data
set names or average allocations in this column, depending on the type of environment.
Chapter 6. Using the reporter
109
If you attempt to merge across MVS/370 and other MVS systems, EPILOG displays a screen message
notifying you that the merge cannot be performed as requested. Separate panels are displayed for
MVS/370 information in each interval; data from other MVS systems are merged into one panel per
interval.
There is a way to merge MVS/370 data with data from other MVS systems. Assume, for example, that you
want to merge data from two MVS/370 systems (SYS1 and SYS2) and two MVS/XA systems (SYS3 and
SYS4). If you issue a command in the following format:
DIS RDAS SYSID((SYS1 SYS2 MERGE), (SYS3 SYS4 MERGE))
EPILOG generates two panels per interval: one for the MVS/370 systems and one for the MVS/XA
systems. Each panel may contain a large number of device statistics per interval, and the two types of
panel will not be displayed on the same screen. You can focus the display by specifying a single volume in
the command, in the following format:
DIS RDAS(V(vvvvvv)) SYSID((SYS1 SYS2 MERGE), (SYS3 SYS4 MERGE))
As a result, the two panels per interval will generally appear on one screen.
Using the MERGE and COMBINE keywords
You can combine records by adding a COMBINE keyword with an associated time span:
DIS RDAS STIME(10) ETIME(11) TODAY SYSID(MERGE) COMBINE(1H)
The result is shown in Figure 79.
+========================== DASD Device Activity =============================+
| From:
10:00 to 11:00 on 10/18/99
Elap = 1:00 H
|
| Sysid:
SYSA SYSF
|
| Combination and Merge of 8 Intervals
|
+-----------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev Avg|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Allcs|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- -----|
|CHKPT1 SYSA 150 00C
1.1
69
0
48
7 14
47.8 17.2 1.0|
|
SYSF 150 00A
1.0
32
0
10
7 15
9.5 22.0 1.0|
|COM001 SYSA D89 055
6.8
16
4
3
6
3
.2 2.0
5.7 134.2|
|
SYSF D89 056
4.5
10
1
2
4
3
.2 1.6
4.0 381.0|
|COM002 SYSA D8A 055
.3
14
0
5
6
3
1.9 2.3
.2 223.5|
|
SYSF D8A 056
1.3
19
1
3
6
9
.3 2.6
3.7 168.1|
+=============================================================================+
Figure 79. The RDAS display using MERGE and COMBINE
In this example, the COMBINE keyword generates a single cross-system display of DASD device activity
for each one hour interval.
When the COMBINE keyword is used without the SYSID(MERGE) keyword, records with different SYSIDs
are not combined.
Tolerance limits
There is a potential problem when data from two systems with unequal collection intervals is displayed
together. In multi-system reporting, the sampling intervals used to collect the data (usually RMF-based)
must be the same for both systems.
The tolerance limit for interval boundaries in 10%. The MERGE keyword will merge data from different
systems if the RMF intervals have a common boundary (within a 10% tolerance factor) based on the first
interval encountered.
For example, assume that you are merging records from SYSA and SYSB. If the first RMF interval
encountered is a SYSA record with a start time of 9:00, the tolerance limit is 10% of 15 minutes, or 1.5
110
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
minutes. EPILOG merges the SYSF record for the corresponding RMF interval if the start time is between
9:00 and 9:01:30. If you attempt to merge data with unequal interval records, EPILOG displays the data
from the system with the earliest start time as a separate panel, and uses the next record to set the
starting time for the next merged interval. When an interval cannot be merged, the following message is
displayed in the panel header, and the interval search proceeds to completion:
This display is incomplete because of incompatible
RMF intervals
You can overcome the interval time tolerance limit by using the COMBINE keyword. For example, if you
add the keyword COMBINE(1H) to the MERGE command, all intervals with a starting time within the one
hour reporting interval will be merged.
Initializing the active datastore list
The reporter builds the active datastore list by reading the initial datastore list, which is a list of data set
names in the EDSLIST member. The EDSLIST member is in a partitioned data set that is identified by the
EPPARM ddname. Comments are allowed in the member, either by the usual asterisk designator in
column 1 or on the same line as the data set name, following its specification. The data set name may
appear anywhere in a line, provided it is only preceded by blanks.
The reporter treats the list as a group of logically concatenated datastores. To deallocate or allocate
another datastore, use the DATASTOR command.
For compatibility, if a job, CLIST, or EXEC specifies an EPVSAM DD statement (without an EDSLIST
member in the EPPARM data set), the task will not abort; it will be treated as an EDSLIST list with a
single EDS. An informational message will be displayed informing you about the change.
If a job, CLIST, or EXEC specifies an EPVSAM DD statement and a EDSLIST member exists in the
EPPARM data set, the EDSLIST member will override the EPVSAM EDS designation and an error
message will be displayed informing you about the override and change.
See the IBM Tivoli OMEGAMON XE on z/OS: Planning and Configuration Guide for other installation and
maintenance issues related to multiple datastore reporting.
Product display functions
EPILOG for MVS provides several other display functions that help to use the product efficiently.
Displaying datastore information
The INQUIRE command allows you to make an online inquiry about the status of the datastore(s). This
section describes how to display information about the EPILOG datastore and Profile datastore. (If you
also have DELTAMON for MVS installed, INQUIRE displays information about the storage of DELTAMON
data. See the DELTAMON Reference Manual, DM53-3904, for details.)
INQUIRE can be entered with or without an operand. Without an operand, it shows you VSAM file
statistics about the EPILOG datastore and Profile datastore.
Chapter 6. Using the reporter
111
Datastore Statistics
For:
Type:
Utilization:
SYSID:
Collection Status:
Reason:
TDNN01.EDS.SYSA.DEC99
EDS
64%
SYSA
AVAILABLE
Allocated Control Intervals
Free Control Intervals
Number of Data Records
Number of Extents
Control Interval Size
Dataset LRECL
High Allocated RBA
Date And Time Range:
15000
4639
345874
1
8192
8185
- 122880K
12/01/99 09:00 - 12/31/99 23:59
By including operands with the INQUIRE command, you can display more specific information about either
the EPILOG datastore or the Profile datastore.
If you include a workload name with INQUIRE, the display shows how much space in the EPILOG
datastore is being taken up by records of that type. For example, if you enter
INQUIRE PGN(2)
the display shows you how much space performance group 2 records are using.
Workload:
SYSID:
Data Store:
Type:
PGN 2
SYSA
TDNN01.EDS.SYSA.DEC99
EDS
First Record Written On - 12/01/99 At
Last Record Written On - 12/31/99 At
Number of Records 3388
Length of 3388 Data Records = 1481675
09:01
23:58
If you enter a workload name and a profile name, the display shows how much space in the Profile
datastore is being used by profile records of that type. You can use an asterisk for the workload and profile
name to survey the entire contents of the Profile datastore for the workload type you specify. For example,
if you enter
INQ JOB(*) PROFNAME(*)
the resulting display looks like this:
112
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Job = MGTJOBS
Date first profile record written:
Date last profile record written:
Total number of profile records =
Total length of profile records =
Job = SORTNY
Date first profile record written:
Date last profile record written:
Total number of profile records =
Total length of profile records =
Profname = MGTJOBS
11/11/99
11/11/99
1
553
Profname = BATCHSORT
11/18/99
11/18/99
1
893
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .
Note: When making inquiries against the Profile datastore, do not specify multiple criteria for batch
workloads (for example, JOB(*) CLASS(A)).
The SUMMARY keyword generates a display panel showing a summary status report for all datastores
that have been defined in the current reporter session.For example, the command
INQUIRE SUMMARY
generates a summary display in the following format:
+==============================================================================+
|
|
|
Summary Data Store Statistics
|
|
|
|
Collection
|
|
Date and Time Range
Type Util SYSID Status
Reason
|
|
------------------------------- ---- ---- ---- --------- --------- |
|
1 TDNN01.EDS.SYSA.DEC98
|
|
12/01/98 09:00 - 12/31/98 23:59 EDS 64% SYSA AVAILABLE
|
|
|
|
2 TDNN01.EDS.SYSA.JAN99
|
|
01/01/99 00:01 - 01/31/99 23:00 EDS 73% SYSA SWITCHED MONTH
|
|
|
|
3 TDNN01.EDS.SYSA.FEB99
|
|
02/01/99 00:02 - 02/17/99 08:15 EDS 21% SYSA ACTIVE
|
|
|
|
4 TDNN01.EDS.SYSF.F1
|
|
01/01/99 00:01 - 01/25/99 07:00 EDS 100% SYSF SWITCHED FULL
|
|
|
|
5 TDNN01.EDS.SYSF.F2
|
|
01/25/99 07:16 - 02/17/99 08:15 EDS 62% SYSF ACTIVE
|
|
|
|
6 TDNN01.EDS.SYSF.F3
|
|
............................... .... .... .... UNAVAIL
ALLOC
|
|
|
|
TDNN01.DEV.PRDS
|
|
PRDS 47% SYSA
|
|
|
|
TDNN01.DEV.PRDS
|
|
PRDS 12% SYSF
|
|
|
|
TDNN01.DEV.DDS
|
|
12/01/98 06:30 - 02/17/99 08:15 DDS 88% SYSA
|
|
|
|
|
|
Press ENTER to Continue
|
|
|
+==============================================================================+
Chapter 6. Using the reporter
113
For each datastore, the panel shows the datastore type, the percent utilization, the date ranges that the
datastore spans, the SYSIDs in the datastore, the current collection status, and the associated reason for
the collector status. If the EDS collection status is invalid or unavailable, the Reason column contains an
abbreviated description of the reason; a complete reason is provided in the message log.
The sequence number is displayed for all EPILOG datastores to enable easy use of the DATASTOR
DROP(n) command where n represents the EDS to be dropped from the current reporter session. This
sequence number is not the same as the sequence number displayed on the collector STATUS display.
The INQUIRE SUMMARY display is automatically invoked at the beginning of a reporter session, and any
time a datastore is allocated or deallocated during a session.
In addition to the amount of space, these displays show you the date range of the records in the
datastore. This information is provided to help you run the datastore maintenance functions. These
functions are described in the EPILOG for MVS Advanced User’s Guide.
For further information concerning the syntax and usage of the INQUIRE command, see the EPILOG for
MVS Command Summary.
Displaying product information
The PRODUCTS reporter command shows information about the EPILOG for MVS and DELTAMON for
MVS monitors you have installed. The PRODUCTS command identifies the version of each product and
shows the expiration dates.
The PRODUCTS command
To check which members and versions of IBM background monitors are installed at your installation, issue
the following one word command:
PRODUCTS
The output of the command looks like this:
C U R R E N T
------------EPILOG/MVS
114
P R O D U C T S
---------------
VERSION 350 EXPIRES 9/15/99
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 7. Service level management
EPILOG for MVS reports help an installation manage the performance and capacity of the data center.
Three standard report jobs are included with the EPILOG product. These jobs can be modified to allow
standard daily or weekly reports on the capacity, performance, or service level of a data center. These jobs
are included in rhilev.REDDATA as members KEPSRDIR, KEPSROPM, and KEPSRSPG.
In addition, a job that generates standard profile/workload comparison reports is also included in member
KEPSRCMP. These reports compare both TSO performance groups and batch jobs to their respective
profiles. See Chapter 10, “Workload Profiling Facility (WPF),” on page 167for further information about the
workload profiling reports.
You should modify the standard reports that are provided to suit the specific requirements of your
installation.
Standard reports
The following sections discuss the sample report jobs provided on the distribution tape. You may want to
run these jobs on your system and review the output to decide which apply to your installation and which
should be modified or eliminated.
Service levels reports (KEPSRDIR)
The standard KEPSRDIR reporting job reports on service levels provided by the data center. This job is
targeted to the data center manager responsible for providing service levels. The job reports on throughput
and response time and should be run weekly. It is broken up into four major areas (total system, TSO,
IMS, and CICS). The reports provided include:
v DIR01-1 TSO trivial response time summary.
v
v
v
v
v
v
v
DIR01-2
DIR01-3
DIR01-4
DIR02-1
DIR02-2
DIR02-3
DIR02-4
Service level of TSO trivial functions.
Service level of overall TSO.
Service level of TSO trivial functions during prime afternoon hours 1:00 PM to 3:00 PM
Overall batch service level analysis.
Class A batch service level analysis (production).
Class T batch service level analysis (test).
Specifically selected batch analysis, for example, PAYROLL.
v DIR03-1 Performance group 1 service levels analysis.
v DIR03-2 Performance group 3 service levels analysis.
v DIR04-1 Overall system degradation analysis.
Jobs and groups reports (KEPSROPM)
The standard KEPSROPM reporting job reports on specific TSO performance groups, production jobs,
tape or disk mounts. This job is designed to be run daily and provide input to the operations manager on
the system. The reports provided include:
v OPM01-1 Period 1 TSO response time summary analysis.
v OPM01-2 Instances of poor first period TSO response time from 13:00 to 15:00.
v OPM01-3 Periods of poor overall TSO response time (greater than 2 seconds).
v OPM01-4 Periods of poor overall TSO response time (greater than 10 seconds).
v OPM02-1 Non-setup jobs requiring disk mounts.
v OPM02-2 Non-setup jobs requiring tape mounts.
v OPM02-3 Quick turnaround jobs utilizing large amounts of CPU.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
115
v
v
v
v
v
OPM03-1
OPM03-2
OPM03-3
OPM04-1
OPM04-2
Selected jobs running at abnormal times.
Selected jobs exceed their usual run times.
Specified jobs execute during prime shift.
Jobs waiting for disk mounts over 10 minutes.
Jobs waiting for tape mounts over 10 minutes.
v
v
v
v
v
v
OPM05-1
OPM05-2
OPM05-3
OPM05-4
OPM05-5
OPM05-6
TSO waiting for auxiliary storage.
Batch jobs waiting for auxiliary storage.
TSO degraded waiting for data sets.
Batch degraded waiting for data sets.
TSO degraded by pending vary online or offlines.
Batch degraded by pending vary online or offlines.
Performance and tuning reports (KEPSRSPG)
The standard KEPSRSPG reporting job reports data on performance, capacity management, and tuning.
This job is designed to be run daily and provide input to the system programming departments responsible
for performance management. The reports provided include:
v SPG01-1 TSO trivial response time summary analysis.
v SPG01-2 TSO response for trivial functions during prime afternoon hours of 1:00 PM to 3:00 PM
analysis.
v SPG01-3 TSO response exceeding 3 seconds during prime shift analysis.
v
v
v
v
v
SPG02-1
SPG02-2
SPG03-1
SPG03-2
SPG03-3
TSO waiting on auxiliary storage.
Batch waiting on auxiliary storage.
Demand page-in degrading TSO response time.
Common page-in degrading TSO response time.
Demand page-in degrading batch turnaround.
v
v
v
v
v
v
v
v
SPG03-4
SPG04-1
SPG04-2
SPG05-1
SPG05-2
SPG05-3
SPG05-4
SPG05-5
Common page-in degrading batch turnaround.
Swap-in causing TSO response time problems.
Swap-in causing batch turnaround problems.
Enqueue-exchange swaps degrading TSO.
Exchange swaps degrading TSO.
Unilateral swaps degrading TSO.
Enqueue-exchange swaps degrading batch.
Exchange swaps degrading batch.
v SPG05-6 Unilateral swaps degrading batch.
v SPG06-1 Excessive enqueues degrading TSO.
v
v
v
v
v
SPG06-2
SPG07-1
SPG07-2
SPG08-1
SPG08-2
Excessive enqueues degrading batch.
Logical channel 1 impacting TSO.
Logical channel 1 impacting batch.
Excessive DASD reserves degrading TSO.
Excessive DASD reserves degrading batch.
v SPG09-1 Control unit 14X impacting TSO.
v SPG09-2 Control unit 14X impacting batch.
v
v
v
v
SPG10-1 I/O queueing impacting TSO workload.
SPG10-2 I/O queueing impacting batch workload.
SPG11-1 MSS staging impacting TSO.
SPG11-2 MSS staging impacting batch.
116
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Workload profile comparisons (KEPSRCMP)
The standard KEPSRCMP reporting job compares workloads with profiles for various TSO performance
groups and batch workloads. This job should be modified to include your installation’s specific performance
groups, jobnames, and assigned profile names. It is designed to be run daily and provide information for
the systems programming departments responsible for performance management. In order to create the
profiles required for the comparisons, however, you must first modify and submit the job supplied in
rhilev.REDDATA(KEPPROFJ), or in member KEDPROFJ for sites where both EPILOG and DELTAMON
are installed. (For further information, see the instructions for creating profiles in Chapter 10, “Workload
Profiling Facility (WPF),” on page 167.)
The standard KEPSRCMP reporting job generates the following reports:
v
v
v
v
CPM01-1
CPM01-2
CPM02-1
CPM02-2
Compares
Compares
Compares
Compares
period one TSO response time with last profile.
overall TSO response time with last profile.
all jobs with their profiles and summarizes.
all jobs with their profiles.
TSO response time
Before you begin to analyze TSO response time problems it is essential that your installation establish
service goals; that is, you must first define what constitutes good response time. This is usually done by
deciding on an acceptable average response time for first period or trivial TSO transactions.
After service level goals have been defined you can use the exception reporting capabilities of EPILOG for
MVS to find intervals where goals were not being met. A degradation detail display will also show the main
components of response time, letting you determine the reasons response time was unacceptable. For
example, the command
DISPLAY PGN(2) PGP(1) YESTERDAY STIME(11) ETIME(15) SUMMARY
shows the main wait reason for each RMF interval yesterday between 11:00 and 3:00. Figure 80 on page
118displays the results of this command.
Chapter 7. Service level management
117
+-----------------------------------------------------------------------------+
|
EPILOG for MVS SAMPLE REPORTS
|
|
SERVICE LEVEL ANALYSIS
|
|
|
|
** S U M M A R Y R E P O R T **
|
|
|
|
REPORT ON TSO RESPONSE TIME YESTERDAY BY RMF INTERVAL
|
|-----------------------------------------------------------------------------|
| LEGEND:
|
|
MAIN_REASON: THE AREA WHERE TSO IS SPENDING MOST OF ITS TIME
|
|
TIME: THE AVERAGE TSO RESPONSE TIME
|
|
*: PLOTS THE IMPACT OF THE MAIN_REASON ON RESPONSE
|
|
-->: PLOTS THE AVERAGE TSO RESPONSE TIME
|
|-----------------------------------------------------------------------------|
| WKLD PERIODS: STARTING ON 4/05/99 AT 11:00 AND ENDING ON 4/05/99 AT 15:00 |
+=============================================================================+
| PERIOD: 04/05/99 11:00 TO 04/05/99 15:00
|
+=============================================================================+
|DATE__START_END____MAIN_REASON___TIME___|0S_____2______4______6______8_____10|
|04/05 11:00 11:15 USING CPU
2.02 S|***---->
.
.
.
.|
|
11:15 11:30 DA 250 TSO001 1.17 S|**---> .
.
.
.
.|
|
11:30 11:45 DA 250 TSO001 1.19 S|**---> .
.
.
.
.|
|
11:45 12:00 CPU WAIT
1.70 S|***---->
.
.
.
.|
|
12:00 12:15 DA 250 TSO001 1.39 S|**---> .
.
.
.
.|
|
12:15 12:30 DA 250 TSO001 1.72 S|**----->
.
.
.
.|
|
12:30 12:45 DA 250 TSO001 0.97 S|*-->
.
.
.
.
.|
|
12:45 13:00 CPU WAIT
2.35 S|***----->
.
.
.
.|
|
13:00 13:15 CPU WAIT
1.36 S|**---> .
.
.
.
.|
|
13:15 13:30 DA 250 TSO001 1.84 S|****--->
.
.
.
.|
|
13:30 13:45 PRVATE PAGEIN 7.06 S|**************-----------> .
.|
|
13:45 14:00 PRVATE PAGEIN 3.40 S|*******-----> .
.
.
.|
|
14:00 14:15 LONG WAIT SWP 2.26 S|****---->
.
.
.
.|
|
14:15 14:30 PRVATE PAGEIN 5.38 S|*******------------> .
.
.|
|
14:30 14:45 PRVATE PAGEIN 1.73 S|***---->
.
.
.
.|
|
14:45 15:00 PRVATE PAGEIN 8.53 S|*********--------------------->
.|
+=============================================================================+
Summary Report Showing Profile of TSO Response Time
Figure 80. Summary report showing profile of TSO time
In this example you can see that TSO response time (all periods) hovered around two seconds, with a few
exceptions, especially in the 13:30 to 15:00 hours time period. Further investigation showed that DASD
backup/restore jobs were being run during this period, and were impacting the performance of the paging
subsystem, causing poor response.
With the TSO full-screen EPILOG reporter you can compare one time period with another. You can
compare the current week’s average during prime shift to last week’s average by issuing the following
sequence of commands:
1. DISPLAY PGN(2) PGP(1) DAY(WEEKDAY) LASTWEEK STIME(9) ETIME(9) COMBINE
2. Press PF14 for Mode: HOLD to appear on the screen, or enter CNTL HOLD.
3. DISPLAY PGN(2) PGP(1) DAY(WEEKDAY) THISWEEK STIME(9) ETIME(17) COMBINE
118
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
------------------------------------------------------------------------------|
|
| Workloads:
Period 1 of Performance Group(2)
|
| Wkld Periods: Starting on 8/15/99 and Ending on 8/21/99
|
| Misc Options: COMBINE AVERAGE PLOTMIN(5)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Performance Group = 2
Period 1
|
| Period: 09:00 on 08/15/99 to 17:00 on 08/21/99
A083 |
| Average of 400 Intervals
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.01 S 6.1|--> .
.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.03 S 13.4|-----> .
.
.
.
.
.
.
.
.|
|Private Page-In Wait
0.03 S 13.4|-----> .
.
.
.
.
.
.
.
.|
|Terminal Output Wait
0.03 S 12.2|---->
.
.
.
.
.
.
.
.
.|
|Waiting for CPU
0.02 S 8.5|--->.
.
.
.
.
.
.
.
.
.|
|Average Trans Time
0.29 S
173659 MVS Transactions Ended
|
+=============================================================================+
------------------------------------------------------------------------------|
|
| Workloads:
Period 1 of Performance Group(2)
|
| Wkld Periods: Starting on 8/22/99 and Ending on 8/25/99
|
| Misc Options: COMBINE AVERAGE PLOTMIN(5)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Performance Group = 2
Period 1
|
| Period: 09:00 on 08/22/99 to 17:00 on 08/25/99
A083 |
| Average of 400 Intervals
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.01 S 6.3|--> .
.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.04 S 13.7|-----> .
.
.
.
.
.
.
.
.|
|Private Page-In Wait
0.04 S 13.7|-----> .
.
.
.
.
.
.
.
.|
|Long Wait
0.04 S 13.7|-----> .
.
.
.
.
.
.
.
.|
|Waiting for CPU
0.02 S 8.4|--->.
.
.
.
.
.
.
.
.
.|
|Terminal Output Wait
0.01 S 5.3|--> .
.
.
.
.
.
.
.
.
.|
|Disk SYSR8J 36A Act
0.01 S 5.3|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
0.31 S
120918 MVS Transactions Ended
|
+=============================================================================+
Figure 81. TSO response time comparison
Now you can compare last week’s average to this week’s. Specific time periods are indicated by the
STIME and ETIME keywords. You could also have chosen a more specific range of days with the SDATE
and EDATE keywords.
You can also use EPILOG to look for intervals during which a particular type of bottleneck became
significant. For example, the command
DISPLAY PGN(2) PGP(1) RIF(QIO(ANY >0.5S))
asks EPILOG to display any intervals in which queuing against any one DASD volume contributed more
than .5 seconds to TSO first period response time. Figure 82 on page 120 shows an example using two
time intervals.
Chapter 7. Service level management
119
------------------------------------------------------------------------------|
|
| Workloads:
Performance Group(2)
|
| Select if : Any Queued I/O is greater than .500 Seconds
|
| Misc Options: SINGLE PLOTMIN(0)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Performance Group = 2
Symbolic Name = T S O
|
| Period: 07:00 to 07:15 on 07/25/99
Elap = 14:51 M A083 |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
.0|.
.
.
.
.
.
.
.
.
.
.|
|Disk SYS021 140 Act
2.53 S 50.0|------------========>
.
.
.
.
.|
|Waiting for CPU
1.68 S 33.3|------------=> .
.
.
.
.
.
.|
|Disk SYS021 140 Que
0.84 S 16.7|------> .
.
.
.
.
.
.
.
.|
|Average Trans Time
5.07 S
5 MVS Transactions Ended
|
+=============================================================================+
+=============================================================================+
| Performance Group = 2
Symbolic Name = T S O
|
| Period: 09:00 to 09:15 on 07/25/99
Elap = 14:50 M A083 |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.15 S 4.8|-> .
.
.
.
.
.
.
.
.
.|
|Disk OMON21 143 Act
1.20 S 38.5|------------===>.
.
.
.
.
.
.|
|Disk OMON21 143 Que
0.84 S 26.9|----------> .
.
.
.
.
.
.
.|
|Waiting for CPU
0.15 S 5.1|--> .
.
.
.
.
.
.
.
.
.|
|ECB Wait
0.14 S 4.5|-> .
.
.
.
.
.
.
.
.
.|
|Private Page-In Wait
0.11 S 3.6|-> .
.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.10 S 3.3|-> .
.
.
.
.
.
.
.
.
.|
|Disk TSO022 151 Act
0.07 S 2.4|.
.
.
.
.
.
.
.
.
.
.|
|Disk SYSR8J 36B Act
0.07 S 2.4|.
.
.
.
.
.
.
.
.
.
.|
|Disk SYS021 140 Act
0.06 S 2.1|.
.
.
.
.
.
.
.
.
.
.|
|Disk TSO021 141 Act
0.06 S 2.1|.
.
.
.
.
.
.
.
.
.
.|
|Waiting for MVS Lock
0.04 S 1.5|.
.
.
.
.
.
.
.
.
.
.|
|Average Trans Time
3.12 S
318 MVS Transactions Ended
|
+=============================================================================+
Figure 82. Effect of queued I/O on TSO response time
Batch jobs
EPILOG can also be used to investigate your batch workload; it can break jobs out by jobname, JES job
class, installation account code, and program name. In this example we issue the command
DISPLAY CLS(A) COMBINE
to combine all the class A jobs. Figure 83 on page 121shows an example of an average job’s degradation
profile.
120
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
------------------------------------------------------------------------------|
|
| Workloads:
Job Class(A)
|
| Misc Options: COMBINE AVERAGE PLOTMIN(5)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Job Class = A
|
| Period: 08:59 on 08/22/99 to 16:25 on 08/25/99
|
| Average of 467 Job Steps
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
9.85 S 7.6|--->.
.
.
.
.
.
.
.
.
.|
|Unilateral Swap
23.33 S 18.0|------->.
.
.
.
.
.
.
.
.|
|Waiting for CPU
21.29 S 16.5|------->.
.
.
.
.
.
.
.
.|
|Disk WORK21 354 Act 17.50 S 13.5|-----> .
.
.
.
.
.
.
.
.|
|Disk WORK21 354 Que 14.65 S 11.3|---->
.
.
.
.
.
.
.
.
.|
|Average Job Step Time
2:09 M
|
+=============================================================================+
Figure 83. Average job step analysis for class A
In this example, we find unilateral swaps to be the main wait reason. An excessive amount of time spent in
the unilateral or exchange swap state often indicates a problem with the IPS definition of the batch
performance group and/or domains (such as the maximum MPL set too low). On the other hand, there
could be too many active batch initiators accepting class A jobs. In some shops where resources are
scarce and TSO response time is very important, it may be normal for batch jobs to spend part of their
time swapped out to make room for higher priority work.
You can also analyze batch jobs that run longer than the required time specified for a job class. For
example, your installation job class A requirements might specify that no job should run longer than 10
minutes. The following command results in a report on all class A jobs which ran longer than the limit:
DISPLAY CLS(A) REPORTIF( ELT(>10M) )
Figure 84 on page 122 shows an example of the output of this command.
Chapter 7. Service level management
121
------------------------------------------------------------------------------|
|
| Workloads:
Job Class(A)
|
| Select if : Elapsed Time is greater than 10 Minutes
|
| Misc Options: SINGLE PLOTMIN(5)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Job = TDOI05A
JES Number = 3664
Program = ENQOBJ
|
| Job Class = A
Account Number = CANDLE
Step Number = 2
|
| Period: 14:29 to 14:40 on 08/22/99
Elap = 11:42 M A083 |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
19.66 S 2.8|-> .
.
.
.
.
.
.
.
.
.|
|Waiting for CPU
6:34 M 56.2|------------==========> .
.
.
.
.|
|Disk OMON21 143 Que
1:24 M 12.0|---->
.
.
.
.
.
.
.
.
.|
|Private Page-In Wait
47.04 S 6.7|--> .
.
.
.
.
.
.
.
.
.|
|Disk WORK21 354 Que 40.72 S 5.8|--> .
.
.
.
.
.
.
.
.
.|
|Job Step Elapsed Time 11:42 M
|
+=============================================================================+
+=============================================================================+
| Job = IMSSMP2
JES Number = 3674
Program = HMASMP
|
| Job Class = A
Account Number = CANDLE
Step Number = 1
|
| Period: 14:41 to 17:51 on 08/22/99
Elap =
3:09 H A083 |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
8:44 M 4.6|-> .
.
.
.
.
.
.
.
.
.|
|Disk IMS012 369 Act 51:05 M 26.9|----------> .
.
.
.
.
.
.
.|
|Disk WORK21 354 Act 41:01 M 21.6|-------->
.
.
.
.
.
.
.
.|
|Disk WORK21 354 Que 37:13 M 19.6|------->.
.
.
.
.
.
.
.
.|
|Waiting for CPU
28:06 M 14.8|-----> .
.
.
.
.
.
.
.
.|
|Job Step Elapsed Time
3:09 H
|
+=============================================================================+
Figure 84. Analysis of jobs running in the wrong class
Tape mounts can sometimes create a large component of degradation for jobs that would otherwise run
very quickly. In the following example, when you issue the command
DISPLAY CLS(T) RIF( TMP(>5M) )
EPILOG searches for all short tape jobs (class T) that spent more than five minutes waiting for a tape
mount to be satisfied, as shown in Figure 85 on page 123.
122
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
------------------------------------------------------------------------------|
|
| Workloads:
Job Class(T)
|
| Select if :
Tape Mount Pending is greater than 5 Minutes
|
| Misc Options: SINGLE PLOTMIN(0)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Job = CWDP06E
JES Number = 3776
Program = IEBGENER |
| Job Class = T
Account Number = CANDLE
Step Number = 1
|
| Period: 17:27 to 17:36 on 08/22/99
Elap =
8:32 M A083 |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.51 S
.1|.
.
.
.
.
.
.
.
.
.
.|
|Tape Mount Pending
8:25 M 98.7|------------============>>>>>>>>>>>>>>>>.|
|Waiting for MVS Lock
3.07 S
.6|.
.
.
.
.
.
.
.
.
.
.|
|Disk MVSA21 150 RSV
3.07 S
.6|.
.
.
.
.
.
.
.
.
.
.|
|Job Step Elapsed Time
8:32 M
|
+=============================================================================+
+=============================================================================+
| Job = TSSA01A
JES Number = 5574
Program = IEBGENER |
| Job Class = T
Account Number = CANDLE
Step Number = 1
|
| Period: 14:27 to 14:37 on 08/25/99
Elap =
9:58 M A083 |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
.0|.
.
.
.
.
.
.
.
.
.
.|
|Tape Mount Pending
9:49 M 98.4|------------============>>>>>>>>>>>>>>>>.|
|Waiting for CPU
2.99 S
.5|.
.
.
.
.
.
.
.
.
.
.|
|Disk SYS021 140 Act
2.99 S
.5|.
.
.
.
.
.
.
.
.
.
.|
|Disk TSO021 141 Que
2.99 S
.5|.
.
.
.
.
.
.
.
.
.
.|
|Job Step Elapsed Time
9:58 M
|
+=============================================================================+
Figure 85. Analysis of tape mounts
You can analyze the effect reserves are having on batch throughput in your installation. The command
DISPLAY CLS(B) RIF( RIO(ANY >5M) )
reports on any job steps that waited over five minutes for any reserve activity if the jobs ran in class B.
Figure 86 shows the output of this command.
------------------------------------------------------------------------------|
|
| Workloads:
Job Class(B)
|
| Select if :
Any Reserved I/O is greater than 5 Minutes
|
| Misc Options: SINGLE PLOTMIN(5)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Job = XXATP005
JES Number = 1927
Program = SORT
|
| Job Class = B
Account Number = 4460
Step Number = 8
|
| Period: 11:31 to 12:22 on 09/06/99
Elap = 51:21 M 33.8 |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
2:30 M 4.9|-> .
.
.
.
.
.
.
.
.
.|
|Disk WORK01 922 Act 12:56 M 25.2|----------> .
.
.
.
.
.
.
.|
|Disk WORK01 922 Rsv 10:34 M 20.6|-------->
.
.
.
.
.
.
.
.|
|Disk MVWK55 35B Act
8:13 M 16.0|------> .
.
.
.
.
.
.
.
.|
|Disk MVWK56 357 Act
3:44 M 7.3|--> .
.
.
.
.
.
.
.
.
.|
|Disk MVWK55 35B Que
3:23 M 6.6|--> .
.
.
.
.
.
.
.
.
.|
|Disk MVWK51 356 Act
3:20 M 6.5|--> .
.
.
.
.
.
.
.
.
.|
|Waiting for CPU
2:49 M 5.5|--> .
.
.
.
.
.
.
.
.
.|
|Job Step Elapsed Time 51:21 M
|
+=============================================================================+
Figure 86. Reserve analysis
To report on any job steps that ran for over an hour last week, issue the following command:
Chapter 7. Service level management
123
DISPLAY JOB(*) LASTWEEK REPORTIF(ELT(>1H))
Figure 87 shows the output of this command.
------------------------------------------------------------------------------|
|
| Workloads:
Job Name(********)
|
| Periods:
Starting on 8/15/99 and Ending on 8/21/99
|
| Select if : Elapsed Time is greater than 1 Hour
|
| Misc Options: SINGLE PLOTMIN(5)
|
|
|
------------------------------------------------------------------------------+=============================================================================+
| Job = FORMTAPE
JES Number = 3152
Job Steps =
1 / 1 |
| Job Class = O
Account Number = CANDLE
Input Queue = 1.45 S |
| Period: 07:02 to 09:23 on 08/20/99
Elap =
2:20 H
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
58.93 S
.7|.
.
.
.
.
.
.
.
.
.
.|
|Swapped with WTOR
1:42 H 73.2|------------============>>>>>> . . .|
|Tape Mount Pending
33:32 M 23.9|---------> .
.
.
.
.
.
.
.|
|Job Elapsed Time
2:20 H
|
+=============================================================================+
+=============================================================================+
| Job = IMSSMP2
JES Number = 1826
Job Steps =
1 / 1 |
| Job Class = A
Account Number = CANDLE
Input Queue = 3:31 M |
| Period: 15:39 to 17:18 on 08/17/99
Elap =
1:39 H
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
24:11 M 24.4|---------> .
.
.
.
.
.
.
.|
|Disk IMS012 369 Act 34:53 M 35.2|------------==> .
.
.
.
.
.
.|
|Waiting for CPU
22:35 M 22.8|---------> .
.
.
.
.
.
.
.|
|Disk WORK21 354 Act 11:06 M 11.2|---->
.
.
.
.
.
.
.
.
.|
|Waiting for MVS Lock
5:03 M 5.1|--> .
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
1:39 H
|
+=============================================================================+
Figure 87. Analysis of long running jobs
124
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 8. Fast-path navigation
A facility of EPILOG in the full-screen mode called fast-path navigation lets you easily and quickly discover
periods of bad system response time and investigate the associated bottlenecks and their causes. Before
describing this facility it will help to have a brief review of some of the basic types of EPILOG displays:
XPG
The cross-performance group display shows how various wait reasons are distributed
across all performance groups in the system.
SUMMARY
The SUMMARY display produces one line of output for each RMF interval, job execution,
or multi-interval. Each line contains the most significant bottleneck along with a plot of the
average transaction response time.
SUMWAIT
The SUMWAIT display groups detailed wait reasons into summary wait categories, such
as I/O waits or JES waits.
DETAIL
The DETAIL degradation display provides an in-depth breakdown of bottlenecks for a
given RMF interval, job execution, or multi-interval.
Resources
These are an array of 12 different pre-defined displays containing resource information
extracted from RMF and EPILOG data sampling, such as CPU activity, paging rates,
DASD activity, etc.
Each of these displays (except XPG) can use the collected data at two different levels of summarization:
SINGLE
At the SINGLE level, the data is shown in the smallest possible unit, either by RMF
interval or by a single execution of a batch job, started task, or TSO session. This is the
default.
COMBINE
COMBINEd displays are those which take more than one RMF interval or job execution
and average it in some meaningful fashion, resulting in displays which consist of more
than one interval or job’s worth of data.
Note: Cross-performance data (XPG) can only be displayed for single intervals.
This gives us nine basic types of panels:
v Combined Interval Summary
DISPLAY PGN(2) TODAY SUMMARY COMBINE(1H)
v Combined Interval Sumwait
DISPLAY PGN(2) TODAY SUMWAIT COMBINE(1H)
v Combined Interval Detail Display
DISPLAY PGN(2) TODAY COMBINE STIME(1100) ETIME(1200)
v Combined Interval Resource Display
DISPLAY RPAG TODAY COMBINE STIME(1100) ETIME(1200)
v Single Interval Summary
DISPLAY PGN(2) TODAY SUMMARY STIME(1100) ETIME(1200)
v Single Interval Sumwait
DISPLAY PGN(2) TODAY SUMWAIT STIME(1115) ETIME(1130)
v Single Interval Detail Display
DISPLAY PGN(2) TODAY STIME(1115) ETIME(1130)
v Single Interval Resource Display
DISPLAY RPAG TODAY STIME(1115) ETIME(1130)
v Single Interval XPG Display
DISPLAY XPG TODAY STIME(1115) ETIME(1130)
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
125
Using fast-path navigation
Fast-path navigation allows you to easily explore the possible displays in full-screen mode in order to
implement thelogical tuning methodology. Let’s begin with an example to illustrate how to use fast-path
navigation.
First let us look at a combined interval summary sumwait display for a given day from 9:00 AM to 5:00
PM:
DIS PGN(2) YDAY CMB(1H) SUMM SUMW
+=============================================================================+
| Performance Group = 2
Symbolic Name = DEV TSO
|
| From 09:00 To 17:00 On 09/06/99
SYSA |
+-----------------------------------------------------------------------------+
|DATE__START__END___MAIN_REASON(*)_TIME(-)|0S_____1______2______3______4______5
|09/06 09:00 10:00 Page + Swap
0.99 S|****--->
.
.
.
|
|
10:00 11:00 Page + Swap
0.68 S|***--> .
.
.
.
|
S
11:00 12:00 Page + Swap
2.16 S|******--------->
.
.
|
|
12:00 13:00 Page + Swap
0.90 S|***--->.
.
.
.
|
|
13:00 14:00 I/O
0.84 S|**---->.
.
.
.
|
|
14:00 15:00 Page + Swap
0.98 S|**---->.
.
.
.
|
|
15:00 16:00 Page + Swap
0.71 S|**---> .
.
.
.
|
|
16:00 17:00 Page + Swap
0.96 S|****--->
.
.
.
|
|----------------------------------------|
|Average for this Workload Display 1.07 S
|
+=============================================================================+
Figure 88. COMBINEd interval SUMMARY display
The format of the SUMMARY display makes it obvious that response time got appreciably worse between
11:00 and 12:00. The next step might be to ask for the SINGLE interval breakdown of the SUMMARY
display, by entering
DIS PGN(2) SUMM SUMW YDAY STIME(11) ETIME(12)
Fast-path navigation allows you to do this more simply by typing an S (for “select single interval display”) in
the first character of the line. (Note the character S on the third line of the display body in Figure 88.) This
has the effect of internally generating a command to produce a SINGLE interval display, as shown in
Figure 89.
+=============================================================================+
| Performance Group = 2
Symbolic Name = DEV TSO
|
| From 11:00 to 12:00 on 09/06/99
SYSA |
+-----------------------------------------------------------------------------+
|DATE__START__END___MAIN_REASON(*)_TIME(-)|0S_____1______2______3______4______5
|09/06 11:00 11:15 Page + Swap
0.84 S|***--->.
.
.
.
|
D
11:15 11:30 Page + Swap
4.98 S|************---------------------->|
|
11:30 11:45 Page + Swap
1.70 S|***--------> .
.
.
|
|
11:45 12:00 Page + Swap
1.12 S|*****--->
.
.
.
|
|----------------------------------------|
|Average for this Workload Display 2.16 S
|
+=============================================================================+
Figure 89. SINGLE interval SUMMARY display
A brief examination shows us that the major problem was in the 11:15 to 11:30 time interval. At this point
we might want to see a full detailed display of bottlenecks during this interval. This could be done by
entering
DIS PGN(2) SUMW YDAY STIME(1115) ETIME(1130)
Fast-path navigation simplifies it by allowing us to type a D (for “detailed degradation” display) in the first
column of the appropriate line. (Note the D on the second line of the display body in Figure 89.) Again, a
126
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
command is generated internally to produce the desired detailed display.
+=============================================================================+
| Performance Group = 2
Symbolic Name = TSO
|
| From 11:15 To 11:30 On 09/06/99
Elap = 15:00 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.23 S 4.7|-> .
.
.
.
.
.
.
.
.
.|
WPage + Swap
2.13 S 42.8|------------=====> .
.
.
.
.
.|
|I/O
1.53 S 30.8|------------>
.
.
.
.
.
.
.|
|SRM Delay (MPL)
0.29 S 5.8|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
4.98 S
865 MVS Transactions Ended
|
|Productivity Index
10%
|
+=============================================================================+
Figure 90. SINGLE interval DETAIL display
Figure 89 on page 126 clearly shows that degradation is primarily due to two factors: paging and swapping
delays and I/O delays. Our next step is to have a closer look at the paging and swapping delays. To do
so, we could enter the following command to display the detailed wait reasons that are responsible for the
degradation:
DIS PGN(2) STIME(1115) ETIME(1130)
We can achieve nearly the same effect by entering W on the line of the most important wait reason (in this
case, Page + Swap). (Note the W on the second line of the display body in Figure 90.)
The resulting display breaks down the summary wait reason into detailed wait reasons, as shown in Figure
92 on page 191.
+=============================================================================+
| Performance Group = 2
Symbolic Name = TSO
|
| From 11:15 To 11:30 On 09/06/99
Elap = 15:00 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.23 S 4.7|-> .
.
.
.
.
.
.
.
.
.|
RPrivate Page-In Wait
1.75 S 35.1|--------------> .
.
.
.
.
.
.|
|Swap Page-In Wait
0.38 S 7.7|--->.
.
.
.
.
.
.
.
.
.|
|
|
|Average Trans Time
4.98 S
865 MVS Transactions Ended
|
|Productivity Index
10%
|
+=============================================================================+
Figure 91. SINGLE interval DETAIL display
Notice that this display does not show a complete detail display; it only shows the detailed wait reasons
associated with the summary wait category we selected.
At this point we will want to see some resource data for the Private Page-In Wait bottleneck, which was
a major contributor to the poor response time. For example, to see the RPAG and RPDS resource panels
we could issue the request directly as:
DIS RPAG RPDS TODAY STIME(1115) ETIME(1130)
To request resource data with fast-path navigation we simply type an R in the first character of the line that
describes the bottleneck. (Note the R on the second line of the display body in Figure 91.) In this case the
result would be a display like the one shown in Figure 93 on page 192.
Chapter 8. Fast-path navigation
127
+=========================== Paging and Storage ==============================+
| Period: 11:15 to 11:30 on 09/06/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
|
|Virtual |
Average Frame Counts
|
Page/sec
| Swap paging/sec |
|
| Memory | Real = Fixed + Non-Fix| In
| Out | In
Out |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| SQA |
512K |
170 |
170 |
|
|
|
|
|
| LPA | 4380K |
416 |
20 |
396 |
2.3 |
|
|
|
| CSA | 3004K |
183 |
62 |
121 |
.5 |
.3 |
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| LSQA|
|
219 |
219 |
|
|
|
|
|
| Pvt | 7552K | 2226 |
81 | 2145 |
6.5 | 4.4 | 41.2 | 41.3 |
| Free|
|
305 |
|
305 |
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| Nuc |
896K |
224 |
224 |
|
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|Total| 16384K | 3743 |
776 | 2967 |
9.4 | 4.8 | 41.2 | 41.3 |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|
Total Pages / second
|
14.2
|
82.5
|
+=============================================================================+
+========================== Page Dataset Activity ============================+
| Period: 11:15 to 11:30 on 09/06/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
| Space
Dev
Dev
%
% In
I/O's
Pages
Trnsfr |
| Type
Adr
VOLSER
Type
Full
Use
per sec
per IO
Time |
| -------- ----------------- ------------ ------ |
| PLPA
150
MVSA21
3380
57.3
7.7
2.0
1.2
.032 |
| Common
140 SYS021
3380
27.0
5.0
.5
1.8
.058 |
| Local
143
TIME21
3380
33.5
26.3
2.3
8.2
.055 |
| Local
153
TIME22
3380
33.3
25.0
2.3
11.6
.009 |
| Local
141
TSO021
3380
32.1
27.9
2.2
9.9
.013 |
| Local
151
TSO022
3380
32.6
23.1
2.3
11.4
.009 |
+=============================================================================+
Figure 92. SINGLE interval resource display
You can control which resource panels should be associated with a given type of bottleneck, using a
mechanism called an Automatic Analysis Matrix, or A-matrix. This facility is described in “Automatic
analysis” on page 161.
By using fast-path navigation, we have traveled down a tree of displays, beginning with the COMBINEd
interval SUMMARY and ending with some SINGLE interval resource displays. We can back out of this tree
by pressing PF3 or PF15. Each time you press one of these PF keys, the display will “pop up” a level,
where you could select another line with S, D, or R.
Table 12 summarizes the navigational facilities of EPILOG.
Table 12. Fast-Path Navigation Commands
From
Display
To
Display
Navigational
Command
Summary
Detail
D
Combined
Single
S
Summary Wait
Detailed Wait
W
Degradation
XPG
X
Degradation
Resource
R
Workload Degradation
DASD Resource
V
Workload Degradation or DASD
Resource
Shared-DASD Degradation (by
Workload)
J
Workload Degradation or DASD
Resource
Shared-DASD Degradation (by
Performance Group)
P
128
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The commands D, S, W, X, and R are illustrated in this section. The commands V, J, and P support
navigation in a multisystem configuration, as explained in “Navigating in a multisystem environment” on
page 131.
As you can see, we could have gone directly from the COMBINEd interval SUMMARY (Figure 88 on page
126) to a COMBINEd interval DETAIL (or to a COMBINEd detail wait display).
For example, rather than typing an S, we could have entered a D to produce a DETAIL display of the
COMBINEd 1-hour interval from 11:00 to 12:00:
+=============================================================================+
| Performance Group = 2
Symbolic Name = DEV TSO
|
| From 11:00 to 12:00 on 09/06/99
SYSA |
| Average of 4 Intervals
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.09 S 4.3|-> .
.
.
.
.
.
.
.
.
.|
RPrivate Page-In Wait
0.44 S 20.3|-------->
.
.
.
.
.
.
.
.|
|Disk OMON21 143 Que
0.27 S 12.5|-----> .
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.17 S 8.1|--->.
.
.
.
.
.
.
.
.
.|
|SRM Delay (MPL)
0.13 S 6.2|--> .
.
.
.
.
.
.
.
.
.|
|Disk OMON21 143 Act
0.11 S 5.1|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
2.16 S
3569 MVS Transactions Ended
|
|Productivity Index
10%
|
+=============================================================================+
Figure 93. COMBINEd interval DETAIL display
As before, we can type an R in the first character of the line representing the bottleneck to be investigated.
This time we will see a combined resource display for the entire hour, as shown in Figure 94 on page 130.
Chapter 8. Fast-path navigation
129
+=========================== Paging and Storage ==============================+
| Period: 11:00 to 12:00 on 09/06/99
Elap =60:00 M
A083 |
| Combination of 4 Intervals
|
+-----------------------------------------------------------------------------+
|
|Virtual |
Average Frame Counts
|
Page/sec
| Swap paging/sec |
|
| Memory | Real = Fixed + Non-Fix| In
| Out | In
Out |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| SQA |
512K |
168 |
168 |
|
|
|
|
|
| LPA | 4380K |
456 |
20 |
436 |
1.7 |
|
|
|
| CSA | 3004K |
189 |
63 |
126 |
.3 |
.2 |
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| LSQA|
|
205 |
205 |
|
|
|
|
|
| Pvt | 7552K | 2233 |
125 | 2108 |
5.2 | 4.0 | 39.1 | 39.1 |
| Free|
|
315 |
|
315 |
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| Nuc |
896K |
224 |
224 |
|
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|Total| 16384K | 3790 |
805 | 2985 |
7.3 | 4.3 | 39.1 | 39.1 |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|
Total Pages / second
|
11.6
|
78.3
|
+=============================================================================+
+========================== Page Dataset Activity ============================+
| Period: 11:00 to 12:00 on 09/06/99
Elap =60:00 M
A083 |
| Combination of 4 Intervals
|
+-----------------------------------------------------------------------------+
| Space
Dev
Dev
%
% In
I/O's
Pages
Trnsfr |
| Type
Adr
VOLSER
Type
Full
Use
per sec
per IO
Time |
| -------- ----------------- ------------ ------ |
| PLPA
150
MVSA21
3380
57.3
5.5
1.5
1.1
.032 |
| Common
140 SYS021
3380
27.0
3.0
.4
1.6
.047 |
| Local
143
TIME21
3380
34.3
21.6
2.1
9.8
.011 |
| Local
153
TIME22
3380
33.6
21.0
2.1
11.1
.009 |
| Local
141
TSO021
3380
31.3
22.5
1.9
10.9
.011 |
| Local
151
TSO022
3380
32.6
21.6
2.0
11.7
.009 |
+=============================================================================+
Figure 94. COMBINEd interval resource display
You can also display cross-performance data for an interval by placing an X on a display line.If the display
line corresponds to a single interval, data is displayed for that interval only. If the line corresponds to a
combined interval or is from a batch job that extends over multiple intervals, separate display panels are
generated for each RMF-based interval in the combined interval or job elapsed time.
When an X on the display screen is used to generate a cross-performance group display, it is generated
as if the COMPLETE keyword had been included in command; that is, all display panels are generated
when the command is executed. Because of this, you should be careful when you use this feature to
navigate from combined intervals or batch job displays. If the combined interval or batch job spans a large
number of single intervals, the display may take some time to be generated.
The RESOURCE command
If a DETAILed degradation display is currently on the screen, you can enter the command RESOURCE to
automatically display the resource panels associated with the worst bottleneck.
For example, entering RESOURCE while a display like Figure 91 on page 127 was on the screen would
be equivalent to using the R navigational command to select resource panels for the “Private Page-in
Wait” wait reason.
RESOURCE has an advantage over R in that you can supply an AMATRIX keyword to override the
current default automatic analysis matrix, or specify a resource panel directly. For further information
concerning the use of this command, see the “The RESOURCE command.”
130
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Navigating in a multisystem environment
Fast-path navigation allows you to navigate from a workload degradation display (if a DASD performance
bottleneck is shown) to an RDAS resource display. You can also navigate from a workload degradation
display or an RDAS resource display to the Shared-DASD Degradation Analysis display, which shows
cross-system usage of a DASD device. The Shared-DASD Degradation Analysis report shows DASD
usage in two ways: by workload (that is, job, started task, or TSO user) and by performance group. The
codes are J for impact by nonperformance group workloads, P for impact by performance group
workloads, and V for the RDAS display of the volume against which the navigation was selected.
Navigating to a shared-DASD display
Multiple datastore reporting provides special benefits in a shared-DASD environment. If you are viewing
workload degradation displays, and you want to investigate a wait reason against a particular volume with
respect to a shared-DASD situation, you can select against the DASD wait reason and display the RDAS
statistics for all systems (as defined by the active datastore list) that were active against the volume within
the workload display’s time period. You can also navigate directly from the Workload Degradation display
to a Workload Impact on DASD Volume display.
For example, suppose that you display the following EPILOG panel:
+==============================================================================+
| Job = $AOASMP1
JES Number = 3722
Job Steps = 2 / 47 |
| Job Class = P
Acct No = 34245
Input Queue = 10:48 M |
| From
02:26 to 02:27 on 02/16/99
Elap =
1:17 M SYSF |
+------------------------------------------------------------------------------+
|Wait_Reason_____________Time______%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
4.37 S
7.7|--->.
.
.
.
.
.
.
.
.
.|
VDisk COM002 145 Que 29.64 S
52.6|------------=========> .
.
.
.
.|
|Waiting for CPU
9.88 S
17.5|------->.
.
.
.
.
.
.
.
.|
|ECB Wait
4.94 S
8.7|--->.
.
.
.
.
.
.
.
.
.|
|Disk OMON24 313 Act 4.94 S
8.7|--->.
.
.
.
.
.
.
.
.
.|
|Job Elapsed Time
56.25 S
|
|Productivity Index =
38%
|
+==============================================================================+
Figure 95. Navigating to the shared-DASD degradation display (part 1 of 4)
The display shows that job $AOASMP1 is spending about 50% of its time waiting for I/O (as shown by the
wait reason DISK COM002 145 Que). You decide to view resource activity for this DASD device.
Follow these steps:
1. Make sure that all systems to be monitored (in this example, SYSA and SYSF) are active in the
reporter session. You can do this in one of two ways.
a. Verify (through the INQUIRE SUMMARY command) that the systems have been defined in the
initial datastore list.
b. Use the DATASTOR ADD command to allocate the datastores for the current session.
For further information about the active datastore list, see “The datastore list” on page 106.
2. You want to analyze the degradation on workload $AOASMP1 due to Queued I/O against Volume
COM002. Place the cursor to the left of the wait reason and enter a V (as shown in Figure 95 at the
line flagged by an arrow in the left margin). The result is an RDAS panel that displays statistics for
each system’s activity against the volume.
Chapter 8. Fast-path navigation
131
+=========================== DASD Device Activity =============================+
| From:
02:15 to 02:30 on 02/16/99
Elap = 15:00 M
|
| Sysid:
SYSA,SYSF
|
| Merge of 2 Intervals
|
+------------------------------------------------------------------------------+
| VOLSER SMF Dev
I/O -------- Time in Milliseconds --------- % Dev Open|
| Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB
DB Util Dsns|
| ------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- ----|
J COM002 SYSA 145 008 84.4
292
0
3 288
1
63.1 3.0|
|
SYSF 145 008
.3
58
0
44
2 12
.7 1.3|
+==============================================================================+
Figure 96. Navigating to the shared-DASD degradation display (part 2 of 4)
This panel shows DASD device activity for all monitored systems (in this case, SYSA and SYSF)
against the specified device (in this case, volser COM002).
The display is accurate only if you have defined to the reporter session all EPILOG datastores that
contain degradation information against the volume during a required time period. In this example, a
shared-DASD environment between SYSA and SYSF is assumed, and datastores for both systems are
defined to the reporter. If, for example, there is another system that accesses devices common to
SYSA and SYSF, you could define it to the current session by using the DATASTOR ADD command.
In this example, the cross-system DASD device activity panel shows that device COM002 is active on
systems A and F. In order to diagnose the cause of the I/O wait shown in Figure 95 on page 131, you
decide to view workloads that were active (in all systems) against this device.
3. To determine the non-performance group workloads in different systems that were active against the
volume during the current display time period, enter J to the left of the volume (as shown in Figure 96).
The display shown in Figure 98 on page 198 is invoked.
+==============================================================================+
| Volume = COM002
Period = 02:15 to 02:30 on 02/16/99
|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system ACTIVE I-O
|
+------------------------------------------------------------------------------+
|Workload_JES #_Sysid_Time________%__|0___1___2___3___4___5___6___7___8___9___0|
|ASMPBKUP J4661 SYSA
5:32 M
82.1 |------------============>>>>>>>>>> . .|
|$AOASMP2 S4533 SYSF 45.98 S
11.4 |-----> .
.
.
.
.
.
.
.
.|
|TDNY30
T4014 SYSA 26.22 S
6.5 |--> .
.
.
.
.
.
.
.
.
.|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system QUEUED I-O
|
+------------------------------------------------------------------------------+
|Workload_JES #_Sysid_Time________%__|0___1___2___3___4___5___6___7___8___9___0|
|$AOASMP1 J4243 SYSF 29.64 S
66.7 |------------============>>> .
.
.
.|
|ASMPBKUP J8812 SYSA 13.94 S
31.4 |------------==> .
.
.
.
.
.
.|
|$AOASMP2 S4533 SYSF
.84 S
1.9 |>
.
.
.
.
.
.
.
.
.
.|
+==============================================================================+
Figure 97. Navigating to the shared-DASD degradation display (part 3 of 4)
In this display, the active I/O and queued I/O status percentages each add up to 100%. You can see
both the heavily active and the heavily inactive workloads accessing the volume on a comparative
basis.
The percentages reflect the portion of the total observed active or queued I/O against volume COM002
that is attributable to each job. In other words, the number of samples for active I/O against COM002
during the time period 2:15 to 2:30 is accumulated from all eligible records on each datastore in the
active datastore list, with each sample being weighted according to the sampling interval time in effect
for each system involved.
In addition to the total accumulation, an accumulation for each job within the system is also performed.
The percentages are then calculated by dividing the job within system accumulations by the total
accumulation across all jobs in all systems. This calculation is applied separately to active, queued and
reserved I/O wait reasons.
132
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
4. To display the impact of performance groups on volume COM002, substitute the navigation code P for
J in step 3 on page 132 above. The panel shown in Figure 98 is displayed.
+==============================================================================+
| Volume = COM002
Period = 02:15 to 02:30 on 02/16/99
|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system ACTIVE I-O
|
+------------------------------------------------------------------------------+
|Perf Grp_______Sysid_Time________%__|0___1___2___3___4___5___6___7___8___9___0|
|HOTBATCH
SYSA
5:32 M
82.1 |------------============>>>>>>>>>> . .|
|REGBATCH
SYSF 45.98 S
11.4 |-----> .
.
.
.
.
.
.
.
.|
|PERF121
SYSA 26.22 S
6.5 |--> .
.
.
.
.
.
.
.
.
.|
+------------------------------------------------------------------------------+
| Relative Usage of selected DASD device for Cross-system QUEUED I-O
|
+------------------------------------------------------------------------------+
|Perf Grp_______Sysid_Time________%__|0___1___2___3___4___5___6___7___8___9___0|
|REGBATCH
SYSF 30.48 S
68.6 |------------============>>>>.
.
.
.|
|HOTBATCH
SYSA 13.94 S
31.4 |------------==> .
.
.
.
.
.
.|
+==============================================================================+
Figure 98. Navigating to the shared-DASD degradation display (part 4 of 4)
The display shows the symbolic names defined for each performance group in the Perf Grp column. If
these names have not been defined, PERFnnnn is displayed, where nnnn is the performance group
number. In this example, the jobs $AOASMP1 and $AOASMP2 have been defined in the same
performance group, REGBATCH. The queued I/O portion of the display contains a rolled up
percentage for REGBATCH for these two jobs, which appeared separately in the non-performance
group variation of the display.
Expanding the time period
Fast-path navigation does not simply carry forward the time period that was used by the workload
degradation display. If it did so, the resulting shared-DASD display would omit system activity that had
been collected into time intervals that span the original time period. To show the full impact of system
activity, EPILOG expands the navigational time period to cover the complete RMF interval that spans the
original time period. This is done in two ways.
In cases where the systems in the active datastore list have synchronized RMF boundaries, EPILOG
expands the time period covered to span the RMF interval boundaries surrounding the workload panel
time period. In cases where the multiple systems have non-synchronized RMF boundaries, EPILOG
expands the time period to the earliest of the RMF boundaries among the systems involved, with a
tolerance factor of 10%.
For example, in the workload degradation display shown in Figure 95 on page 131, the time period starts
at 2:26 and ends at 2:27. However, the Shared-DASD display in Figure 97 on page 132 shows that volser
COM002 is active in two systems that do not synchronize on the same boundary. SYSA writes RMF
records at 2:15 and 2:30, while SYSF writes RMF records at 2:14 and 2:29.
The tolerance factor is 10% of the RMF interval, or 1.5 minutes. Since the expiration time in SYSF is less
than one minute earlier than the expiration time in SYSA, EPILOG expands the time period to the earlier
RMF interval. In other words, it uses the start time of the earliest RMF interval as the start time of the
navigational time period. The resulting time period is 2:14 to 2:30. As a result of the time boundary
expansion, system activity will be included in the Shared-DASD degradation analysis if the data collection
for that activity started between 2:14 and 2:30.
Note: To ensure writing of started task, TSO user and batch job data on each RMF interval, STCINT,
TSOINT and BATCHINT must be specified in the EPILOG collection options in KEPOPTN. For
further information, see the discussion of the INTERVAL keyword in Table 4 on page 36.
Chapter 8. Fast-path navigation
133
Viewing workload utilization directly
You can generate the Shared-DASD Degradation display directly without navigating from a workload
display, by using the JDAS and PDAS keywords with the DISPLAY command. This feature is similar to
fast-path navigation from a workload degradation display, but allows you greater flexibility in defining the
time period to be monitored. For a description of the JDAS and PDAS keywords, see “Shared-DASD
analysis” on page 44.
134
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Part 2. Advanced reporting features
Part 2 describes the advanced reporting features of EPILOG for MVS. It includes sections that describe
features such as A-Matrix definitions, exception thresholds, the Workload Profile Facility, exporting
historical data, and reporting with SAS graphics.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
135
136
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 9. Advanced reporting functions
EPILOG for MVS provides advanced functions that allow the experienced user to customize reporter
operation for maximum efficiency. This chapter describes the following types of advanced reporting:
v “Setting reporter options”
v “Using the reporter in batch” on page 142
v “Workload exception filters” on page 143
v “Resource exception filters” on page 155
v “Automatic analysis” on page 161
Setting reporter options
You can set four types of reporter options for a given session:
v PF key definitions
v A-matrix definitions
v Summary wait category definitions
v Productivity indexing definitions
All of these features have defaults which are defined in several members of the rhilev.REDPARM
partitioned data set. For PF keys and A-matrices, there are reporter commands to edit these default
members (or create your own member) without exiting from the reporter session. To edit the summary wait
category or productivity index defaults (or create your own member), you should use an external editor
such as ISPF EDIT or TSO EDIT to create or modify appropriate members of rhilev.REDPARM.
Data set naming conventions
See the appropriate chapter in EPILOG for MVS Customization Guide for a detailed explanation of data
set naming conventions.
PF key definitions
Each of the PF keys is assigned a function in EPILOG. Some issue control commands, others have
ISPF-like functions, and a few have been set to issue DISPLAY commands. These functions are simply
defaults; you can change the function associated with most PF keys using the PFK command described in
this section.
If you are using the reporter in split-screen mode, two keys have been reserved for the SPLIT and SWAP
ISPF functions. By default, PF2 is reserved for splitting the screen, and PF9 for swapping screens. To
modify these assignments, you must return to the Split Screen Options Menu.
The following figure describes the default function for each PF key.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
137
CANDLE CORP.
10/18/99 14:36
Mode: PAGE
1 of 2 PFK DISPLAY
CMD==>
*******************************************************************************
P R O G R A M
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
PFK
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
F U N C T I O N
K E Y S
'HELP'
'RESERVED FOR ISPF SPLIT SCREEN'
'BACK'
'CONTROL RECALL'
'CONTROL TITLE'
'DIS RALL TODAY COMBINE STIME(9) ETIME(17)'
'CONTROL SCROLL UP'
'CONTROL SCROLL DOWN'
'RESERVED FOR ISPF SWAP SCREEN'
'DIS PGN(2) TODAY SUMMARY COMBINE(1H)'
'DIS PGN(2) TODAY RIF(RESP(>1S)) AUTO'
'PFK'
'HELP'
'CONTROL MODE'
'BACK'
'CONTROL RECALL'
'CONTROL TITLE'
'DIS RALL TODAY COMBINE STIME(9) ETIME(17)'
'CONTROL SCROLL UP'
'CONTROL SCROLL DOWN'
'CONTROL LOG'
'DIS PGN(2) TODAY SUMMARY COMBINE(1H)'
'DIS PGN(2) TODAY RIF(RESP(>1S)) AUTO'
'PFK'
Figure 99. PF key defaults
A brief explanation of the default settings follows.
PF1/PF13
HELP (no operands)
PF2
RESERVED FOR ISPF SPLIT SCREEN - For details about using the ISPF split-screen
feature, see “Split-screen mode under ISPF” on page 20.
PF3/PF15
BACK - Return to the previous display.
PF4/PF16
CONTROL RECALL - Display the last command issued.
PF5/PF17
CONTROL TITLE - Display the report title block.
PF6/18
DISPLAY - Sample command to display all resource panels for today, averaged together
from 09:00 to 17:00.
PF7/PF19
CONTROL SCROLL UP - Scroll back, toward the top, one report or screen at a time.
PF8/PF20
CONTROL SCROLL DOWN - Scroll forward, toward the bottom, one report or screen at a
time.
PF9
RESERVED FOR ISPF SWAP SCREEN
PF10/22
DISPLAY - Sample command to display an hourly summary of today’s degradation for
performance group 2.
PF11/23
DISPLAY - Sample command to display all intervals today when the TSO response time of
performance group 2 was greater than one second.
PF12/PF24
PFK - Display the current PFK definitions.
PF14
CONTROL MODE - Advances through display modes (PAGE, ROLL, HOLD, ONE). Each
depression of the PF key advances to the next mode.
PF21
CONTROL LOG - Logs the current screen to the EPLOG data set.
138
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The default PF key definitions are contained in the $DEFAULT member of the rhilev.REDPARM data set.
The PFK command described in the next section allows you to change this default member, or to create a
new one which you can invoke during the reporter session.
When the PFK command is processed, the syntax of the PF key definition is checked, and any errors are
flagged with an appropriate message. If you use an external editor to create or modify these members,
syntax checking will be done when the PFK function is executed, and it may not always be obvious that
the error was caused by the faulty PF key definition. For this reason, we recommend that you use the PFK
command described in the next section rather than an external editor to create and maintain PF key
definitions.
The PFK command accepts a 1- to 7-character member name as an operand. The member name you
enter is then prefixed with a dollar sign ($) to generate the actual member name in rhilev.REDPARM. For
example, if you enter:
PFK SAVE (MYPFKS) . . .
The name of the member created in rhilev.REDPARM is $MYPFKS.
The PFK command
The PFK command allows you to perform four basic functions:
v modify a PF key definition
v create a pseudo-PF key definition
v save new PF key definitions in a member of rhilev.REDPARM
v load a set of PF key definitions from a member of rhilev.REDPARM
The syntax of the command is shown below.
PFK [nn 'command']
or
PFK [SAVE(cccc)]
or
PFK [READ(cccc)]
where nn = a PF key number (1-99) and cccc = a PFK member name.
nn ‘command’
If the PF key number is entered with a command, the command function
is assigned to that PF key. You can create up to 99 pseudo-PF key
definitions in this way.
nn
If the pseudo-PF key number is entered by itself, the command associated
with that PF key is executed.
SAVE(cccc)
Saves the changes in the specified PFK member.
READ(cccc)
Reads in the specified PFK member, and uses those PF key definitions for
the current reporter session.
If the command is entered without any keywords, the current PF key definitions are displayed.
Examples
PFK
Displays current PF key definitions.
7
Chapter 9. Advanced reporting functions
139
Executes the command associated with PF7 (by default, ‘CONTROL SCROLL UP’).
PFK 22 'DISPLAY PGN(2) TODAY'
Assigns the specified command to PF22.
PFK SAVE(MYPFKS)
Saves the current PF key definitions in the $MYPFKS member of the rhilev.REDPARM partitioned data
set. (Note that the name you specify is prefixed with a dollar sign ($) to name the member in
rhilev.REDPARM.)
PFK READ(MYPFKS)
Reads in the $MYPFKS member of rhilev.REDPARM and uses it as the current PF key definitions.
Summary wait categories
The summary wait category definitions are invoked under two circumstances:
1. When the SUMWAIT keyword is included on the DISPLAY command, and
2. When a summary wait category is used as a wait reason in exception analysis (that is, with the
REPORTIF keyword).
The purpose of these definitions is to group detailed wait reasons into summary wait categories. These
categories are defined in a member of the rhilev.REDPARM partitioned data set called DEFAULTS.
You can create your own member of summary wait definitions using an external editor (such as ISPF EDIT
or TSO EDIT) to copy and modify this member. When you do so, observe the following guidelines:
1. Each definition associates a detailed wait reason with a summary wait reason. Valid detailed wait
reason codes and their associated summary wait categories are described in Table 14 on page 146.
2. Both the detailed wait reason code and the summary wait reason must be enclosed in single quotes.
(The equal sign between them is optional.)
3. Each summary wait reason consists of a long and short definition. The long definition is used on
workload degradation displays, and can be up to 20 characters in length. The short form is recognized
by the REPORTIF exception keyword and used to construct the XPG display, and is 4 characters long.
(If you do not define a short form for a summary wait reason, then the first 4 characters of the wait
reason name are used to construct the short form.) We recommend that you define short forms for
each summary wait reason to avoid confusion.
4. You must be careful in naming the summary wait category so that the short form (whether explicitly
specified or derived from the first four characters of the full name) does not match any of the detailed
wait reason short forms. If a short form matches a long form, this creates an ambiguity for the
command parser when the term is used with the REPORTIF keyword.
The only exception to this occurs when only one detailed wait reason is defined for a summary wait
category. This is the case with the CPUTIL summary wait category, which consists only of the Waiting
for CPU (CPU) detail wait reason.
You invoke the new definitions with the SET command. For example, if you copied DEFAULTS to a
member called MYSUMS in rhilev.REDPARM, and then modified the summary wait definitions in it, you
would invoke it during a reporter session by issuing the following command:
SET SUMDEF(MYSUMS)
The definitions in MYSUMS would then be used anytime the SUMWAIT keyword was used.
To reset the definitions back to the defaults, simply enter:
SET SUMDEF(DEFAULTS)
140
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Productivity index definitions
In order to calculate the productivity index on a workload degradation display, the reporter uses the
productivity index definitions contained in a member of the rhilev.REDPARM partitioned data set called
DEFAULTP. The definitions in the DEFAULTP member group detailed wait reasons into three categories:
Productive
States are those in which the workload is executing in some fashion, and
are not considered degradation. By default, Using CPU, Active I/O, and
STIMER waits are considered productive states.
Unproductive
States are those in which the workload is waiting to execute, but is being
prevented from doing so because a resource is not available. All detailed
wait reasons are considered unproductive unless they are specifically
defined otherwise in the DEFAULTP member.
Idle
States are those in which the workload is not doing productive work, but is
not being prevented from doing so. By default, terminal input waits are
considered an idle state.
Each line in the productivity index associates a detailed wait reason with one of these categories.
Remember, if a wait reason is not defined, it is considered unproductive by default.
If you wish to set up your own set of productivity index definitions, copy the DEFAULTP member into
another member of rhilev.REDPARM and then modify your own definition member. Observe the following
guidelines:
1. Each line in the definitions associates a detailed wait reason code with one of two categories (PROD
and IDLE). Valid detail wait reason codes are described in Table 14 on page 146.
2. Both the detail wait reason code and the category must be enclosed in single quotes. (The equal sign
between them is optional.)
3. The definition must begin with the PRODIDX keyword.
The productivity index shown on the display screen is calculated by dividing the time spent doing
productive work by the time spent productively and unproductively:
Prod. Index =
Productive reasons
Productive + Non-productive reasons
Figure 100. Productivity index calculation formula
Time spent in idle states is not relevant to degradation, and is disregarded in the calculation of the
productivity index.
You invoke the new definitions with the PRDXDEF keyword and the DISPLAY or SET commands. For
example, if you copied DEFAULTP to a member called MYINDEX in rhilev.REDPARM, and then modified
the productivity index definitions in it, you could invoke it during a reporter session by issuing the following
command:
SET PRDXDEF(MYINDEX)
The definitions in MYINDEX would then be used to calculate the productivity index anytime a workload
degradation display was generated. To reset the definitions back to the defaults, simply enter:
SET PRDXDEF(DEFAULTP)
Chapter 9. Advanced reporting functions
141
Using the reporter in batch
A sample jobstream that invokes the reporter in batch is distributed in rhilev.REDDATA(KEPPROC), or in
rhilev.REDDATA(KEDPROC) if both EPILOG and DELTAMON® are installed at your site. This jobstream
performs the necessary allocations and reads the input commands from a data set pointed to by the
EPSYSIN DD statement. Output is directed to a data set pointed to by the EPMSG DD statement.
Because JCL is system-level dependent, we recommend that you read the KEPPROC or KEDPROC
member and tailor it to your own needs.
There are two reporter functions that are especially useful in batch mode: the PAGESEP command and
the LIMIT keyword. These are described below.
The PAGESEP command
The PAGESEP command allows you to design and print page separators between batch reports. The
keywords used with this command are summarized in the following table.
Table 13. Batch report page-separator keywords
Keyword
Short
BLOCK
DUPLICATE
TITLE
TITLE2
TITLE3
DUP
Description
Operand
Formats the specified report title in 5x7 block letters.
A report title
Specifies the number of separator pages to print. (default=1)
0 to 9
Produces single, double, or triple spaced report titles.
A report title
The following example illustrates how the PAGESEP command is used to set up a title page.
PAGESEP DUP(3) BLOCK('ROOM 1') TITLE3('SPECIAL REPORT') TITLE('TO THE DIRECTOR')
Note that TITLE, TITLE2, TITLE3 and BLOCK may be used on the same command.
The PAGESEP command above produces the following page separator three times on the data set
identified by the EPREPT DD statement.
*******************************************************************************
*
*
*
*
*
****
***
*** *
*
*
*
*
*
* *
* *
* ** **
**
*
*
*
* *
* *
* * * *
* *
*
*
**** *
* *
* *
*
*
*
*
* *
*
* *
* *
*
*
*
*
* * *
* *
* *
*
*
*
*
*
* ***
*** *
*
*****
*
*
*
*
*
*
SPECIAL REPORT
*
*
TO THE DIRECTOR
*
*
*
*
*
*******************************************************************************
Figure 101. Sample PAGESEP output
142
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
The LIMIT keyword
In addition to using the PAGESEP command to create separator pages, you may also want to use the
LIMIT keyword with the DISPLAY command under certain circumstances. The LIMIT keyword determines
the maximum number of panels in the output report for a given DISPLAY command. For example, if you
enter:
DISPLAY PGN(2) SIF(RESP >2S) LASTWEEK LIMIT(20)
you may not know beforehand how many intervals will meet the exception criteria. However, because the
LIMIT keyword is set to 20, only the first twenty intervals will be included in the output. This keyword puts
a safety net on commands that have the potential to produce voluminous unwanted output.
By default, LIMIT is set to 10 when the reporter is run in batch.
Workload exception filters
The DISPLAY command accepts exception filters to select only records that meet certain selection criteria.
This section describes exception filters for workloads. Resource exception filters are described in
“Resource exception filters” on page 155.
In performance group analysis, you may want to select only intervals where response time was greater
than two seconds, or where private page-in waits accounted for less than 20 percent of the total
transaction’s lifetime. For batch jobs, you may only be interested in analyzing long-running jobs (say,
where the elapsed time exceeds 1 hour), or those executions of a particular program in which the I/O
waits accounted for more than 25% of the program’s run time.
To specify exception filters, add the SELECTIF or REPORTIF keyword to the DISPLAY command, along
with a valid wait reason and appropriate selection criteria.SELECTIF and REPORTIF can be shortened to
SIF and RIF, respectively. The syntax for both of these keywords is shown below.
SELECTIF
or
REPORTIF
(keyword1(pnnq),keyword2(pnnq) ...)
Figure 102. Exception Analysis Keyword Syntax
keyword
Defines the exception filter. Valid exception filters can be:
v Detail wait reasons, which are described in Table 14 on page 146.
v Summary wait reasons, which are described in Table 7 on page 53.
v Response time (RESP) for performance groups.
v Elapsed time (ELAP) for batch jobs, started tasks, or TSO user sessions.
pnnq
Sets the quantity at which the exception is reached. Note that this expression must be enclosed by
parentheses (as shown) or preceded by a blank space.
p
> for greater than (default), < for less than.
nn
The value at which the exception is reached, expressed as a decimal number (25,
.5, 37.6).
q
Units (default is %)
%
percent
<10 (less than 10 percent-- default)
<10% (less than 10 percent)
10 (greater than 10 percent--default)
S
seconds
Chapter 9. Advanced reporting functions
143
<1.5S (less than 1.5 seconds)
>56S (greater than 56 seconds)
M
minutes
10M (greater than 10 minutes)
H
hours
As shown in Figure 102 on page 143, more than one exception filter can be specified in a
command.
If you want to display all periods for which no waits of a certain type are recorded, add the prefix
NO to the exception type keyword numeric argument. For example,
NOSWI
eliminates degradation caused by Waiting to be swapped in from the display. (SWI is the short
name for the wait reason keyword SWAPIN, as shown in Table 14 on page 146.) Note that the
keyword NOSWI is the equivalent of
SWI(<1%)
because EPILOG rounds a wait percentage of less than 1 to 0.
NO is not valid when used with the ELAPSED, RESPONSE, and CPUTIL keywords.
Examples
To display all intervals today where performance group 2 first-period response time was greater than 1
second, enter:
DISPLAY PGN(2) PGP(1) TODAY
SIF(RESPONSE(>1S))
To display all those intervals where response time was greater than .5 seconds and private page-in waits
made up greater than 30% of the average transaction wait time, enter:
DISPLAY PGN(2) PGP(1) TODAY
SIF(RESP(>.5S) PAG(>30%))
To display degradation data for all jobs run last week with elapsed time greater than 30 minutes, enter:
DISPLAY JOB(*) LASTWEEK
SIF(ELAP(>30M))
To display degradation data for all runs of the ARUPDATE job last month in which the elapsed time was
greater than 30 minutes, enter:
DISPLAY JOB(ARUPDATE) LASTMONTH
SIF(ELAP(>30M))
To display degradation data for all TSO sessions yesterday which were longer than 1 hour, enter:
DISPLAY TSO(*) YESTERDAY SIF(ELAP(>1H))
To display degradation data for all started tasks beginning with CICS in which CPU waits accounted for
more than 5% of the total degradation, enter:
DISPLAY STC(CICS*) YESTERDAY SIF(CPW(>5))
To display degradation data for all started tasks beginning with CICS in which I/O waits accounted for
more than 10% of the total degradation, enter:
DISPLAY STC(CICS*) YDAY RIF(I/O(>10))
I/O is a summary wait category defined in the default summary wait member of the rhilev.REDPARM
partitioned data set.
144
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
If you are not using the default members, the summary wait reason included in the exception request must
be defined as a valid summary wait category. The default summary wait categories are described in
Table 7 on page 53.
You can optionally include the SUMWAIT keyword with the command, even if you are using a summary
wait category as the exception criteria. If you do not include SUMWAIT, the display will show detail wait
reasons even though a summary wait category was used as the exception criteria.
Note that most of these examples use the SELECTIF keyword (SIF) instead of REPORTIF (RIF). In most
cases, REPORTIF could have been used in place of SELECTIF. (The example involving summary waits is
the only exception; summary waits may only be specified with REPORTIF, never with SELECTIF.)
The differences between these keywords are described in “SELECTIF and REPORTIF processing” on
page 151.
Response time and elapsed time (RESPONSE and ELAPSED)
The RESPONSE and ELAPSED keywords (RESP and ELAP) allow you to select performance group
records by response time, and other workloads (batch jobs, started tasks, and TSO sessions) by elapsed
time. These keywords cannot be entered with a percentage operand. They require a specific time, such as
2S, 30M, or 1H. Typical commands using these keywords are:
DISPLAY PGN(2) SIF(RESP(>2S)) LASTWEEK
DAYOFWK(WEDNESDAY)
This command displays only those intervals on Wednesday of last week in which the response time of
performance group 2 was greater than 2 seconds. (Presumably, performance group 2 defines a group of
online TSO users.)
DISPLAY JOB(PAYROLL) STEP RIF(ELAP(>45M)) LASTWEEK
This command displays only those steps of the PAYROLL jobs last week that ran longer than 45 minutes.
DISPLAY TSO(*) RIF(ELAP(>1H)) LASTWEEK
This command displays only those TSO sessions that ran longer than 1 hour.
DISPLAY PNAME(SHIFT1) JOB(*) RIF(ELAP(>1H)) LASTWEEK
This command displays only those batch job profile records with a PNAME of SHIFT1 for which the
elapsed time exceeded 1 hour. (The PNAME (or PROFNAME) keyword is described in “Creating profiles”
on page 168.)
The RESPONSE and ELAPSED keywords function similarly when used with the COMPARE command,
which compares workload degradation to profile degradation. (The COMPARE command is described in
“Comparing workloads with profiles” on page 183.) They can, however, accept the time operand with a
percentage. Some typical examples are shown below.
COMPARE PGN(2) PNAME(SHIFT1) RIF(RESP(>2S)) LWK DAY(WED)
This command compares workload degradation of performance group 2 last week on Wednesday to the
most recent profile record for performance group 2 with PNAME SHIFT1. Only those intervals in which the
response time of performance group 2 exceeded 2 seconds are displayed.
When the percent sign is included with the time operand, both the profile record and the workload record
are used to make the selection. In the previous example, the selection is based only upon the workload
record when a specific time is stated.
For example, if you enter
COMPARE PGN(2) PNAME(SHIFT1) RIF(RESP(>25%)) LWK
DAY(WED)
Chapter 9. Advanced reporting functions
145
only those comparisons in which the response time of the workload is 25% greater than that of the profile
are displayed.
Detail wait reasons
Table 14 shows a list of valid detail wait reasons that can be used with the SELECTIF and REPORTIF
keywords.For each keyword, the table shows the short name, the summary wait reason category (as
defined in Table 7 on page 53), and a description of the wait reason or execution state.
MVS/SP 4.2 introduces the Advanced Program-to-Program Communication (APPC) transaction processor
service. APPC transaction processors are treated as started tasks by the EPILOG collector and reporter.
Table 14. Detail wait reason codes
Keyword
Short
Summary
Description
ACTIVEIO
AIO
I/O
Delay due to an active I/O in progress. See the note following the table.
APPCWAIT
APS
SWAP
Delay due to voluntary swap-out of an idle APPC address space (MVS/SP
4.2 and above).
AUXSTOR
AUX
SWAP
Delay due to swaps caused by a shortage of auxiliary space. Auxiliary
space shortage is due to a high allocation of local page slots.
BACKUP
BKP
HSM
Delay due to waiting for HSM to execute a data set backup request.
CANCEL
CAN
JES
Delay due to a job cancel request issued to JES.
CENTSTOR
CSS
SWAP
Delay caused by swap-out to improve central storage availability (MVS/SP
4.2 and above).
CILOCATE
CIL
HSM
Delay due to waiting for HSM to execute a locate request for a control
interval.
COMMON
COM
PGSW
Delay due to address spaces waiting for a PLPA or common page-in.
CPUTIL
CPU
CPU
Time spent using CPU.
CPUWAIT
CPW
CPW
Delay due to address spaces waiting on the active CPU dispatching queue.
High percentages here may mean that the address space is low priority or
that there is a shortage of CPU cycles.
DELETEDS
DLD
HSM
Delay due to waiting for HSM to delete a data set.
DELETEJOB
DEL
JES
Delay due to a job delete request issued to JES.
DETECTED
DET
SWAP
Delay due to swaps caused by a detected wait. A detected wait swap
occurs when a user address space has not executed in more than 8 SRM
seconds or 2 seconds, whichever is less (in current versions of MVS), and
has not issued a long WAIT.
DISKMOUNT
DMP
OPER
Delay due to address spaces waiting for an operator to mount a disk.
ECBSTIMER
ECS
MISC
Delay due to ECB waits with an STIMER TASK or REAL pending (see
ECBWAIT).
ECBWAIT
ECB
MISC
The address space has issued a voluntary wait for some reason. Examples
are an IMS message region waiting for work, a CICS region waiting for
work, or even OMEGAMON; for MVS waiting between screen refreshes.
High percentages of ECB waits may indicate a program error, normal
voluntary waits, or, if the Candle Subsystem is not running, a system
reconfiguration.
ENQEXC
EEX
SWAP
Delay due to swaps performed to make room for users that are enqueued
on a resource required by other users.
ENQUEUE
ENQ
ENQU
Delay due to waiting for an enqueue. See the note following the table.
EXCHANGE
EXG
SWAP
Delay due to exchange swaps. Exchange swaps are performed when one
user in a given domain must be swapped out to make room for another
user in the same domain that has a higher recommendation value.
146
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 14. Detail wait reason codes (continued)
Keyword
Short
Summary
Description
HLIST
HLS
HSM
Delay due to waiting for HSM to execute an HLIST command.
JOBSTATUS
JST
JES
Delay due to a job status request issued to JES that has been queued.
LONGWAIT
LON
SWAP
Delay due to swaps caused by user-requested long waits.
MIGRATE
MIG
HSM
Delay due to waiting for HSM to migrate a data set.
MVSLOCK
LCK
MISC
Delay due to address spaces waiting to acquire a local or global MVS lock.
OUTLONG
OLS
SWAP
Delay caused by swap-out to enable the swap-in of an address space that
has been swapped out too long (MVS/SP 4.2 and above).
PAGEIN
PAG
PGSW
Delay due to address spaces waiting for a private page-in operation.
QUEUEDIO
QIO
I/O
Delay due to waiting for queued I/O. See the note following the table.
READCTL
RCR
HSM
Delay due to waiting for HSM to read a control data set record.
READY
RDY
SRMD
SRM Delay (MPL); this can occur when a domain is at the maximum MPL
and a unilateral or exchange swap must occur to get the ready work
swapped in. The READY delay occurs while waiting for SRM to allow the
swap. When the ASM has scheduled a swap-in, any subsequent delay will
then show up in the Swap Page-In (SWAPIN) category.
REALSTOR
RST
SWAP
Delay due to swaps caused by a shortage of real pageable frames.
RECALL
RCL
HSM
Delay due to waiting for HSM to recall a data set.
RECOVER
RCV
HSM
Delay due to waiting for HSM to recover a backup data set.
REQUEST
REQ
SWAP
Delay due to requested swaps.
REQUEUE
RQU
JES
Delay due to a job requeue request to JES.
RESRVEIO
RIO
I/O
Delay due to wait for RESERVEd I/O (MVS/370 only).
SRMRTO
RTO
MISC
Delay imposed by SRM to meet the response time objective set in the IPS.
STAGING
MSS
MISC
Waiting for a mass storage volume to be staged.
STIMER
STI
MISC
The address space has issued an STIMER and is voluntarily waiting for it
to end. This is typically not considered degradation since it is a voluntary
wait.
SWAPIN
SWI
PGSW
The address space has been given to the ASM queue and is waiting for
MVS to swap it into storage. A high value for this wait can be caused by
I/O problems such as a slow I/O device or DASD contention.
SYSOUT
PSO
JES
Delay due to a process SYSOUT request issued to JES that has been
queued.
TAPEMOUNT
TMP
OPER
Delay due to tape mounts. The percentage indicated in the degradation
data reflects the percentage of time in which the tape mount pending
condition was in effect. Note that this may not always represent
degradation.
For example, an application may issue a tape mount request some time
before it actually requires I/O to the data on that tape. (This is often the
case with such applications as CICS, which can perform asynchronous
I/O.) Under these conditions, the job continues running while the tape
mount pending condition is in effect, with no degradation to that job.
Remember also that the tape mount pending condition is in effect from the
time that the tape mount request is issued to the operator console to the
time that the tape is actually mounted and the tape drive is made READY.
For those installations that use them, this time may actually be spent
waiting for a tape management system, not a system operator.
Chapter 9. Advanced reporting functions
147
Table 14. Detail wait reason codes (continued)
Keyword
Short
Summary
Description
TERMIN
TIN
TERMI
Delay due to swaps caused by waiting for terminal input.
TERMOUT
TOU
SWAP
Delay due to swaps caused by a TSO session that is waiting for an output
buffer or waiting for the user to press Enter after “***” is displayed in line
output mode.
TRANSWAP
TSW
SWAP
Delay due to swap-outs that occur because an address space was made
non-swappable. (When an address space is made non-swappable, it has to
be swapped out and then swapped back in with non-swappable status.)
UNILATERAL
UNI
SWAP
Delay due to swaps that occur because the MPL for a domain exceeded
the target MPL.
WTORWAIT
WTO
OPER
Delay due to a long or detected wait swap with WTOR pending.
SYSPAGE
SPS
SWAP
Delay caused by swap-out to reduce the system page fault rate (MVS/SP
4.2 and above).
Note: The ACTIVEIO, ENQUEUE, QUEUEDIO, and RESRVEIO keywords all involve specific I/O and
enqueue waits, and may be entered with additional operands. These operands are described in
Table 16 on page 150.
Summary wait reason codes
Table 15 shows the default summary wait categories and the detail wait reasons included in each category.
Table 15. Default summary wait categories
Summary Wait
Categories
Detail Wait Reasons
CPU
Using CPU
CPW
Waiting for CPU
I/O
Active on an I/O device
Waiting for a logical channel
Queued on an I/O device
Waiting for a reserved I/O device
SWAP
Auxiliary storage shortage swaps
Detected wait swaps
Enqueue exchange swaps
Exchange swaps
User-requested long wait swaps
Real storage shortage swaps
Requested swaps
Terminal output wait swaps
Transition swaps
Unilateral swaps
APPC wait swaps
Central storage swaps
Out too long swaps
System paging swaps
PGSW
Waiting for a PLPA or common page-in
Waiting for a private page-in
Waiting to be swapped in
148
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 15. Default summary wait categories (continued)
Summary Wait
Categories
Detail Wait Reasons
HSM
Waiting
Waiting
Waiting
Waiting
Waiting
Waiting
Waiting
Waiting
for
for
for
for
for
for
for
for
a data set backup
a C/I locate
a data set deletion
an HLIST user command
a data set migration
a control data set record
a data set recall
backup data set recovery
JES
Waiting
Waiting
Waiting
Waiting
Waiting
for
for
for
for
for
a
a
a
a
a
OPER
Waiting for a disk mount operation
Waiting for a tape mount operation
Swapped with WTOR pending
MISC
ECB waits with STIMER TASK or REAL pending
Voluntary waits
Waiting to acquire an MVS lock
SRM delay to meet the response time objective
Waiting for a mass storage volume to be staged
Waiting for a STIMER to end
ENQU
Waiting for an enqueue
SRMD
Waiting for SRM to allow a swap
TERMI
Waiting for terminal input
job cancellation
job deletion
queued job status request
job requeue request
queued SYSOUT request
I/O and enqueue exceptions
Exceptions involving I/O and enqueue waits may be entered with an additional operand to identify the
specific device type and address or enqueue type. The format for doing so is shown below.
DIS wkl SIF(keyword1(type pnnq) keyword2(type pnnq) ...)
DIS wkl RIF(keyword1(type pnnq) keyword2(type pnnq) ...)
where:
wkl
Specifies a workload type (for example, PGN(2) or JOB(USERB)).
keyword
Specifies is an I/O or enqueue wait reason (ACTIVEIO, ENQUEUE, QUEUEDIO, or
RESRVEIO).
type
Specifies a device type and address or enqueue type. Valid device and enqueue type
keywords are described in Table 16 on page 150.
p
Specifies > or < (greater than or less than); it must be preceded by a blank or a delimiter.
nn
Specifies a decimal number (25, .05, 37.6); it cannot be preceded by a blank space.
q
Specifies a unit type (% = percent, S = seconds, M = minutes, H = hours).
Examples
To display all intervals in which performance group 2 was performing active I/O to VSRESA more than 5%
of the time, enter:
DISPLAY PGN(2) SIF(AIO(VOLUME(VSRESA)>5%)) THISWEEK
Chapter 9. Advanced reporting functions
149
To display all intervals in which started task IMSCNTL waited for queued I/O to any device more than 10%
of the time, enter:
DISPLAY STC(IMSCNTL) SIF(QIO(ANY >10%)) INTERVAL YDAY
To display all intervals in which any TSO user session waited for an enqueue more than 1 second, enter:
DISPLAY TSO(*) SIF(ENQ(ANY >1S)) INTERVAL TODAY
Valid device and enqueue operands are shown in Table 16.
Table 16. I/O and enqueue operands
Operand
Short
Description
ALLWAITS
Select if the sum of all the I/O or ENQ waits exceeds the threshold.
ANY
Select if any I/O or ENQ wait exceeds the threshold.
CU(nnnnX)
Exception applies only to the specified control unit (nnnn = the control unit
number; the suffix X is required). (MVS/370 only) Note 1
DASD(cuu or nnnn)
D (cuu or
nnnn)
Exception applies only to specified DASD device (cuu = the device
address; nnnn = the device number). Note 1
DISK(cuu or nnnn)
D (cuu or
nnnn)
Exception applies only to specified DASD device (cuu = the device
address; nnnn = the device number). Note 1
EVERY
Select if all I/O or ENQ waits exceed the threshold.
LCH(n)
L(n)
Exception applies only to specified logical channel (n = the logical channel
number). (MVS/370 only) Note 1
MAJOR(enqname)
M(enqn)
Exception applies only to specified major enqueue name. This operand
requires an additional argument. Note 2
TAPE(cuu or nnn)
T (cuu or
nnnn)
Exception applies only to specified tape device (cuu = the device address;
nnnn = the device number). Note 1
UNIT(cuu or nnn)
U (cuu or
nnnn)
Exception applies only to specified disk or tape device (cuu = the device
address; nnnn = the device number). Note 1
VOLUME(volser)
V(volser)
Exception applies only to device with specified volser. Note 1
Notes:
1. These options are allowed with the NO prefix. (Only select if no waits were seen for this wait reason.)
2. A list of available arguments for the MAJOR operand is shown in Table 17.
Enqueue major names are defined during the customization procedure in the OPTIONS statement
contained in rhilev.REDDATA(KEPOPTN). See the EPILOG for MVS Customization Guide for further
details.
The defaults that are supplied with the product are described in Table 17. Keep in mind that if these
defaults have been customized during product installation, you will have to use the enqueue names
defined in the modified version.
Table 17. Enqueue major name defaults
Major Name
Description
MISC SYS
Miscellaneous SYS prefixed enqueue. Any detected enqueue that does not match the names
specified with the ENQUEUE keyword on the OPTIONS statement is put into one of two
categories: MISC SYS or MISC USR. MISC SYS contains system enqueues. If you specify this
category as a major name with REPORTIF or SELECTIF, you must enclose it in quotes because it
contains a blank space.
150
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 17. Enqueue major name defaults (continued)
Major Name
Description
MISC USR
Miscellaneous USR prefixed enqueue. Any detected enqueue that does not match the names
specified with the ENQUEUE keyword on the OPTIONS statement is put into one of two
categories: MISC SYS or MISC USR. MISC USR contains user enqueues. If you specify this
category as a major name with REPORTIF or SELECTIF, you must enclose it in quotes because it
contains a blank space.
SYSDSN
System major for data set integrity.
SYSIEA01
Dump data set.
SYSIEFSD
Scheduler and allocation serialization.
SYSIEWLP
Linkage editor syslmod serialization.
SYSIGGV1
Master catalog open.
SYSIGGV2
VSAM catalog serialization.
SYSIKJBC
Broadcast data set serialization.
SYSSMF01
SMF data set serialization.
SYSVSAM
VSAM CI serialization.
SYSVTOC
DADSM VTOC update serialization.
SYSZJES2
JES2 spool space. For JES3 systems, modify the major name default to SYSZJES3.
SYSZVARY
CPU reconfiguration.
SELECTIF and REPORTIF processing
Special considerations arise when exception thresholds are applied to combined data. Specifically, the
question arises whether to apply the exception filter before or after the data is combined. The SELECTIF
and REPORTIF keywords allow you to control when the filter is applied.
During display processing, two types of combining take place:
1. Implicit combining is done under four circumstances.
v When multiple performance group numbers are specified as operands to the PGN keyword, data for
these performance groups is averaged together to form a report group.
v When the AVERAGE and TOTAL keywords are used, interval data is combined together for the
entire time period (as specified with the time/date keywords).
v When the JOB or STC keyword is used without the STEP modifier, step records are added together
to form a single display panel for the job or task.
v If data is being collected at intervals and the JOB, STC, or TSO keyword is used without the
INTERVAL modifier, multiple intervals may be combined into a single display panel. For batch jobs
and started tasks, this may occur even if the STEP keyword is used, since one step may span two
or more intervals.
In each case, individual data records are combined together to form a display.
2. Explicit combining is the result of using the COMBINE keyword. For performance groups, this keyword
combines together RMF-based intervals into larger reporting intervals. For batch workloads, it
combines together multiple executions of a job or step type.
The SELECTIF keyword applies the exception filter before any combining (implicit or explicit) takes place.
The REPORTIF keyword applies the exception filter after all combining takes place.
Chapter 9. Advanced reporting functions
151
Since performance group degradation data is collected and kept as individual performance group records
for each RMF-based interval, the SELECTIF keyword applies the exception filter to these records. Since
batch workload information is stored as step level records, SELECTIF applies the filter to individual job
steps.
There is one exception to this rule; it occurs when batch job, started task, and TSO session data are
combined. In these cases, implicit data combining may take place at two levels:
v If data was collected at RMF-based intervals and the INTERVAL keyword is not specified in the
DISPLAY command, the intervals are combined into larger units, such as steps, jobs, tasks, or
sessions.
v If data was not collected at intervals, but a job or started task consisted of multiple steps, then data was
written to the EDS at end-of-step. If the DISPLAY command does not request step level information,
then these step records are combined into a single display panel for each job or started task that meets
the criteria.
Exception criteria specified with the SELECTIF keyword will be applied after these types of implicit
combining have been done.
For example, if JOBA was collected at intervals, and you specify
DISPLAY JOB(JOBA) YESTERDAY SELECTIF(ELAP(>30M))
only one display panel would be generated for each run of JOBA yesterday.
Assuming that the job spanned more than one interval, multiple interval records in the EDS would be
combined to produce each display panel. The SELECTIF criteria would not be applied to the individual
interval records. It would be applied to each job record that was created by combining those interval
records.
If the job was not being collected at intervals, but it consisted of multiple steps, multiple step records would
be combined to produce each display panel. The SELECTIF criteria would be applied to the job record,
not to the individual step records that were combined to produce it.
REPORTIF processing is done just like it is in other cases; the criteria is applied to the combined job
record after the COMBINE keyword has been applied.
The example following the figure illustrates how the difference in processing can affect the data in the
display.
152
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Step 1
┌─────────────────┐
│ Read another
│
│
│
│
│
│
EDS
│
└────────┬────────┘
│
│
│
│
│
│
│
│
│
│
│
│
│
│
│
│
│
┌─────────────────┐
│
│ Does the record │
No
│
│ meet SELECTIF ├───────────┘
│
criteria?
│
└────────┬────────┘
│ Yes
│
│
│
│
│
│
│
│
Step 2
│
┌─────────────────┐
│
Step 3 │ Are there more │
└───────────┤ records to be
│
Yes
│
COMBINEd?
│
└────────┬────────┘
│ No
│
│
│
│
│
│
│
│
┌─────────────────┐
│ COMBINE data
│
│ into a larger │
│ interval record │
└────────┬────────┘
│
│
│
│
│
│
│
│
│
Step 4
Step 5
Step 6
┌─────────────────┐
│
│ Does the record │
No
│
│ meet REPORTIF ├─────────────────┘
│
criteria?
│
└────────┬────────┘
│ Yes
│
┌─────────────────┐
│
DISPLAY
│
│
│
│
the data
│
└─────────────────┘
Figure 103. Exception processing on combined performance group data
The steps shown in Figure 103 are explained below.
Step 1
A record is selected from the EPILOG datastore based upon date/time and workload
criteria. This record usually represents one RMF interval.
Step 2
The record is examined to determine whether it meets the SELECTIF exception threshold.
If it does meet the threshold, it is passed to the next step. If the record does not meet the
threshold, it is discarded and the process returns to step 1.
Step 3
If there are more records that belong in the combined interval, the process returns to step
1. If not, the process continues to step 4.
Step 4
The records that have not been eliminated by step 2 are combined into a single record
that represents the combined interval.
Step 5
The record produced in step 4 is examined to determine whether it meets the REPORTIF
exception threshold. If it does meet the threshold, it is passed to the next step. If the
record does not meet the threshold, it is discarded and the process returns to step 1.
Step 6
The data is displayed.
Chapter 9. Advanced reporting functions
153
Note: Although the DISPLAY command syntax allows you to use both SELECTIF and REPORTIF in the
same command, we recommend that you use either one or the other. The following example
illustrates how these keywords differ, and you can see how confusing the results would become if
both were used.
Suppose that you were looking at performance group 2 response time from 10:00 AM to 1:00 PM,
combining 15 minute RMF intervals into 1 hour intervals, and were only interested in intervals in which the
response time was greater than 1 second. The following table shows how the data gets selected and
combined when the SELECTIF keyword is used. (For the purposes of the example, assume that the
number of transactions per interval remains constant.)
15-Minute
Interval Value
SIF(>1S)
Combine
Final Value
10 to 11
0.1
2.0
0.3
0.7
---2.0
-------
2.0
2.0
11 to 12
0.6
0.3
0.2
0.5
-------------
----
----
12 to 1
1.4
1.8
0.7
0.3
1.4
1.8
-------
1.6
1.6
Hourly Interval
Now, we’ll see how the same intervals are selected when the REPORTIF keyword is used.
Hourly Interval
15-Minute
Interval Value
Combine
RIF(>1S)
Final Value
10 to 11
0.1
2.0
0.3
0.7
0.775
----
----
11 to 12
0.6
0.3
0.2
0.5
0.4
----
----
12 to 1
1.4
1.8
0.7
0.3
1.05
1.05
1.05
You can see from these two tables that interval selection can vary widely depending on whether the
SELECTIF or REPORTIF keyword is used. The interval from 10:00 to 11:00 was evaluated with a
combined response time of 2.0 seconds when the SIF keyword was used, and was eliminated by the
exception criteria when the RIF keyword was used. Also, the interval from 12:00 to 1:00 was evaluated to
two different values, depending on whether RIF or SIF was being used to define the exception criteria.
When you apply exception criteria to combined intervals, you should be aware of the differences between
the RIF and SIF keywords, and use the keyword that is appropriate to your situation.
154
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Resource exception filters
The DISPLAY command accepts exception filters for the resource display panels. Resource exception
filters are defined by using DISPLAY command with the REPORTIF keyword.
The REPORTIF keyword
The REPORTIF keyword (RIF, for short) allows you to specify exception criteria for resources by entering
a field name, a comparator, and a value. When the resource record is retrieved from the EPILOG
datastore, the value of that field is compared to the value entered on the command. If the value meets the
criteria, the record is processed. If not, the record is bypassed. In other words, the RIF keyword selects
only records that meet the exception criteria.
For example, if you enter
DISPLAY RCPU RIF(PCTCPTCB > 10%)
only RCPU records in which the PCTCPTCB value (percent CPU used by TCBs) is greater than 10 are
displayed.
This selection can be done at the level of the panel or the line. RCPU is an example of panel selection. If
a field fails a RIF test, the entire panel is bypassed. However, other resource panels, such as RDAS, are
organized differently, and look more like tables. For these tabular resource types, the RIF test is applied to
each line, and a line is included on the display only if it passes the RIF test. The panel is bypassed only if
none of the lines on it pass the RIF test.
You can use RIF selection criteria for each of the resource types. RIF is supported on the DISPLAY,
EXTRACT, OBTAIN, and SET commands.
The REPORTIF keyword is also used in exception reporting for workloads, as described in “Workload
exception filters” on page 143. Unlike workload exception reporting, however, resource exception filtering
does not use the SELECTIF (SIF) keyword.
RIF command syntax
The RIF keyword syntax for resources is shown in Figure 104.
REPORTIF (keyword1(pnnq),keyword2(pnnq) ...)
Figure 104. REPORTIF keyword syntax
Where:
keyword
Defines the exception filter, such as device percent busy (DPCTBUSY) or I/O rates
(DIOPFSEC).
pnnq
Sets the quantity at which the exception is reached.
p
the comparator to be used
nn
the value to be matched or used as a threshold
q
units
As indicated in Figure 104, REPORTIF supports multiple expressions within the parentheses. These
expressions are separated by any of the standard EPILOG delimiters (commas, blanks, and so on), and
are logically ANDed together during processing. Multiple RIF expressions on a command are also allowed,
and are logically ANDed as well.
Chapter 9. Advanced reporting functions
155
Exception keywords
The exception keyword defines the field to act as the exception criteria or threshold. In general, you can
select using any data value appearing on the display panel, but not on data values that are selected with
other keywords, such as start date, start time, system ID, and so on.
Valid resource exception keywords are noted in Chapter 13, “Report element tables,” on page 233. You
can also determine if a given keyword supports resource exception filtering by consulting Appendix A,
“Data dictionary,” on page 301.
Comparators
You can use both character and sign comparators. Valid character comparators are shown below.
GT
GE
EQ
NE
LE
LT
Valid comparator signs are shown below.
GT
>
GE
>= => !< <! ^< <^
EQ
= ==
NE
!= =! <> >< ^= =^
LE
<= =< !> >! ^> >^
LT
<
The following rules apply to comparators.
v One or more blank spaces are required to separate a character comparator from the keyword and value
in the expression. Sign comparators do not require blank spaces to separate them from the keyword
and value.
v For two-character sign comparators, order is insignificant. (For example, >= is equivalent to =>.)
v Two NOT symbols are supported (^ and !), as well as a special NOT EQUAL symbol ( <> ). In addition,
a NOT symbol may be specified with a GREATER THAN or LESS THAN symbol ( !< or ^< ).
v For both character and sign comparators, parentheses are optional. If used, however, the parentheses
must be balanced.
Values
There are three types of values: numeric, character strings, and hex strings.
Numeric Values
Numeric values may be specified as decimal or integer numbers. The
value in the record is generally of greater precision than the value
displayed on the screen. It is important to be aware of potential differences
in precision when interpreting RIF criteria.
For example, suppose that the value of the field % Dev Busy (Percent
Device Busy) on the RDAS report is displayed as 7.2, but the internally
stored value is the floating point number 7.2432.... You decide to use
the % Dev Busy field to create a RIF filter, using the expression RIF
(DPCTBUSY>7.2). (The term DPCTBUSY is the data element name for the
indicated field.) Although the displayed value seems to be excluded by the
RIF filter, the record will be selected, because the internal value meets the
RIF condition.
Character Strings
Character strings may be specified in their entirety for exact matches, or
may use masks for pattern matching.
An asterisk in the middle of a pattern fixes the length of the matching
string, except if there is an asterisk at the end of the pattern.
156
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Examples
Hex Strings
OMON*
Matches any character string beginning with OMON.
OMON*A
Matches any 6-character string beginning with OMON and
ending with A.
Hex strings are used for device addresses or device numbers, and can be
treated as hex numbers or character strings. They are evaluated
numerically if a greater-than or less-than comparator is used, and are
treated as character strings if a equal-to or not-equal comparator is
specified.
For example, if you specify
DIS RDAS RIF(DEVADD >= 0000, DEVADD <= 000F)
the value is treated as a hex number, and only those devices between 000
and 00F are included on the RDAS display.
The following command produces the same effect:
DIS RDAS RIF(DEVADD = 00*)
In this case, however, the device address is treated as a character string.
Hex strings are different than ordinary character strings in that the length
is always an even number of characters. MVS/370 users with 3-character
device addresses will have to prefix the device address with a leading
zero when specifying it as a character string.
For example, suppose you enter the following command:
DIS RDAS RIF(DEVADD = 1*)
If you are working in an MVS/XA or MVS/ESA environment, all devices in
the range 1000 to 1FFF are selected.
However, if you are working in an MVS/370 environment, no devices are
selected. To specify devices 100 to 1FF, you would have to enter
DIS RDAS RIF(DEVADD = 01*)
If you enter
DIS RDAS RIF(DEVADD >= 100, DEVADD < 200)
all devices in the range 100 to 1FF are selected, since the hex string is
treated as a hex number. (No leading zeros are required to process the
hex string as a number.)
To keep the distinction between hex numbers and character strings,
masking of hex strings is allowed only with the EQ and NE. If an asterisk
is included in the value of any other comparator, it is flagged as invalid.
Units
Units are optional; if no unit is specified, the value is assumed to be in the unit shown on the display panel
and documented in the manual.
If a unit is specified, it must be compatible with the units of the data element that is selected. For example,
if you select the data element PCTCPTCB, you can specify the value as n% or n, with no difference in
interpreted meaning.
Chapter 9. Advanced reporting functions
157
For time and storage units, you can specify a compatible unit other than that on the display panel, and it
will be converted to the correct unit for the test. For example, if the value specified is ordinarily shown in
milliseconds, and you specify 1S, the value is automatically scaled to milliseconds for comparison with the
value in the record.
Valid time units are:
us
microseconds
ms
milliseconds
s
seconds
m
minutes
h
hours
d
days
For storage units, you can also specify any valid unit of space, and it will be converted to the standard unit
for the keyword before the RIF test. Valid space units are:
b
bytes
Kb
kilobytes
Examples
To select all RCPU records in which the %TCB exceeds 30, enter
DIS RCPU RIF(PCTCPTCB > 30)
To select all devices with volsers beginning with OMON, enter
DIS RDAS RIF(VOLSER EQ OMON*)
To select all devices with a DCUBUSY greater than or equal to 1000, enter
DIS RDAS RIF(DCUBUSY GE 1S)
The DCUBUSY value is collected and stored in milliseconds; the value specified was automatically
converted from seconds to milliseconds for the comparison.
How exception criteria are applied
There are two types of EPILOG resource displays: tabular displays and panel-oriented displays. Tabular
displays contain columns of data, and the same information is repeated for a list of entries on the panel.
The RDAS display is an example of a tabular display.
158
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+========================== DASD Device Activity ==============================+
| From:
10:00 to 11:00 on 10/18/99
Elap = 1:00 H
SYSA |
| Combination of 4 Intervals
|
+------------------------------------------------------------------------------+
|Volume SMF Dev
I/O
-------- Time in Milliseconds -------- % Dev
Avg|
|Serial id
# LCU Rate Total = IOSQ+Pend+Conn+Disc CUB DB Util Allcs|
|------ ---- --- --- ---- -------- ---- ---- ---- ---- ---- ----- ------|
|CHKPT1 SYSA 150 00C
1.1
69
0
48
7 14
47.8 17.2
1.0|
|CHKPT2 SYSA 300 00F
.7
34
0
0 10 24
2.5
1.0|
|COM001 SYSA D89 055
6.8
16
4
3
6
3
.2 2.0
5.7 234.2|
|COM002 SYSA D8A 055
1.3
19
1
3
6
9
.3 2.6
3.7 368.1|
|COM003 SYSA D8B 055
2.6
12
2
2
3
5
.3 1.0
1.9 103.1|
|DB2002 SYSA 14D 00C 12.1
35
12
0
6 17
29.6 144.6|
|EPIL01 SYSA 321 00F
2.5
15
0
5
3
7
4.8
2.3
1.7|
|HSM001 SYSA 307 00F
.3
23
0
0 12 11
.8
1.2|
|HSM002 SYSA 322 00F
1.6
27
3
5
5 14
4.3 5.4
8.1|
|HSM003 SYSA 9C2 046
.2
29
0
0 15 14
.6
.1|
|IMS100 SYSA 14B 00C
6.3
26
1
1
6 18
1.0 15.2 27.4|
|IMS210 SYSA 161 00C
27
0
1
1 25
35.0|
|MIS001 SYSA 313 00F
.3
18
0
1
4 13
1.3
.6
8.3|
|MIS002 SYSA 312 00F
7.1
28
5
0
8 15
16.7 71.8|
|MIS003 SYSA 306 00F
.2
6
0
0
4
2
.1 22.6|
|MP310A SYSA D8C 055
7.0
21
5
1 11
4
.3 .6 10.0 351.5|
|MTLIB1 SYSA D80 055
17
0
1
3 13
.6
|
|MTLIB2 SYSA 15B 00C
.4
15
0
1
3 11
.7
.8 19.0|
|MTLIB3 SYSA 14C 00C
1.6
17
0
1
6 10
1.1 2.6 25.0|
+==============================================================================+
Figure 105. DASD resource display (MVS/ESA)
Panel-oriented displays are generally non-columnar, and each type of information is given only once. RINF
is an example of a panel-oriented display.
+======================= General System Information ==========================+
| From:
9:00 to 9:15 on 03/06/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
| MVS SP3.1.0e Model=3090
Serial=203237
Mode=Partitioned
|
| IPS=AA
ICS=AA
OPT=XA
RMF 4.1.1
|
| Partition Name=MVSA
Partition Num=1 Physical Processors=2
|
| Processor Weight=850
Processor Complex Utilization=193.4
|
+=============================================================================+
Figure 106. Typical system resource display (with logical partitioning)
The RIF test is applied differently to tabular displays and panel-oriented displays.
v For tabular displays, each line of the display is subjected to the RIF test. If the keyword value from the
line fails the test, it is omitted from the display. If it passes, it is included.
If all panels are suppressed, the EB620 message is issued.
v For panel-oriented displays, if the keyword values from the panel fail the RIF test, the entire panel is
suppressed. (This is a standard way to handle RIF processing.) If all panels are suppressed, the EB620
message is issued.
For example, if you enter
DIS RDAS YDAY STIME(9) ETIME(0915) RIF(DPCTBUSY > 30%)
only devices that were more than 30% busy appear in the output display. If all lines of an RDAS display
fail the test, the entire panel is suppressed.
The following is a list of resource displays that have the RIF test applied to the entire display panel.
v RCPU
v RINF
v RPAG
v RSWA
Chapter 9. Advanced reporting functions
159
If the RIF test fails, the panel is entirely suppressed.
The following is a list of resource displays that have the RIF test applied to each line of the display.
v RCCH
v RCHN
v RDAS
v RDOM
v RLCU
v RPDS
v RPGN
v RSDS
v RSRM
v RSWR
v RVLF
If the RIF test fails, only the lines that failed the test are suppressed.
RCPU and RLCU processing
There are two panels that do not fall neatly into the category of a tabular or panel-oriented display: RCPU
and RLCU. This is because these panels contain both tabular and panel-level data.
A sample RCPU panel follows.
+============================== CPU Activity =================================+
| From
13:45 to 14:00 on 07/19/99
Elap =15:00 M
SYSX |
+-----------------------------------------------------------------------------+
|
Interrupt
|
|
|
|
Min Max
Avg Ready Users Logical CPU Utilization
Rate %TPI |
|
--- --- ---- ----------- ------------------------- --------- ---- |
| Batch 18
23 20.3 In =
1.2 SRB= 10.2% CPU1 = 48.7%
144.1 3.95 |
| STC
54
59 56.6 Out =
.0 TCB= 63.1% CPU2 = 47.8%
279.9 4.86 |
| TSO
8
8
8.0
MVS= 23.3%
|
|
|
|
Average Logical Processor Utilization = 48.3%
|
|
Physical CPU Utilization For This Partition = 48.3%
|
+=============================================================================+
Figure 107. Sample CPU resource display (with logical partitioning)
Notice that most of this panel fits the definition of being panel-oriented. However, the right side of the
display contains a list of processors, showing the percent utilization, interrupts, and TPI instructions for
each. This is a tabular kind of organization, and the data cannot be handled adequately by simple
panel-level selection.
In this case, you can specify CPU utilization, interrupts, or TPI instructions on the RIF command, and
display the panel if any of the processors satisfies the condition.
A sample RLCU panel follows.
160
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+===================== Logical Control Unit Activity =========================+
| From:
9:00 to 9:15 on 03/06/99
Elap =15:00 M
SYSA |
+-----------------------------------------------------------------------------+
|
I/O
I/Os
Avg #
|
|
Processor
/sec
I/O Q
|
|
-----------------|
|
00
199.80
.08
|
|
01
194.84
.05
|
|
|
|
I/Os
% All
CHPID
|
|
/sec
Avg # Ch Path
% CU
Ints
|
| LCU Deferred I/O Q
Busy
Busy CU CHPID /sec
Comment
|
| --- -------- ----- ---------- --- ----- ------ -------------------- |
| 00C
.000
.00
.00
.00 011
01
.31
|
|
.00
02
.31
|
|
.00 071
07
.32
|
|
.00
08
.30
|
| 00F
.000
.00
.00
.28 031
03
1.19
|
|
.60
04
1.28
|
|
.35 091
09
1.26
|
|
.28
0A
1.19
|
+=============================================================================+
Figure 108. Sample I/O queuing resource display (3090 processor)
There are two lists in this panel: a list of I/O processors at the top and a list of LCUs at the bottom.
You can specify the I/Os/sec and the Avg # I/O Q from the top, thereby selecting the panel if any of the
I/O processors satisfies the condition.
However, suppose that you have specified one RIF condition for the top part of the panel, and another for
a field on the bottom part. If the RIF condition pertaining to the upper part of the display panel is met, the
panel is displayed, even if no lines meet the criteria. If the top part of the panel fails a RIF test, the entire
panel is suppressed.
SET processing
RIF expressions can be entered on the SET command. When a subsequent DISPLAY, EXTRACT, or
OBTAIN command is processed, the RIF criteria previously entered with SET are applied to the command.
If the DISPLAY, EXTRACT, or OBTAIN command contains another RIF expression, it is logically ANDed to
the RIF criteria previously entered with SET. If another SET command is entered, the criteria entered are
logically ANDed with the criteria previously SET. SET criteria are removed with the SET CLEAR command.
Most exception keywords are unique to a given resource type, but a few, such as VOLSER, DEVADD, and
others appear in more than one resource panel. This introduces an ambiguity with the SET command. For
example, consider the following command:
SET RIF(VOLSER EQ OMON*)
VOLSER is a valid field on three different resource types: RDAS, RPDS, and RSDS. In cases such as
this, the exception filter will act on all resource types for which it is defined.
Automatic analysis
Fast path navigation, which is described in theEPILOG for MVS Basic User’s Guide, offers a way to
explore your performance environment under manual control. EPILOG provides another facility known as
Automatic Analysis which can automate much of this process. This is done by combining the EPILOG
exception filtering capabilities with a user-defined table called an Automatic Analysis matrix, or A-matrix.
The Automatic Analysis matrix
The Automatic Analysis matrix is a user-definable table which tells EPILOG what resource panels you
consider relevant to each type of bottleneck. For example, if the main cause of poor response time was
Chapter 9. Advanced reporting functions
161
waiting for private paging, you might want to see the Paging and Storage (RPAG) and Page Dataset
Activity (RPDS) panels; you probably would consider most of the other panels irrelevant.
The user assigns each A-matrix a 1- to 7-character name; EPILOG prefixes the name with “@” and keeps
it as a member of the rhilev.REDPARM partitioned data set.
When you invoke a resource display from a degradation display during system navigation (by placing an R
in the first character of a line), you are actually invoking a default A-matrix contained in this data set. The
matrix is contained in the @DEFAULT member of the rhilev.REDPARM partitioned data set.
Each row of the matrix represents a wait reason identified by its 3-character abbreviation (AIO, AUX,
COM, etc.). The names of the resource panels (RCHN, RCPU, etc.) associated with each wait reason are
shown in the order in which they will be displayed. This matrix is created, modified, and displayed with a
set of utility commands that are described in “Setting reporter options” on page 137.
A-Matrix definitions
A-matrices are standard members of the rhilev.REDPARM partitioned data set, and so may be added,
deleted, or edited as normal PDS members. EPILOG also provides a set of commands to create, delete,
list, and replace A-matrix tables from the EPILOG reporter.
When these commands are used, the syntax of the entry is checked, and any errors are flagged with an
appropriate message. If you use an external editor to create or modify these members, syntax checking
will be done when the A-matrix is invoked, either explicitly or implicitly. If an error is detected when the
A-matrix is invoked implicitly, it may be not always be obvious that the error was caused by the faulty
A-matrix. For this reason, we recommend that you use the commands described in this section rather than
an external editor to create and maintain A-matrix members.
Each of these commands accepts a 1- to 7-character member name as an operand. The member name
you enter is then prefixed with an “at” sign (@) to generate the actual member name in rhilev.REDPARM.
For example, if you enter:
CREATEM (MYMAT) . . .
The name of the member created in rhilev.REDPARM is @MYMAT.
The following sections describe each of the A-matrix member utility commands: CREATEM, DELETEM,
LISTM, and REPLACEM.
Create an A-matrix (CREATEM)
The CREATEM command (abbreviated CMAT) creates a new Automatic Analysis member. The first
operand is a 1- to 7-character A-matrix name and is required. You must also enter one or more wait
reasons, each followed by the resource panel names which are to be associated with it.
CMAT
A-matrix-ID
wait-reason(resource-panels ...) ...
For example, the following command creates a new A-matrix called SHIFT1 which will display the RINF,
RPDS, and RPAG resource panels whenever waiting for local page-in (PAG) is the most significant wait
reason:
CMAT SHIFT1 PAG(RINF RPDS RPAG)
Valid 3-character wait reason codes are defined in Table 14 on page 146. Valid resource panel names are
defined in Table 9 on page 63.
After you create the A-matrix member with CREATEM, you use the REPLACEM command to add more
lines to the matrix.
162
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Delete an A-matrix (DELETEM)
The DELETEM command (abbreviated DMAT) is used to delete an existing automatic analysis member
from the rhilev.REDPARM data set. The only required operand is a 1- to 7-character A-matrix name.
DMAT
A-matrix-ID
List an A-matrix (LISTM)
The LISTM command (abbreviated LMAT) lists an existing automatic analysis member from the
rhilev.REDPARM data set. The only operand is a 1- to 7-character A-matrix name.
The LISTM command syntax is shown below.
LMAT
A-matrix-ID
Replace an A-matrix (REPLACEM)
The REPLACEM command (abbreviated RMAT) is used to add, modify, or delete wait reasons from a
specified A-matrix. The syntax of the command requires a 1- to 7-character A-matrix name and a valid wait
reason code. If a resource panel name is included with the wait reason code, the wait reason is added to
the matrix. (If the wait reason is already defined in the matrix, the definition is replaced.) If the wait reason
code is entered without a resource panel name, it is deleted from the matrix.
The REPLACEM command syntax is shown below.
RMAT
A-matrix-ID
wait-reason(resource-panels ...)
Notice that you can specify multiple resource panels for a given wait reason.
For example, the following command associates the resource panel RCPU with the CPW (waiting for
CPU) wait reason in A-matrix MYMAT. (If there was already a definition for the CPW wait reason in
MYMAT, it is replaced.)
RMAT MYMAT CPW(RCPU)
Any existing entries within MYMAT will be left unchanged (except for CPW, which will be completely
replaced by the new specification). The 3-character IDs which represent all the possible wait reasons can
be found in Table 14 on page 146. The resource panel names can be found in Table 9 on page 63.
You can only modify or add resource panels for one wait reason at a time. To change more than one wait
reason, enter multiple REPLACEM commands.
Also, you cannot use REPLACEM to modify the only wait reason in a matrix. To get around this, you can
either add a temporary wait reason to the matrix and then delete it when you are finished, or delete and
create a new matrix with the DELETEM and CREATEM commands.
AUTOMATIC resource displays
You can use these A-matrices in various ways. The System Navigator uses the default when an R is
entered to decide which resource panels should be displayed for the selected bottleneck. (This feature is
described in theEPILOG for MVS Basic User’s Guide.) EPILOG has another feature called Automatic
Resource Reporting which automatically generates the resource panels associated with the most
significant wait reason.
For example, suppose we wanted to look for all intervals today where the response time for TSO was
greater than 1.5 seconds. If weadd the AUTOMATIC keyword (or AUTO for short) to the request, EPILOG
automatically finds the most significant wait reason (that is, the one with the highest percentage) and looks
it up in the DEFAULT Automatic Analysis matrix. Based on what it finds there, it automatically generates
certain resource panels for the corresponding interval.
Chapter 9. Advanced reporting functions
163
In the following example, the most significant wait reason was Private Page-In Wait. The first two
resource panels for this wait reason, as specified in the DEFAULT A-matrix, are Paging and Storage
(RPAG) and Page Dataset Activity (RPDS). These automatically appear when you enter AUTO as part of
the command:
DIS PGN(2) TODAY RIF( RESP(>1.5S)) AUTO
The resulting displays might look like the following figures:
+=============================================================================+
| Performance Group = 2
Symbolic Name = TSO
|
| From 11:30 to 11:45 on 09/07/99
Elap = 15:00 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.06 S 4.0|-> .
.
.
.
.
.
.
.
.
.|
|Private Page-In Wait
0.26 S 16.4|------> .
.
.
.
.
.
.
.
.|
|Terminal Output Wait
0.18 S 11.3|---->
.
.
.
.
.
.
.
.
.|
|Waiting for CPU
0.15 S 9.9|--->.
.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.12 S 7.7|--->.
.
.
.
.
.
.
.
.
.|
|Disk DEV023 36D Act
0.11 S 7.4|--> .
.
.
.
.
.
.
.
.
.|
|Disk DEV022 151 Que
0.08 S 5.1|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
1.60 S
705 MVS Transactions Ended
|
|Productivity Index
15%
|
+=============================================================================+
+=========================== Paging and Storage ==============================+
| Period: 11:30 to 11:45 on 09/07/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
|
|Virtual |
Average Frame Counts
|
Page/sec
| Swap paging/sec |
|
| Memory | Real = Fixed + Non-Fix| In
| Out | In
Out |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| SQA |
512K |
173 |
173 |
|
|
|
|
|
| LPA | 4380K |
405 |
20 |
385 |
3.2 |
|
|
|
| CSA | 3004K |
172 |
61 |
111 |
.8 |
.5 |
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| LSQA|
|
223 |
223 |
|
|
|
|
|
| Pvt | 7552K | 2407 |
177 | 2230 |
8.3 | 5.9 | 41.0 | 40.9 |
| Free|
|
235 |
|
235 |
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
| Nuc |
896K |
224 |
224 |
|
|
|
|
|
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|Total| 16384K | 3839 |
878 | 2961 | 12.4 | 6.5 | 41.0 | 40.9 |
+-----+--------+--------+--------+--------+--------+--------+--------+--------+
|
|
|
Total Pages / second
|
18.9
|
81.9
|
+=============================================================================+
Figure 109. Automatic analysis example (part 1 of 2)
+========================== Page Dataset Activity ============================+
| Period: 11:30 to 11:45 on 09/07/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
| Space
Dev
Dev
%
% In
I/O's
Pages
Trnsfr |
| Type
Adr
VOLSER
Type
Full
Use
per sec
per IO
Time |
| -------- ---------------------------- ------ |
| PLPA
150
MVSA21
3380
57.3
13.0
2.7
1.2
.040 |
| Common
140 SYS021
3380
27.5
7.3
.8
1.6
.054 |
| Local
143
TIME21
3380
35.7
27.6
2.6
9.2
.011 |
| Local
153
TIME22
3380
35.3
28.0
2.6
9.2
.012 |
| Local
141
TSO021
3380
32.9
24.2
2.4
10.1
.010 |
| Local
151
TSO022
3380
33.1
28.4
2.6
9.1
.012 |
+=============================================================================+
Figure 110. Automatic analysis example (part 2 of 2)
The other resource panels associated with private page-in waits (RCHN, RCPU, and RINF) would be
displayed after the RPAG and RPDS.
If, when using AUTO, you specify other exception criteria in addition to response time, EPILOG displays
the resource panel(s) for those other criteria, even if they do not reflect the most significant bottleneck. To
164
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
illustrate, in the example below, we have asked only for intervals with poor response where waiting for
CPU (CPW) makes up more than 10 percent of an average transaction. Even though CPW is not the most
significant wait reason (disk active and disk queued for device 36D are higher), AUTO produces a
resource panel for CPW (RCPU in this case).
DIS PGN(2) TODAY RIF( RESP(>1.5S) CPW(>10%) ) AUTO
The resulting display might look like this:
+=============================================================================+
| Performance Group = 2
Symbolic Name = TSO
|
| From 12:30 to 12:45 on 09/07/99
Elap = 15:00 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.09 S 5.6|--> .
.
.
.
.
.
.
.
.
.|
|Disk DEV023 36D Act
0.22 S 13.2|-----> .
.
.
.
.
.
.
.
.|
|Disk DEV023 36D Que
0.21 S 12.2|---->
.
.
.
.
.
.
.
.
.|
|Waiting for CPU
0.20 S 12.0|---->
.
.
.
.
.
.
.
.
.|
|Private Page-In Wait
0.16 S 9.5|--->.
.
.
.
.
.
.
.
.
.|
|Swap Page-In Wait
0.12 S 7.2|--> .
.
.
.
.
.
.
.
.
.|
|Disk DEV022 151 Act
0.12 S 7.1|--> .
.
.
.
.
.
.
.
.
.|
|Average Trans Time
1.72 S
501 MVS Transactions Ended
|
|Productivity Index
25%
|
+=============================================================================+
+============================== CPU Activity =================================+
| Period: 12:30 to 12:45 on 09/07/99
Elap =15:00 M
A083 |
+-----------------------------------------------------------------------------+
|
Min
Max
Avg
Ready Users
CPU Utilization
|
|
----------------------------------------------|
| Batch
0
1
.4
In =
1.3
SRB =
3.6 %
CPU2 = 43.8 %
|
| STC
27
28
27.5
Out =
.1
TCB = 28.4 %
|
| TSO
12
19
15.1
MVS = 11.8 %
|
+=============================================================================+
Note that the AUTO keyword is only valid when you request a DETAILed degradation display (that is, if
you omit the SUMMARY keyword).
By default, the A-matrix called DEFAULT selects the associated resource panels. You can override this
with theAMATRIX keyword along with AUTO. In the example below, EPILOG uses the A-matrix called
MYMAT instead of DEFAULT to make its decisions:
DISPLAY PGN(2) TODAY AUTO AMATRIX(MYMAT)
The RESOURCE command
If a DETAILed degradation display is currently on the screen, you can enter the command RESOURCE to
automatically display the resource panels associated with the worst bottleneck. These resource panels can
be selected in three ways:
1. If you enter the RESOURCE command without an operand, it uses the DEFAULT A-matrix to
determine which resources to display.
2. If you enter the name of another A-matrix with the command, it uses that A-matrix to determine the
resource panels.
3. If you enter one or more resource keywords (RPGN, RCHN, etc.), the specified resource panels are
displayed.
For example, suppose that you have navigated to a detail workload display, such as the one shown in
Figure 111 on page 166.
Chapter 9. Advanced reporting functions
165
+=============================================================================+
| Performance Group = 2
Symbolic Name = TSO
|
| From 11:15 To 11:30 On 09/06/99
Elap = 15:00 M SYSA |
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
0.23 S 4.7|-> .
.
.
.
.
.
.
.
.
.|
RPrivate Page-In Wait
1.75 S 35.1|--------------> .
.
.
.
.
.
.|
|Swap Page-In Wait
0.38 S 7.7|--->.
.
.
.
.
.
.
.
.
.|
|
|
|Average Trans Time
4.98 S
865 MVS Transactions Ended
|
|Productivity Index
10%
|
+=============================================================================+
Figure 111. SINGLE interval DETAIL display
In such a case, entering RESOURCE while the display is on the screen is equivalent to using the R
navigational command to select resource panels for the Private Page-In Wait wait reason.
RESOURCE has an advantage over R in that you can supply an AMATRIX keyword to override the
current default automatic analysis matrix, or specify a resource panel directly:
RESOURCE AMATRIX(A-matrix-ID)
or
RESOURCE RDAS
RESOURCE has the disadvantage that you cannot specify which bottleneck you want to analyze; it always
provides resource data for the wait reason with the highest percentage. Of course, you can always use the
DISPLAY command to choose exactly which resource panels you want to look at.
166
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 10. Workload Profiling Facility (WPF)
This chapter describes the EPILOG for MVS Workload Profiling Facility (WPF). It begins with an overview
of the facility and the profiling process. The chapter then provides a complete and detailed survey of WPF
commands. With a full range of examples, it illustrates the ways you can use WPF to your best advantage.
You will be guided step-by-step through the profiling process. We’ll examine how to decide which
workloads’ performance should be tracked— which workloads are critical in your system. Then we’ll
discuss how to determine which data should be included in a workload’s profile— which records are
representative of performance and which are not. After generating some profiles, we’ll look at profile
displays that show the performance of a workload over a period or that compare a particular workload with
a specific profile.
This chapter also briefly discusses the initialization and maintenance of the Profile datastore (PRDS).
Complete installation requirements and maintenance procedures are described in theEPILOG for MVS
Program Directory.
WPF overview
WPF is the profiling component of EPILOG. It selects and averages degradation data for any workload
over a designated time period so that you can track its performance and spot any trends towards poor
performance.
The WPF and its auxiliary components are described below.
Workload Profiling Facility
A facility for creating profiles. Using a batch job, it selects and averages
data from the EPILOG datastore according to user specifications. EPILOG
datastore records can be selected for any of the valid workload types:
v Performance groups
v Batch jobs
v Started tasks
v TSO user sessions
The Profile datastore
A VSAM data set that can be defined when installing EPILOG or anytime
thereafter.
The EPILOG reporter
The same report facility used for standard EPILOG reports can also be
used to access the Profile datastore and generate two basic types of
reports:
v Profile Reports
v Profile/Workload Comparison Reports
These reports can be displayed online or generated in batch.
Using WPF
This section explains how to use the Workload Profiling Facility. It demonstrates how you can use it to
profile the performance of selected workloads and then generate profile reports either online or in batch.
Figure 112 on page 168 provides an overview of how to use WPF.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
167
┌────────────────────────────┐
│
│
│
Accumulate
│
│
Historical Degradation
│
│
Data
│
│
│
└─────────────┬──────────────┘
│
│
┌────────────────────────────┐
│
│
│
Create Profiles
│
│
for
│
│
Selected Workloads
│
│
│
└─────────────┬──────────────┘
│
│
┌────────────────────────────┐
│
│
│
Compare Workload's
│
│
Current Degradation
│
│
with Profile
│
│
│
└────────────────────────────┘
Figure 112. Overview of Using the Workload Profiling Facility
Because profiles are derived from EPILOG datastore records, the profiling process actually begins when
you start collecting degradation data with the EPILOG collector. After accumulating sufficient data to
generate meaningful samples, the Workload Profiling Facility’s PROFILE command allows you to create
profiles. Each profile contains the averaged degradation data for a particular workload over a period of
time. With the EPILOG reporter’s DISPLAY command, you can use this data to generate Profile Reports
that represent a workload’s typical performance. Finally, with the reporter’s COMPARE and SETP
commands, you can use the reporter to generate Profile/Workload Comparison Reports that measure a
workload’s current degradation data against its profile.
These four commands—PROFILE, DISPLAY, COMPARE, and SETP—are all you need to use the EPILOG
Workload Profiling Facility.
Before you begin
Before you use WPF, you must have defined a VSAM cluster for the Profile datastore (PRDS). This cluster
is defined and initialized with the EPILOG datastore during product installation. Complete instructions for
initializing and maintaining this datastore are provided in the IBM Tivoli OMEGAMON XE on z/OS:
Planning and Configuration Guide.
Creating profiles
Once your Profile datastore has been defined, you can create profiles for any performance group, batch
job, started task, or TSO user(s). Although you can use WPF to profile any workload on your system, you’ll
find it especially useful for tracking the performance of critical workloads.
What qualifies as a critical workload?
v Any batch job that for one reason or another must run within certain time constraints.
v Any performance group with response time service-level objectives.
v Any started task or group of TSO users that must run within defined performance parameters.
168
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
An example might be a nightly stream of batch jobs that must be completed before CICS can be brought
up in the morning, or a performance group consisting of of TSO users that require response times below a
certain limit. These are the types of workloads that you’ll probably want to profile.
For a complete list of workload types and the keywords necessary to define them to WPF, see the
Workload Keywords table in theEPILOG for MVS Command Summary booklet.
A sample batch job for creating profiles
Use the PROFILE command to create profiles based on data in the EPILOG datastore. You should create
profile records in batch mode. To do so, submit the job supplied in rhilev.REDDATA(KEPPROFJ) or
rhilev.REDDATA(KEDPROFJ) if DELTAMON for MVS is installed.
The sample batch job in the KEPPROFJ member invokes the profile routine, EPIPROF, which selects the
required data from the EPILOG datastore, creates profiles, and writes them out to the Profile datastore.
Before submitting the job, you must modify the JCL to meet your installation’s requirements. Directions for
doing so are provided in the job’s coded comments. In addition, after familiarizing yourself with the
PROFILE command described below, you may also want to modify the job’s PROFILE commands to suit
your own needs.
The PROFILE command: creating profiles
The PROFILE command (PROF) creates a high-level profile by averaging degradation data for a specified
workload. See the EPILOG for MVS Customization Guide for information on deleting workload profiles.
The PROFILE command and its keywords are illustrated below:
PROFILE {workload} {PROFNAME(cc...cc)} [time period] [SYSID(cccc,...)] [OUTLIER(n) | NOOUTLIER] [SELECTIF(filter)] [PLOT | NOPLOT] [TEST]
Figure 113. PROFILE Command Format
As you can see in Figure 113, the only required keywords are a workload identifier and a profile name.
Descriptions of all keywords and keyword types follow.
workload
Identifies the batch job, performance group, started task, or TSO user
session for which you are creating a profile. In addition to these specific
types of workload keywords, you can profile degradation data for all
workloads on the system by using the SYSTEM keyword as the workload.
For a complete list of workload keywords, see “Workload keywords” on
page 171. You can only use one of these keywords at a time with the
PROFILE command. All of these keywords are valid on the PROFILE
command except STEP and INTERVAL.
Depending on your workload specification, a single PROFILE command
can be used to create more than one profile. For example, if you specify
two or more workload values of the same type, WPF generates a profile
for each workload specified; if you specify an asterisk (*) for a given
workload, WPF generates a profile for each workload that qualifies.
PROFNAME(cc...cc)
Defines a 1- to 16-character profile name that you assign to the workload.
If your profile name (PNAME for short) contains any blanks or special
characters, you must enclose it within quotation marks. If you enter
Chapter 10. Workload Profiling Facility (WPF)
169
PNAME without an operand, WPF assigns the profile the PNAME default
(i.e., automatic). This default is used when creating profiles for multiple
workloads of one type, such as JOB(*).
Keep in mind that the profile is identified by a unique combination of
profile name, workload, time period, and SYSID. Thus, you should select a
name that is meaningful and easy to remember.
time period
Specifies the date and time interval for which you want WPF to select
data. The date/time keywords are summarized in “Date and time
keywords” on page 171.
SYSID(cccc,...)
Specifies an SMF system ID for the workload being profiled. Required only
when the records in your EPILOG datastore have different SYSIDs and
you want to create a profile for a workload running under a specific
SYSID. If you specify multiple SYSIDs, for example (SYSA,SYSB), you
will generate individual profiles for each SYSID. If you do not specify a
SYSID, all records that satisfy your other selection criteria will be included
in the profile regardless of SYSID.
OUTLIER(n)
Excludes degradation records from the profile that are not representative
of the sample. For example, records for jobs that went into a loop or that
were cancelled prematurely. Such records generally lie outside the
representative elapsed- or response-time range (that is, the range of times
that are most typical of the sample). These records are called outliers.
Your OUTLIER (or OUTL(n)) value (0, 1, 2, or 3) determines which outliers
are excluded. It specifies a number of QSTEPS above and below the
representative time range. (A QSTEP is the difference between the lowest
and highest time value in this representative range.) Any outlier that lies
beyond this range of QSTEPS above and/or below the representative
range is excluded.
For example, if your representative range has a low record of 6 minutes
and a high record of 8 minutes, the QSTEP is 2 minutes. Thus, if you
were to specify OUTLIER(1), any outlier that had an elapsed time less
than 4 minutes or greater than 10 minutes would be excluded.
If you do not specify an OUTLIER value, the default is NOOUTLIER and
no data records will be excluded.(See “Selecting meaningful profiles” on
page 173 for a fuller treatment of OUTLIERS.)
SELECTIF(filter)
Specifies your criteria for filtering records from the EPILOG datastore. The
SELECTIF keyword must be followed by a valid wait reason and
appropriate selection criteria. Use any of the exception threshold keywords
described in “Workload exception filters” on page 143.
PLOT
Generates a graph with the workload’s elapsed or response times plotted
on a horizontal axis and their frequency on a vertical axis. Outlier ranges
are also indicated on the horizontal axis to assist you in selecting an
appropriate OUTLIER value. PLOT is the default. To overide it, specify
NOPLOT. See “Using the PLOT keyword” on page 175 for a sample
graph.
TEST
Instructs WPF not to write the profile it has generated to the Profile
datastore. You can thus create and display profiles while testing for the
most appropriate OUTLIER value without wasting datastore space.
We recommend that you specify the PLOT keyword, or leave it as the
default, when using TEST so that you can see the different ranges of
outliers generated by your test profile.
170
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Workload keywords
The following keywords are used to specify a performance group, batch job, started task, or TSO user.
Table 18. Workload keywords
Keyword
Short
Description
Operand
ACCOUNT
ACCT
Selects batch job or TSO session degradation data by installation
account code.
1-12 Char (See Note)
ACCOUNT is not compatible with the STARTTSK keyword.
CLASS
CLS
Selects batch job degradation data by JES job class.
1 Char (See Note)
CLASS is not compatible with the STARTTSK or TSOUSER
keywords.
JOBNAME
JOB
Selects batch job degradation data by jobname. The jobname can
be further qualified by JES job number; any number of JES
numbers may be entered within parentheses following the job
number.
1 - 8 Char (See Note)
Example: JOB(PAYROLL(J721))
PERFGROUP
PGN
Selects degradation by performance group number.
0-9999
PGPERIOD
PGP
Selects one or more performance periods. This keyword must be one or more
used in conjunction with the PGN keyword, and is not allowed with performance periods
the SYM keyword.
PGPERIOD is not valid with the SET or SETP commands.
PROGRAM
PGM
Selects batch job, started task, or TSO session degradation data
by program name.
1 - 8 Char (See Note)
STARTTSK
STC
Selects started task degradation data by task name.
1 - 8 Char (See Note)
SYMBOLIC
SYM
Selects degradation for a performance group by user-defined
symbolic name.
1 - 8 Char (See Note)
SYSTEM
SYS
DISPLAYs system-wide degradation.
TSOUSER
TSO
Selects degradation data for TSO users.
1 - 7 Char (See Note)
Note: Generic formats are valid for this keyword parameter. (That is, asterisks (*) may be used as a
wildcard character.)
Date and time keywords
These keywords control data selection by date and time range.
Table 19. Date and time Keywords
Keyword
Short
Description
Operand
BAND
BAND
Displays data between the start and end times
on the specified days. (default value)
Chapter 10. Workload Profiling Facility (WPF)
171
Table 19. Date and time Keywords (continued)
Keyword
Short
Description
Operand
DAYOFWK
DAY
Displays data only for the specified days. The
operands used with this keyword can be
abbreviated to an unambiguous short form. For
example, WEDNESDAY can be abbreviated to
WED or just W, but SATURDAY can only be
shortened to SA. (This keyword may not be
used when deleting records from the Profile
datastore with the EXCLUDE command.)
ALL
WEEKDAY or WKDAY
WEEKEND or WKEND
MONDAY
TUESDAY
WEDNESDAY
THURSDAY
FRIDAY
SATURDAY
SUNDAY
ENDDATE
EDATE
Specifies the end date of the display. If no
ENDDATE is specified, data is selected up to
the most recent record in the database.
Date in any of the formats
shown following the table.
ENDTIME
ETIME
Specifies the end time of the display. Defaults
to 23:59:59 on the specified ENDDATE.
Time in any of the formats
shown following the table.
EXDATE
EXD
Excludes the specified date(s) from the display. Date(s) in any of the formats
shown following the table.
LASTMONTH
LMN
Displays data from the first day to the last day
of the previous month.
LASTWEEK
LWK
Displays data from Monday to Sunday of the
previous week.
LASTYEAR
LYR
Displays data from the first day to the last day
of the previous year.
RANGE
RANGE
Displays data from the start time and date to
end time and date.
STARTDATE
SDATE
Specifies the start date of the display. If no
STARTDATE is specified, data is displayed
from the first record in the database.
Date in any of the formats
shown following the table.
STARTTIME
STIME
Specifies the start time of the display. If no
STARTTIME is specified, data is displayed
from the first record on the specified
STARTDATE.
Time in any of the formats
shown following the table.
THISMONTH
TMN
Displays data from the first day of the current
month to the most recent record in the
database.
THISWEEK
TWK
Displays data from the first day of the current
week to the most recent record in the
database.
THISYEAR
TYR
Displays data from the first day of the current
year to the most recent record in the database.
TODAY
TDAY
Displays data from the beginning of the current
day to the most recent record in the database.
YESTERDAY
YDAY
Displays data from the beginning of the
previous day to the last record of the previous
day.
Date and time formats
Dates can be entered in either of two formats;
v Julian
172
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
yyddd
v Gregorian
(ddd = 001 through 366)
mmddyy
mm/dd/yy
mm/dd
(assume the current year)
dd
(assume the current month and year)
Note: If slashes are used with Gregorian dates, leading zeros are not required (that is, 7/2/99 is an
acceptable date entry). If slashes are not used, leading zeros are required.
Times must be entered in 24-hour format (i.e., 1300 instead of 1:00 PM), and can be abbreviated as
follows:
h
One-digit hour without a leading zero
hh
Two-digit hour
hmm or h:mm
Hours and minutes without a leading zero
hhmm or hh:mm
Hours and minutes
hhmmss or hh:mm:ss
Hours, minutes, and seconds
You can also use relative times and dates by inserting a plus or minus sign (+ or -) in front of the date or
time operand. For an explanation of relative dates and times, refer to theEPILOG for MVS Basic User’s
Guide.
Selecting meaningful profiles
Because a profile represents a workload’s average performance, the uniformity of the records used to
generate that average determines how meaningful the profile really is. A representative sample should
contain only those records that are typical of a workload’s performance. In other words, it should not
contain records of what you know to be erratic performance.
(Remember that records with extremely high elapsed or response times are almost always being degraded
by a factor that is short lived and not typical of ordinary degradation.)
For example, suppose you wanted a profile of your installation’s daily SMF analysis job for the past two
weeks. If the sample for your profile contained elapsed-time records that were all between one minute
(1:00) and one minute and thirty seconds (1:30), you would probably consider these records as
representative of the job’s performance. A profile based on this sample would thus be meaningful.
On the other hand, assume that the sample for your profile contained two records in which the SMF
analysis job went into a loop. These records might show very high elapsed times that would skew the
average upwards. Similarly, if the sample contained a record in which the job was cancelled, the short
elapsed time of that record would skew the average downward. Excluding these records, however, would
make the sample more representative and thus would yield a more meaningful profile of the SMF analysis
job.
Let’s look at exactly how including or excluding three such records affects the profile of the SMF analysis
job.
The following example shows how including three records that are not representative of the job’s elapsed
time (0:38M, 2:04M; and 7:01M) significantly increases the average. It also shows that excluding them
from the sample, yields a more meaningful average for the profile record.
Chapter 10. Workload Profiling Facility (WPF)
173
JOB ELAPSED TIME (in minutes)
MON
TUE
WED
THU
FRI
SAT
MON
TUE
WED
THU
FRI
0:38M
1:12M
1:11M
1:10M
1:14M
1:09M
1:02M
1:33M
1:32M
2:04M
7:01M
---Average: 2:09M
excluded
1:12M
1:11M
1:10M
1:14M
1:09M
1:02M
1:33M
1:32M
excluded
excluded
---Average: 1:00M
Using the OUTLIER keyword
When creating profiles with WPF, the OUTLIER keyword enables you to exclude unrepresentative records
when averaging data. Let’s see how this is done.
When you issue a PROFILE command, it selects EPILOG datastore records for the workload and time
period that you specify. It also makes sure that the records it selects meet the selection criteria that you
specify with the SIF keyword. These records are your sample, and their average yields the average
elapsed or response time for your profile.
Ensuring that this sample contains only records that are representative of the workload is the function of
OUTLIER. After the EPILOG datastore records have been selected, OUTLIER processing allows you to
filter out any records that you know are not typical of the workload. Such records are known as outliers.
More precisely, an outlier is any record that lies outside the range ofrepresentative records. For WPF, we
have determined this representative range to be one that includes at least the middle 50% of the values.
We determine this range as follows.
1. We sort in ascending order the EPILOG datastore records selected by the PROFILE command, and
assign each one a rank. We’ll illustrate this step using the elapsed time records shown above:
0:38 1:02 1:09 1:10 1:11 1:12 1:14 1:32 1:33 2:04 7:01
------------------------------------------------------1
2
3
4
5
6
7
8
9
10
11
ELAP
RANK
2. We then use a version of the following formula to calculate the positions of quartile values (Q).
Let n = the number of records selected by PROFILE.
Q1 = (n + 1) / 4
Q2 = 2 * (n + 1) / 4
Q3 = 3 * (n + 1) / 4
Since the sample we are using for illustration contains 11 records, the positions of our quartile values
are calculated as follows:
Q1 = (11 + 1) / 4
Q2 = 2 * (11 + 1) / 4
Q3 = 3 * (11 + 1) / 4
174
or rank 3
or rank 6
or rank 9
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Using the same sample, we see that rank 3 points to an elapsed time of 1:09 minutes, and thus Q1
assumes that value. In the same fashion, Q2 (our median), which points to rank 6, assumes the value
1:12 minutes, and Q3, which points to rank 9, assumes the value 1:33. We can also see that our
representative range includes the records that are most typical of the analysis job’s elapsed time.
0:38 1:02 1:09 1:10 1:11 1:12 1:14 1:32 1:33 2:04 7:01
------------------------------------------------------1
2
3
4
5
6
7
8
9
10
11
|
|
|
Q1
Q2
Q3
-----representative range-----
ELAP
RANK
3. We then calculate the QSTEP by subtracting Q1 from Q3:
0:38 1:02 1:09 1:10 1:11 1:12 1:14 1:32 1:33 2:04 7:01
------------------------------------------------------|
|
Q1(1:09)
Q3(1:33)
----- Q Step = 1:33 - 1:09-----
ELAP
QUARTILE
S
4. We then multiply the QSTEP by the OUTLIER value you specify (0, 1, 2, 3) to determine which outliers
should be excluded.
Thus, in our example, specifying OUTLIER(1) would exclude from the sample any records with an
elapsed time 24 seconds less than 1:09 minutes (Q1) and/or 24 seconds greater than 1:33
minutes(Q3).
Note: For the purpose of illustration, these numbers have been rounded off.
Using the PLOT keyword
To simplify your choice of OUTLIER value, the PLOT default generates a graph of the sample and shows
the different ranges of OUTLIER values. The graph is similar to Figure 114 on page 176.
Chapter 10. Workload Profiling Facility (WPF)
175
+------------------------------ELAPSED TIME PLOT------------------------------+
| Job = SMFANAT
Profname = SMFANAT
|
|
|
+-----------------------------------------------------------------------------+
|
|
|
|
|
F 10+
|
|
R
|
|
|
E
|
|
|
Q
|
|
|
U
|
|
|
E 5+
|
|
N
|
|
|
C
|
*
|
|
Y
|
*
|
|
|*
*
***
**
*
|
|
+---------+----------+---------+---------+---------+
|
|
35.79 S
58.75 S
1:21 M
1:44 M
2:07 M
2:30 M
|
|
OUTLIER(0) RANGE ( 1:10 M , 1:33 M ) INCLUSIVE.
|
|
OUTLIER(1) RANGE (47.32 S , 1:56 M ) INCLUSIVE.
|
|
OUTLIER(2) RANGE (37.68 S , 2:19 M ) INCLUSIVE.
|
|
OUTLIER(3) RANGE (37.68 S , 2:42 M ) INCLUSIVE.
|
|
|
+-----------------------------------------------------------------------------+
Figure 114. Sample elapsed time plot
In the above figure, the elapsed times are plotted on the horizontal axis, and their frequency on the vertical
axis. Note that the scale for elapsed times is calculated to include those values that fall within the four
levels of outliers. Therefore, because of space limitations, extreme values are sometimes not displayed.
Thus, the extremely unrepresentative elapsed time from our example (7:01) does not appear on the plot. If
you do not want a plot generated, you should specify the NOPLOT keyword.
Using the TEST keyword
Depending on the purpose of your profile, you can select the OUTLIER level that will make your profile
most meaningful. Before writing the profile record to the Profile datastore, you may want to run a few
preliminary tests using different OUTLIER values to see which one yields the profile that meets your
needs. You can run these tests simply by specifying the TEST keyword and leaving PLOT as the default,
when you issue the PROFILE command. The TEST keyword instructs WPF not to write the profile to the
Profile datastore.
Examples of using the PROFILE command
This section demonstrates how to generate various types of profile records. Note that whenever a sample
command has been written on more than one line, we have used the hyphen (-) as a continuation
character.
Generating a standard profile
PROFILE PGN(2) PNAME('TSO USERS') LASTWEEK
This command creates a profile that contains the averaged data for performance group 2 from the
previous week and assigns it the name TSO USERS. Because no selection filters are specified, WPF
selects all EPILOG datastore response time records for performance group 2. Similarly, because no
OUTLIER criteria is specified, all the selected EPILOG datastore records are included in the average.
176
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Generating a preliminary test profile
PROFILE JOB(PAYROLL) PNAME('NITE BATCH') LASTWEEK BAND STIME(1800) ETIME(0300) PLOT TEST
This command creates a profile containing the averaged degradation data for all jobs with the name
PAYROLL run last week between the hours of 6 p.m. and 3 a.m. The PNAME keyword assigns the profile
a name that clearly identifies the data it contains–NITE BATCH.
Because this profile is being created for the first time, both the PLOT and TEST keywords are specified.
The PLOT keyword (which is a PROFILE default) generates a graph with the workload’s elapsed times
plotted on a horizontal axis and their frequency on a vertical axis. (See “Using the PLOT keyword” on
page 175 for an example.) Finally, the TEST keyword suppresses the writing of this profile to the Profile
datastore since it is being used only for determining an accurate OUTLIER value.
Generating a profile using optional keywords
PROFILE TSO(USER04) PNAME('PRIME SHIFT') LASTWEEK BAND STIME(0900) ETIME(1700) SIF(ELAPSED(>30m)) OUTLIER(3)
This command generates a profile that contains the averaged degradation data for all TSO sessions with
user ID USER04 that ran last week between 9 a.m. and 5 p.m. and that had an elapsed time greater than
30 minutes. PNAME assigns the profile the name PRIME SHIFT. Before the data is averaged,
OUTLIER(3) excludes any job that falls outside the third range of outliers.
Generating multiple profiles
PROFILE PGN(2,12) PNAME(TSO) LASTWEEK BAND STIME(0900) ETIME(1700) SIF(ELAPSED(>30m)) OUTLIER(3)
This command is equivalent to two separate PROFILE commands that specify PGN(2) and PGN(12).
In other words, it creates two profiles. Each profile contains the averaged degradation data for one of the
specified performance groups for last week between 9 a.m. and 5 p.m., where the elapsed time exceeded
30 minutes. PNAME assigns the profiles the name TSO.
Before the data is averaged, OUTLIER(3) excludes any transaction record that falls outside the third range
of outliers.
Here are some other sample commands that create multiple profile records:
PROFILE JOB(DB3*) . . .
Creates separate profile records for all jobs that begin with DB3, and that meet the other criteria.
PROFILE TSO(DEV*) . . .
Chapter 10. Workload Profiling Facility (WPF)
177
Creates separate profile records for all TSO sessions in which the TSO user ID begins with DEV, and that
meet the other criteria.
PROFILE STC(CICS*) . . .
Creates separate profile records for all started tasks that begin with CICS, and that meet the other criteria.
Generating profiles for workloads having multiple SYSIDs
PROFILE PGN(2) PNAME('TSO DEV') SYSID(SYSA,SYSB) THISMONTH OUTLIER(3)
Because the SYSID keyword specifies two SMF system IDs (SYSA,SYSB), individual profiles are
generated for performance group 2 under SYSA and performance group 2 under SYSB. PNAME assigns
both profiles the profile name TSO DEV. Each profile contains the workload’s averaged degradation data
for the current month. OUTLIER(3) excludes from the data sample any records that fall outside the third
outlier interval.
This same command without the SYSID keyword would combine the averaged degradation data for all
workloads that meet the other selection criteria, regardless of SYSID, and would generate only one profile.
This example assumes, of course, that the EPILOG datastore contains records with different SYSIDs.
Generating profiles for an entire workload type
PROFILE TSO(*) PNAME('PRIME SHIFT') LASTWEEK BAND STIME(0900) ETIME(1700) SIF(ELAPSED(<30m)) OUTLIER(3)
Because an asterisk (*) is specified with TSO, this command creates a profile for each TSO session
satisfying the selection criteria. PNAME assigns all such profiles the name PRIME SHIFT. Each profile
contains the averaged degradation data for each TSO ID that had a session last week between 9 a.m. and
5 p.m. where the elapsed time was less than 30 minutes.
Before the data is averaged, OUTLIER(3) excludes any job record that falls outside the third range of
outliers.
Multiple profiles are thus created with a single command. The result is the same as having specified a
PROFILE command for each TSO ID on the system.
Using the PNAME default
PROFILE JOB(*) PNAME LASTWEEK SIF(ELAPSED(>30m)) OUTLIER(3)
178
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
This command creates a profile for each job in the system that ran last week where the elapsed time was
greater than 30 minutes.
Since the user has not specified a value for PNAME, WPF assigns the PNAME default in lieu of a specific
profile name. To identify profiles created with the PNAME default in other WPF commands, simply specify
the PNAME keyword without any operand. Using the PNAME default simplifies the creation of profiles for
generic workloads.
OUTLIER(3) excludes from the data sample any records that fall outside the third outlier interval.
Displaying profiles
The Workload Profiling Facility uses the EPILOG reporter function to display profile reports. This section
assumes that you are already familiar with the DISPLAY command, which is explained in theEPILOG for
MVS Basic User’s Guide. You should consult that manual for complete explanations of the DISPLAY
format keywords as well as for basic operating procedures.
The profile display
There are two types of profile displays:
v The Standard Profile Report
v The Profile Report with Device Detail
The standard profile display
The standard profile report display is similar in format and content to the standard EPILOG degradation
display. A sample profile display follows:
+======================== Profile Report =====================================+
| Job = SORTNYP
Profname = PRIME SHIFT
|
| Period: 00:01 on 09/24/99 to 24:00 on 09/30/99
Sysid = SYSA |
| Outlier(3) Bounds(23.09 S, 1:59 M)
|
| EDS records: 17 Included and 1 Excluded
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
1.72 S 4.7 |-> .
.
.
.
.
.
.
.
.
.|
|I/O Active
17.40 S 48.0 |------------=======>.
.
.
.
.
.|
|Waiting for CPU
8.70 S 24.0 |---------> .
.
.
.
.
.
.
.|
|Enqueue
3.48 S 9.6 |--->.
.
.
.
.
.
.
.
.
.|
|ECB Wait
1.88 S 5.2 |--> .
.
.
.
.
.
.
.
.
.|
|Avg Job Elapsed Time 36.25 S
Average of 16 Jobs
|
|Productivity Index
22%
|
+=============================================================================+
Figure 115. Sample Profile Report display
This display was generated with the following command:
DISPLAY JOB(SORTNYP) PROFNAME('PRIME SHIFT') LASTWEEK
Its header is very similar to a standard degradation display. However, there are some differences. First, the
title Profile Report is used to distinguish it from non-profile displays. Also, the profile name specified on
the DISPLAY command (PRIME SHIFT) appears in the upper-right-hand corner. The second line shows the
time period covered by the profile, followed on the right by the SYSID. Finally, the third and fourth lines,
unique to profile displays, show the the OUTLIER criteria used to create the profile (OUTLIER(3)); the high
and low bounds of the sample (Bounds(23.09 S, 1:59 M)); and the number of EPILOG datastore records
included and excluded after OUTLIER processing (EDS Records: 17 Included and 1 Excluded).
The degradation section of the standard profile report display has almost the same format as the EPILOG
degradation display except that device wait reasons are listed generically rather than by unit address or
Chapter 10. Workload Profiling Facility (WPF)
179
volser, and the total number of batch jobs or transactions included in the profile is displayed in the
lower-right corner. The productivity index is the last line on the display.
When displaying job profiles, you will notice that sometimes the number of EPILOG datastore records
included after OUTLIER processing is different from the number of jobs included in the profile. (For
example, in our sample Profile Report, 17 EDS records were included for a profile of 16 jobs.) The reason
for this difference is that EDS records are kept for each job step. Thus, there may be more than one EDS
record for a single job. Similarly, for performance group profiles, the number of records included after
OUTLIER processing is generally different from the number of transactions included in the profile. This is
because EDS records are kept for each interval and each interval usually includes multiple transactions.
The profile report with device detail
If you want a particular profile report to display the unit addresses or volsers for the I/O devices listed
under wait reasons, you should specify the IODEVICE keyword when you issue the DISPLAY command.
For example, if we had specified IODEVICE on the same command used in the previous example, we
would have produced the following report:
+======================== Profile Report =====================================+
| Job = SORTNYP
Profname = PRIME SHIFT
|
| Period: 00:01 on 09/24/99 to 24:00 on 09/30/99
Sysid = SYSA |
| Outlier(3) Bounds(23.09 S, 1:59 M)
|
| EDS records: 17 Included and 1 Excluded
|
+-----------------------------------------------------------------------------+
|Wait_Reason_____________Time_____%_|0___1___2___3___4___5___6___7___8___9___0|
|Using CPU
1.72 S 4.7 |-> .
.
.
.
.
.
.
.
.
.|
|Disk WORK27 147 Act 9.28 S 25.6 |----------> .
.
.
.
.
.
.
.|
|Waiting for CPU
8.70 S 24.0 |---------> .
.
.
.
.
.
.
.|
|Disk WORK21 754 Act 5.22 S 14.4 |-----> .
.
.
.
.
.
.
.
.|
|SYSDSN
Enqueue
3.48 S 9.6 |--->.
.
.
.
.
.
.
.
.
.|
|ECB Wait
1.88 S 5.2 |--> .
.
.
.
.
.
.
.
.
.|
|Avg Job Elapsed Time 36.25 S
Average of 17 Jobs
|
|Productivity Index
22%
|
+=============================================================================+
Figure 116. Sample Profile Report with device detail
As you can see, this display is identical to the previous one except for the listing of disk wait reasons by
unit address.
When using the IODEVICE keyword, you should keep in mind that, in the profiles of some workloads, a
considerable number of devices may be associated with the various wait reasons. Therefore, specifying
IODEVICE on a DISPLAY command may produce a rather lengthy report. To limit the display to only
significant wait reasons, use the PLOTMIN keyword.
The DISPLAY command: generating profile displays or reports
Use the DISPLAY command (DIS) to generate a profile display or batch report for the specified workload
and profile name (PROFNAME). The DISPLAY command and its keywords are illustrated below:
DISPLAY {workload} {PROFNAME(cc...cc)} [time period] [SYSID(cccc,...)] [REPORTIF(filter)] [IODEVICE] [display options]
Figure 117. DISPLAY Command Format
180
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
KEYWORDS: (In the following list, only workload and PROFNAME are required.)
workload
Identifies the batch job, performance group, started task, or TSO user
session for which you want to display a profile. In addition to these
specific types of workloads, you can display a system-wide profile using
the SYSTEM keyword as the workload. For a complete list of workload
keywords, see “Workload keywords” on page 171. You can only use one
of these keywords at a time with the PROFILE command. All of these
keywords are valid on the PROFILE command except STEP and
INTERVAL.
PROFNAME(cc...cc)
Defines the 1- to 16-character profile name (or PNAME(ccc)) you assigned
to the workload on the PROFILE command. If you created the profile with
the PNAME default (that is, automatic), simply specify PNAME without any
operand. Asterisks (*) may be used as a wildcard characters for this
operand.
time period
Specifies the date and time interval covered by the profiles you want to
display. This time period does not refer to the date on which you created
the profile. Use any of the date/time keywords, as listed in “Date and time
keywords” on page 171, except BAND and EXDATE. If you specify a start
date and end date, you will display any profiles for the specified workload,
profile name, and, if specified, SYSID for that interval or any time period
within that interval.
If you specify only a start date, you will display any profiles for the
specified workload, profile name, and, if specified, SYSID, for any interval
beginning with that or any subsequent date.
If you specify only an end date, you will display any profiles for the
specified workload, profile name, and, if specified, SYSID, for any interval
that ends with that date.
If you do not specify a time period, you will display all profiles for the
specified workload, profile name, and, if specified, SYSID.
SYSID(cccc,...)
Specifies the SMF System ID of the profile you want to display. Required
only whenthe profiles in your Profile datastore have different SYSIDs and
you want to display a profile with a specific SYSID. If you specify multiple
SYSIDs, for example (SYSA,SYSB), you display any profile with a
corresponding SYSID that meets your other display criteria. If no SYSID is
specified, you display all profiles that satisfy your other display criteria
regardless of SYSID.
DISPLAY assumes that you have previously created a profile for any
SYSID(s) that you specify.
REPORTIF(filter)
Specifies your criteria for displaying profiles. The REPORTIF keyword
must be followed by a valid wait reason and appropriate reporting criteria.
Use any of the exception-threshold keywords described in “Workload
exception filters” on page 143. (REPORTIF(filter) can also be expressed
as RIF(filter)).
IODEVICE
Allows you to override the generic listing of device wait reasons, the
default, with a detailed I/O device listing. See “The profile report with
device detail” on page 180 for a sample display. (IODEVICE can also be
expressed as IODEV.)
display options
Specifies the options you want to use to format your profile display. Use
any of the following options:
COMPLETE
Chapter 10. Workload Profiling Facility (WPF)
181
LIMIT
LOGOFF
LOGON
SUMWAIT
PLOTMIN
TWAITOFF
TWAITON
These keywords are described in the EPILOG for MVS Basic User’s
Guide.
The following keywords are not valid when displaying profiles.
AMATRIX
AUTOMATIC
AVERAGE
COMBINE
DETAIL
EXWAIT
INTERVAL
MAXSCALE
SIF
SINGLE
STEP
SUMMARY
TOTAL
Examples of using the DISPLAY command
This section demonstrates how to generate various types of profile displays. It assumes that the
commands are being used interactively to generate online displays. The same commands can be used in
batch mode to generate profile reports. Note that whenever a sample command has been written on more
than one line, we have used the hyphen (-) as a continuation character.
Generating a profile display
DISPLAY PGN(2) PNAME('TSO USERS')
This command displays all profiles for performance group 2 with the profile name TSO USERS. If the
Profile datastore contains profiles for this workload and profile name under more than one SYSID, a profile
is displayed for each ID.
Generating a profile display for a particular time range
DISPLAY JOB(DEV*) PNAME('PRIME SHIFT') SDATE(03/01/99) EDATE(03/31/99)
This command displays the profile(s) with the profile name PRIME SHIFT for every job with a name that
begins with DEV, that ran between the 1st and 31st of March 1999. The number of profiles displayed
depends on the frequency of the profiling. For example, if the profile were created as a weekly run, then
there would be one profile for each week in March.
182
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Generating a profile display for particular SYSIDs
DISPLAY JOB(PRSNL) PNAME('PRIME SHIFT') SYSID(SYSA,SYSB) LASTMONTH
Because the SYSID keyword specifies two SMF system IDs, SYSA and SYSB, this command displays two
profiles: one for SYSA, the other for SYSB. Both profiles have the profile name PRIME SHIFT and contain
the averaged degradation data from the previous month for the job named PRSNL.
This example assumes that the Profile datastore contains some profiles with the SYSID SYSA and others
with the SYSID SYSB. If SYSID were not specified, a profile that meets the other selection criteria would
be displayed for each ID under which a profile was created.
Generating profile displays for an entire workload type
DISPLAY STC(*) PNAME(*)
Because an asterisk (*) is used to specify the workload and the profile name, this command displays all
profiles for all started tasks.
Generating profile displays with the PNAME default
DISPLAY STC(*) PNAME
Since no profile name is specified, this command displays only those started task profiles that were
created with the PNAME default (that is, automatic).
Generating a profile display with device detail
DISPLAY JOB(ADMIN01) PNAME(PAYROLL) SDATE(03/01/99) EDATE(03/31/99) IODEVICE
Because the IODEVICE keyword is specified, this command includes in the display the individual unit
addresses or volsers for any device that is listed as a wait reason. See “The profile report with device
detail” on page 180 for a sample display. This command displays any profiles named PAYROLL that were
created for ADMIN01 jobs, for any time period between March 1st 1999 and March 31st 1999.
Comparing workloads with profiles
To assess a particular workload’s recent performance, you may want to compare its current degradation
with a profile of its past degradation. For example, you may want to see a comparison of the last run of
the payroll job with a profile of payroll jobs for the past two months. Or you may want to compare this
week’s response times for TSO users with a profile of last month’s TSO response time.
Chapter 10. Workload Profiling Facility (WPF)
183
With the Workload Profiling Facility’s COMPARE and SETP commands, you can make such comparisons
between workloads and profiles and produce easy-to-read displays of the required information.
Comparisons can be made online or in batch mode. However, if a comparison requires processing large
amounts of data (for example, a generic-workload or a one-month time period specification), we
recommend batch mode. The sample batch job provided in rhilev.REDDATA(KEPSRCMP) contains several
COMPARE commands. These commands measure various TSO performance groups and batch jobs
against their respective profiles.
COMPARE displays
The COMPARE command produces two basic types of displays:
v The Profile/Workload Comparison Report
v The Profile/Workload Comparison Summary Report
The Profile/Workload Comparison report
The Profile/Workload Comparison report compares the degradation data for a selected workload(s) with a
given profile. This comparison is based on the EPILOG datastore data for the workload and the PROFILE
datastore data for the profile. When doing generic comparisons (e.g., comparing all batch jobs to their
profiles), workloads are displayed in descending order of their worst (highest elapsed or response time)
deviation from the profile. Below is a sample Profile/Workload Comparison report:
+======================== Profile/Workload Comparison ========================+
| Job = MGTMN01
Prof Job = MGTMN01
Profname = MGTJOBS
|
| Prof Period: 00:01 to 12:10 on 10/03/99
Sysid = SYSA |
| Wkld Period: 09:29 to 09:30 on 11/05/99
Sysid = SYSA |
+-----------------------------------------------------------------------------+
|
Times
Delta Plot (in Seconds)
|
|Wait_Reason____________Profile_Workload|16.59___ 8.29_____0_____ 8.29___16.59|
|Using CPU
2.48 S
0.26 S | .
.
.
. <-|
.
.
.
. |
|Waiting for CPU
9.80 S 26.39 S | .
.
.
. |+++++++++++++++> |
|Enqueue
0.95 S 11.73 S | .
.
.
. |+++++++++> . . |
|I/O Queued
1.19 S
2.93 S | .
.
.
.
|> .
.
.
. |
|ECB Wait
1.43 S
2.93 S | .
.
.
.
|> .
.
.
. |
|Swap Page-In Wait
0.23 S
n/a
| .
.
.
. <|
.
.
.
. |
|I/O Reserved
0.71 S
n/a
| .
.
.
. <|
.
.
.
. |
|SRM Delay (MPL)
1.67 S
n/a
| .
.
.
. <|
.
.
.
. |
|Waiting for MVS Lock
2.15 S
n/a
| .
.
.
. <-|
.
.
.
. |
|I/O Active
14.34 S
n/a
| . <------------|
.
.
.
. |
|Elapsed Time
34.99 S 44.27 S | .
.
.
. |+++++++> . . |
|Productivity Index
53%
41%
|
+=============================================================================+
Figure 118. Sample Profile/Workload Comparison report
This display was produced with the following command:
COMPARE JOB(MGTMN01) PROFNAME(MGTJOBS) YDAY RIF(ELAPSED(>10%))
This display compares the EPILOG datastore records for all jobs with the name MGTMN01 that ran
yesterday to the last profile created for MGTMN01 jobs with the profile name MGTJOBS.
The report header displays the name of the workload (JOB=MGTMN01) being compared, in the upper-left
corner, and the name of the profile (MGTJOBS) used in the comparison, in the upper-right corner.
Between the two, is the name of the workload used to create the profile (MGTMN01). The next two lines
show the time period, elapsed time, and SYSID for the profile, followed by the same information for the
workload.
To the left, the body of the report displays a parallel list of the major wait reasons for both the profile and
the workload. The wait reasons are listed in descending order of the workload’s highest elapsed-time
184
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
deviation from the profile. When a wait reason is significant for the profile and not for the workload, or
vice-versa, n/a replaces the time figure under the appropriate heading. This is the case in the above
example, where Swap Page-In Wait was a major wait reason for the profile but not for the workload.
The last two lines of the display show the difference in the elapsed time and productivity index for the job
and the profile. (Batch jobs, started tasks, and TSO user sessions show elapsed time at the bottom.
Performance groups show average response time instead of elapsed time.)
To the right, the body of the report graphically displays the Delta Plot, or how the workload deviated from
the profile, in hours, minutes, or seconds. Each of the wait reasons is plotted with either plus signs (+) or
minus signs (-). These signs indicate the the degree of deviation above (+) or below (-) the profile. The
scale of the plot is determined by the difference between the highest wait-reason or elapsed- or
response-time value. This value represents the high end and the low end of the delta plot; half of this
value is the middle-high and the middle-low point.
Profile/Workload Comparison with device detail listing
If you want a particular comparison report to display the unit addresses or volsers for the devices listed
under wait reasons, you should specify the IODEVICE keyword when you issue the COMPARE command.
For example, if we had specified IODEVICE on the same command used in the previous example, we
would have produced the following report:
+======================== Profile/Workload Comparison ========================+
| Job = MGTMN01
Prof Job = MGTMN01
Profname = MGTJOBS
|
| Prof Period: 00:01 to 12:10 on 10/03/99
Sysid = SYSA |
| Wkld Period: 09:29 to 09:30 on 11/05/99
Sysid = SYSA |
+-----------------------------------------------------------------------------+
|
Times
Delta Plot (in Seconds)
|
|Wait_Reason____________Profile_Workload|16.59___ 8.29_____0_____ 8.29___16.59|
|Using CPU
2.48 S
0.26 S | .
.
.
. <-|
.
.
.
. |
|Waiting for CPU
9.80 S 26.39 S | .
.
.
. |+++++++++++++++> |
|Disk WORK21 9E4 QUE
0.23 S
2.93 S | .
.
.
.
|+> .
.
.
. |
|ECB Wait
1.43 S
2.93 S | .
.
.
.
|> .
.
.
. |
|Disk COM001 146 ACT
0.23 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk MVSA21 150 ACT
0.23 S
n/a
| .
.
.
. <|
.
.
.
. |
|Swap Page-In Wait
0.23 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk DEVO24 322 ACT
0.23 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk DEVO26 144 RSV
0.23 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk COM002 152 ACT
0.23 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk COM003 157 RSV
0.47 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk DEVO27 147 QUE
0.71 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk WORK22 9E1 ACT
0.71 S
n/a
| .
.
.
. <|
.
.
.
. |
|MISC USR
Enqueue
0.95 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk DEVO26 144 ACT
0.95 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk TSO022 320 ACT
0.95 S
n/a
| .
.
.
. <|
.
.
.
. |
|SRM Delay (MPL)
1.67 S
n/a
| .
.
.
. <|
.
.
.
. |
|Disk WORK21 9E4 ACT
2.15 S
n/a
| .
.
.
. <-|
.
.
.
. |
|Waiting for MVS Lock
2.15 S
n/a
| .
.
.
. <-|
.
.
.
. |
|Disk DEVO27 147 ACT
8.60 S
n/a
| . . <-------|
.
.
.
. |
|Elapsed Time
34.99 S 44.27 S | .
.
.
. |+++++++> . . |
|Productivity Index
43%
10%
|
+=============================================================================+
Figure 119. Sample Profile/Workload Comparison report with device detail
As you can see, this display is similar to the previous one. The only difference is the listing of disk wait
reasons by unit address.
When using the IODEVICE keyword, you should keep in mind that a considerable number of devices may
be associated with the various wait reasons. Therefore, specifying IODEVICE on a COMPARE command
may produce a rather lengthy report. To limit the number of wait reasons displayed, you can use the
LINECNT(nnnnn) keyword, which allows you to specify the maximum number (0 - 32767) of wait-reason
lines. The default is LINECNT(20).
Chapter 10. Workload Profiling Facility (WPF)
185
The Profile/Workload Comparison Summary report
The second type of comparison report lists, in descending order, the workloads that deviated most from
their profiles. A sample summary report follows:
+======================== Compare Profile Summary ============================+
| Profile Job = ************
Profname = (automatic)
|
| Job = ************
|
| Profile Period:
00:01 on 11/11/99 to 24:00 on 11/11/99
|
| Workload Period: 00:02 on 10/03/99 to 15:52 on 11/11/99
|
+-----------------------------------------------------------------------------+
|
Delta
Workload
Profile
|
| Workload
Time
Run Time Sdate
Stime Run Time Sdate
Edate |
| USER08AA
18:31 M 19:13 M 10/03/89 09:29 42.01 S 11/11/99 11/11/99|
| USER12A
11:29 M 12:11 M 10/29/89 08:53 42.09 S 11/11/99 11/11/99|
| USER07A
10:47 M 12:35 M 10/28/89 12:39
1:47 M 11/11/99 11/11/99|
| USER08AA
9:21 M 10:03 M 10/18/89 11:20 42.01 S 11/11/99 11/11/99|
| USER07A
8:31 M 10:18 M 11/07/89 15:33
1:47 M 11/11/99 11/11/99|
| USER12A
6:42 M
7:24 M 10/14/89 11:57 42.09 S 11/11/99 11/11/99|
| USER07A
6:22 M
8:10 M 10/16/89 10:47
1:47 M 11/11/99 11/11/99|
| USER07A
3:43 M
5:31 M 10/30/89 15:36
1:47 M 11/11/99 11/11/99|
| USER08AA
3:36 M
4:18 M 10/18/89 10:54 42.01 S 11/11/99 11/11/99|
| USER08AA
3:33 M
4:15 M 10/11/89 11:08 42.01 S 11/11/99 11/11/99|
| USER08AA
3:11 M
3:53 M 10/11/89 10:58 42.01 S 11/11/99 11/11/99|
| USER12A
2:50 M
3:32 M 10/25/89 11:32 42.09 S 11/11/99 11/11/99|
| USER12A
2:49 M
3:31 M 10/14/89 12:18 42.09 S 11/11/99 11/11/99|
| USER08AA
2:41 M
3:23 M 10/23/89 11:30 42.01 S 11/11/99 11/11/99|
| USER08AA
2:40 M
3:22 M 10/08/89 09:19 42.01 S 11/11/99 11/11/99|
| USER07A
2:36 M
4:24 M 10/28/89 11:43
1:47 M 11/11/99 11/11/99|
| USER08AA
2:33 M
3:15 M 10/10/89 14:31 42.01 S 11/11/99 11/11/99|
| USER08AA
2:25 M
3:07 M 11/11/89 13:45 42.01 S 11/11/99 11/11/99|
| USER07A
2:10 M
3:57 M 11/04/89 11:08
1:47 M 11/11/99 11/11/99|
| USER07A
2:01 M
3:48 M 10/30/89 09:47
1:47 M 11/11/99 11/11/99|
+=============================================================================+
Figure 120. Sample Profile/Workload Comparison Summary report
This report was produced with the following command:
COMPARE JOB(*) PROFNAME SUMMARY
The report header shows the name of the workload (Profile Job = *) and (Job= *) together with the
profile name (Profname = (automatic)) specified on the COMPARE command. The next two lines show
the time periods for the profile and the workload. In the above example, the asterisks given for the
workload indicate that all jobs for the specified time period and profile name should be included in the
comparison. The substitute profile name, (automatic), indicates that no specific profile name was given,
and that the user expects the PNAME default. As a result, all profiles created with the PNAME default are
automatically included in the comparison process.
For each workload included in the report, the following information is supplied:
v The delta time, or the amount of elapsed or response time by which it deviates from its profile. (For
batch jobs, started tasks, and TSO session, the delta time is based upon the difference in elapsed times
between the profile and the workload. For performance groups, delta time is based upon the difference
in average response times.)
A minus sign (-) indicates that the workload’s elapsed or response time was less than the profile’s.
v The workload’s run time, start date, and start time.
v The profile’s run time, start date, and end date.
The COMPARE command: comparing workloads with profiles
Use the COMPARE command (COM) to compare the degradation data for a given workload with that of a
profile. Unless preceded by a SETP command directing otherwise, the COMPARE command chooses the
186
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
most current profile for the workload and PROFNAME that you have specified. The COMPARE command
and its keywords are illustrated below:
COMPARE {workload} {PROFNAME(cccc)} [time period] [SYSID(cccc,...)] [REPORTIF(filter)] [IODEVICE] [LINECNT(nnnnn)] [SUMWAIT] [SUMMARY]
Figure 121. COMPARE command syntax
KEYWORDS: (In the following list, only workload and PROFNAME are required.)
workload
Identifies the group of jobs or performance group you are selecting for
comparison with a profile. You can use any of the workload-type
keywords, as defined in “Workload keywords” on page 171, except STEP
and INTERVAL.
PROFNAME(cccc)
Defines the 1- to 16-character profile name (or PNAME(ccc) of the profile
against which you want to compare the EPILOG datastore records.
Typically, this is the name you assigned to the specified workload when
you created its profile. If you created the profile with the PNAME default
(that is, automatic), simply specify PNAME without any operand. Note that
you cannot use the asterisk as a substitute for a profile name or for any
part of a profile name (for example, TSO*).
time period
Specifies the date and/or time interval for which you are selecting the
specified workload’s degradation data (EPILOG datastore records). Use
any of the date/time keywords as defined in “Date and time keywords” on
page 171. If no time period is given, all EPILOG datastore records for the
specified workload are selected.
Note that your time period specification applies only to the selection of
EPILOG datastore data. Unless you have indicated otherwise on a prior
SETP command, the profile selected is the most current profile for the
workload, PROFNAME, and SYSID that you have specified on the
command.
SYSID(cccc,...)
Specifies the SMF System ID for the workload you are comparing.
Required onlywhen the records in your datastores contain different
SYSIDs and you want to compare a workload and profile with a specific
SYSID. If you specify multiple SYSIDs, for example (SYSA,SYSB), you
will compare each workload-profile combination with a corresponding
SYSID that meets your other comparison criteria. If no SYSID is specified,
the specified workload and the most recent profile that meet your other
comparison criteria will be compared regardless of SYSID. If you want to
compare a profile that has a SYSID different from that specified on the
COMPARE command, you may do so by setting the SYSID of the profile
with the SETP command.
REPORTIF(filter)
Specifies the criteria that workload records must satisfy to be included in a
comparison. The REPORTIF keyword (or RIF(filter)) must be followed by a
valid wait reason and appropriate reporting criteria. Use any of the
exception-threshold keywords described in “Workload exception filters” on
page 143. In addition, you may apply the filter (ELT(>nn%) or
Chapter 10. Workload Profiling Facility (WPF)
187
RESP(>nn%)) to indicate that you want to compare only those workload
records with an elapsed (ELT) or response (RESP) time nn% greater than
the profile’s.
IODEVICE
Allows you to override the generic listing of device wait reasons, the
default, with a detailed I/O device listing. (IODEVICE can also be
expressed as IODEV.) See “Profile/Workload Comparison with device
detail listing” on page 185 for a sample display.
LINECNT(nnnnn)
Specifies the maximum number of wait-reason lines that you want to
display. When specified together with SUMMARY, LINECNT (or
LCNT(nnnnn) limits the number of records to be displayed. You may
specify any number from 0 to 32767. The default is LINECNT(20). Note
that Using CPU and the averaged response- or elapsed-time lines will
always be displayed and are not counted as wait reasons.
SUMWAIT
Indicates that summary wait categories should be displayed instead of
detailed wait reasons. (SUMWAIT can also be expressed as SUMW.)
SUMMARY
Indicates that all data should be displayed in summary format. That is, the
following information will be displayed:
v the amount of elapsed or response time by which the workload deviates
from its profile
v the workload’s run time, start date, and start time
v the profile’s run time, start date, and end date
(SUMMARY can also be expressed as SUM.)
See “The Profile/Workload Comparison Summary report” on page 186 for
a sample display.
Notes:
1. The COMPARE command can only be used to compare a specific workload type to a profile.
2. The REPORTIF filter applies only to the EPILOG datastore records for the specified workload and not
to the profiles.
Examples of using the COMPARE command
This section demonstrates how to compare the current degradation for a workload with a given profile.
Note that whenever a sample command has been written on more than one line, we have used the
hyphen (-) as a continuation character.
Comparing a workload with a profile
COMPARE JOB(PAYROLL) PNAME(PAY) TODAY RIF(ELAPSED(>15%))
This command displays a report comparing the current day’s job named PAYROLL with a profile created
for the same workload. Because no SETP command was issued prior to this command, the basis for
comparison is the most current profile. In addition, the RIF keyword indicates that only those EPILOG
datastore records with elapsed times which exceed the profile’s by 15% should be displayed.
See “The Profile/Workload Comparison report” on page 184 for a sample display.
188
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Generating a Profile/Workload Comparison Summary report
COMPARE JOB(PAYROLL) PNAME(PAY) TODAY STIME(1700) ETIME(0700) RIF(ELAPSED(>15%)) SUMMARY
This command compares the same data as the preceding report but displays the data in summary format;
only the following information is displayed:
v the amount of elapsed or response time by which the workload deviates from its profile
v the workload’s run time, start date, and start time
v the profile’s run time, start date, and end date
(See “The Profile/Workload Comparison Summary report” on page 186 for a sample display.)
Displaying the workloads that deviate most from their profiles
COMPARE JOB(*) PNAME SUMMARY
This command also generates a summary report. However, because an asterisk (*) has been substituted
for a job name and the PNAME is set to default, the report displays, in descending order, the workloads
that deviated most from their profiles. Because no LINECNT is specified, up to twenty workloads are listed.
This report has the same format as the sample shown in “The Profile/Workload Comparison Summary
report” on page 186.
Comparing workloads and profiles for different SYSIDs
COMPARE JOB(PAYROLL) PNAME(PAY) SYSID(SYSA,SYSB) LASTWEEK STIME(1700) ETIME(0700)
Because multiple SYSIDs are specified, qualifying workload-profile combinations for both SYSIDs (SYSA
or SYSB) are compared. Thus, for SYSA and SYSB, this command displays a report comparing the
EPILOG datastore records for the job PAYROLL from the previous week between the hours of 5 PM and 7
AM, against the most recent profile for the corresponding SYSID, that has the name PAY and that was
created for the job PAYROLL.
The SETP command: setting default values for the COMPARE
command
You can use the SETP command to establish a default workload, time period, and/or SYSID value for
subsequent COMPARE commands. The COMPARE command then uses this value to select the profile
from the Profile datastore (PRDS). These defaults remain in effect until changed by a subsequent SETP
command or cancelled by SETP CLEAR.
The SETP command and its keywords are illustrated below:
Chapter 10. Workload Profiling Facility (WPF)
189
SETP
workload time period SYSID(cccc,...)
CLEAR
-
Figure 122. SETP command format
KEYWORDS: (In the following list, all keywords are optional.)
workload
Identifies the batch job or performance group that should be used in
selecting the profile for a comparison. Use any of the workload type
keywords, as defined in “Workload keywords” on page 171, except STEP
and INTERVAL. You may not specify multiple workloads.
time period
Specifies the date and/or time interval of the profile to be used in the
comparison. This date does not refer to the date on which you created the
profile, but rather to the time period covered by the profile. Use any of the
date/time keywords as defined in “Date and time keywords” on page 171.
SYSID(cccc,...)
Specifies the SMF System ID that should be used to select the profile(s)
for a comparison. You may specify multiple SYSIDs, for example
(SYSA,SYSB), to select a profile for each SMF ID.
CLEAR
This keyword cancels previously established SETP defaults.
Notes:
1. SETP is required when the profile you want for the comparison has a workload name, time period, or
SYSID different from the most current profile for PROFNAME on the COMPARE command.
2. Until cancelled by SETP CLEAR, values established by SETP commands are combined to create a
cumulative default value. However, when you specify a workload value of any type with a SETP
command, it replaces any workload previously established. Similarly, a time period value specified on a
SETP command replaces a duplicate type established by a previous SETP command.
For example, PGN(2) STIME(0900) ETIME(1700) established with one SETP command would be
combined with SDATE(6/1) EDATE(6/30) established with another. However, PGN(12) established with
a subsequent SETP command would replace the previously set workload default value PGN(2).
Similarly, STIME(1200) established with a subsequent SETP command would replace the previously
set time period default value STIME(0900).
Examples of using the SETP command
This section demonstrates how to use the SETP command to specify default values for the profile used by
the COMPARE command. Note that whenever a sample command has been written on more than one
line, we have used the hyphen (-) as a continuation character.
Setting a default time period for profile selection
SETP
SDATE(04/01/99) EDATE(04/30/99)
This command specifies the time period for the profile to be used in the comparison. As a result, the
profile must include degradation data from April 1st to April 30th inclusive, or for any time period between
those dates.
190
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Comparing a workload with a profile other than the most current
SETP JOB(DEV0) LASTMONTH
COMPARE JOB(DEV12) PNAME(DEVBATCH) TODAY RIF(ELAPSED(>10%))
The SETP command specifies the workload and time period for the profile to be used in the comparison.
As a result, the COMPARE command generates a report comparing the EPILOG datastore records for
DEV12 jobs run today and that have elapsed times which exceed the profile’s by 10%. The basis of
comparison is the most current profile with the name DEVBATCH that was created for DEV0 jobs run last
month.
Comparing a workload with profiles having different SYSIDs
SETP SYSID(SYSA,SYSB) LASTMONTH
COMPARE JOB(DEV0) PNAME(DEVBATCH) TODAY RIF(ELAPSED(>10%))
The SYSIDs (SYSA and SYSB) and the time period (LASTMONTH) specified on the SETP command are
the defaults that the COMPARE command uses to select the profiles for the comparison. The selection
criteria specified in the COMPARE command (TODAY and RIF) are applied to the workload.
As a result, this COMPARE command generates a report comparing the EPILOG datastore records for
DEV0 jobs run today, with elapsed times that exceed the profile’s by 10%, against the most current
profiles for either SYSA or SYSB with the name DEVBATCH that were created for DEV0 jobs run during
the previous month.
Chapter 10. Workload Profiling Facility (WPF)
191
192
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 11. Exporting EPILOG data
The OBTAIN utility exports data from an EPILOG datastore to a sequential output file where it can be used
by mainframe and PC application packages. This chapter shows how to use the OBTAIN command to run
the OBTAIN utility.
The OBTAIN utility
To export data with the OBTAIN utility, you can either submit a batch job or execute a CLIST. Two sample
files are included on the product tape to facilitate this process. One sample file,
rhilev.REDDATA(KEPOBT), contains sample JCL to submit an OBTAIN job; the other,
rhilev.REDDATA(KEPOBTC), contains a sample CLIST. The JCL or CLIST must be modified to identify
your output data sets. The data to be exported is identified by EPILOG and OBTAIN commands and
keywords in the EPSYSIN data set for the OBTAIN utility.
Using the OBTAIN utility
Follow these steps to export data using the OBTAIN utility.
1. Edit the sample JCL or CLIST. Set the system parameters for your installation.
The output file for the data that is exported by OBTAIN can be defined in one of two ways:
v Define an available data set in the OUTFILE DD statement. This is the default method; it does not
require the OUTFILE keyword in the OBTAIN command.
v Define one or more available data sets by defining one or more DD statements. Each DD statement
must correspond to a DD name in the OUTFILE keyword of an OBTAIN command in the OBTAIN
command stream.
2. Create a data set containing OBTAIN and SET commands and keywords to describe the data that you
want to export. Specify the data set name of the OBTAIN command file in the EPSYSIN DD statement.
3. Submit the JCL or execute the CLIST.
Output will be written to the file specified on the OUTFILE DD statement in the JCL file, or to a file defined
by the ddname specified on an OUTFILE keyword in the command file.
Syntax
Use the syntax illustrated below when entering OBTAIN and SET commands in the EPSYSIN data set.
(The OBTAIN command can be abbreviated to OBT.)
OBTAIN {report-name, ...}
{FORMAT(INT|COL|PC}
[ELEMENTS(field-name, ...)]
[OUTFILE(ddname)]
[REPLACE|APPEND]
[any DISPLAY keyword]
Figure 123. OBTAIN and SET command syntax
This section describes the keywords that are required by the OBTAIN command, as well as keywords that
are unique to OBTAIN. All of the OBTAIN keywords (except the report-name keywords) can also be used
with the SET command.
report-name
These keywords identify the resource and workload reports from which data is to be
exported. For a list of the resource keywords that you can use with the OBTAIN command,
see Table 9 on page 63. For a list of the workload keywords that you can use, see
Table 18 on page 171. You must specify at least one report-name keyword in each
OBTAIN command.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
193
If you specify more than one report-name keyword in an OBTAIN command, you must
enclose the keywords in parentheses; then, the data from the reports will be joined on
their common fields. In most cases, reports will be joined on starting date and time.
Some EPILOG reports (such as RCPU) generate one record per interval; others (such as
RDAS) generate multiple records per interval. If you use more than one report-name
keyword with an OBTAIN command and the keywords mix single-record-per-interval
reports and multiple-record-per-interval reports, single-record-per-interval elements will be
repeated until the interval changes. Joining resource reports and workload reports in an
OBTAIN command is not recommended; in most cases, the resulting OBTAIN command
will not export any data.
FORMAT(INT|COL|PC)
This keyword specifies the format of the data to be written to the output file. Three options
are available for the FORMAT keyword: INT, COL, and PC. INT is the default format.
INT writes binary data formatted as described below in “Output files generated by
FORMAT(INT)” on page 195. The INT option is advisable if you will be writing your own
programs to read the OBTAIN output file.
COL writes printable EBCDIC data arranged in columns. The COL option is advisable if
the output file will be used as input to a mainframe graphics or statistics package, such as
GDDM/ICU¬, or if the output file will be downloaded and used as input to a PC software
package such as Lotus®; 1-2-3®;. Your installation’s download utility will handle the
EBCDIC-to-ASCII conversion.
PC writes printable EBCDIC data arranged in columns. The PC format is the same as the
COL format, except that data elements in the COL format are arranged in columns, and
data elements in the PC format are delimited with double-quotation marks (for character
strings) and blanks (for numbers). Use the format that is best suited for the software
application you will be using to process the exported data. For examples illustrating the
differences in the COL and PC formats, see “Output file layouts” on page 195.
ELEMENTS(field-name,...)
Identifies the elements to be selected from the specified reports. Each EPILOG field name
is associated with an element name; in addition, there are data fields that are not
displayed on EPILOG reports, but which are available for OBTAIN processing. For a
description of element names available for OBTAIN processing sorted by report-name
keyword, see Chapter 13, “Report element tables,” on page 233.
If ELEMENT (ELEM for short) is omitted or ELEMENT(*) is specified, all elements for the
named reports are selected.
Field names specified by ELEM on an OBTAIN command are extracted only for that
command. Field names specified by ELEM on one or more SET commands are
cumulative. They will be extracted on all subsequent OBTAIN commands for reports that
contain them. SET ELEM(*) or SET ELEM() clears previous elements activated by SET
commands.
OUTFILE(ddname)
Specifies the name on a DD or FILE statement that defines the data set to be used as the
output file for the OBTAIN command. Use this keyword if you want to override the default
name OUTFILE or to export data to more than one data set during the execution of an
OBTAIN command.
For example, suppose that you want to export data to two data sets, DSN1 and DSN2.
You can do this by creating two DD statements, DSN1 and DSN2, that define the data
sets DSN1 and DSN2. You must also add the keywords OUTFILE(DSN1) and
OUTFILE(DSN2) to the corresponding OBTAIN commands. (Each OBTAIN command can
define only one OUTFILE.)
194
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
If you do not use the OUTFILE keyword, OBTAIN uses the data set that is defined in the
OUTFILE DD statement as the output file.
REPLACE|APPEND
Specifies whether obtained data should overwrite (REPLACE) records in an existing output
file with the same name, or add (APPEND) records to the end of the file.
When REPLACE is used on an output file with DISP=MOD, the obtained data is added to
the end of the file. The existing data is not overwritten.
DISPLAY keywords
In addition to the keywords described here, you can use all of the DISPLAY keywords with
the OBTAIN and SET commands. For a description of the DISPLAY keywords, see the
EPILOG for MVS Basic User’s Guide.
Output file layouts
Each OBTAIN command writes data to a sequential output file that has variable length records. The format
of the output file is controlled by the FORMAT keyword on an OBTAIN or SET command in the command
file.
Output files contain two or three record types, depending on the format of the file. This section shows a
sample record and provides detailed field descriptions for each record type in each of the three output file
formats. A sample output file is provided for each of the three output file formats in “Output file examples”
on page 197.
Output files generated by FORMAT(INT)
This output file contains three record types: the comment record, the attribute record, and the observation
record. The comment records contain the text of the commands; the attribute record defines the elements
being selected; and the observation records contain the element values selected from the datastore.
Your output file will contain one or more comment records for each OBTAIN command in the command
file. A sample comment record is shown below.
0089216F 0083413F *
OBTAIN RINF OUTFILE(SPFTMP.RINF) REPLACE -
The record contains the following fields:
v Four-byte packed decimal date field containing the Julian date on which the data was extracted.
v
v
v
v
Four-byte time stamp containing the time of day that the data was extracted.
One-byte record-type indicator of *, indicating that this is a comment record.
One-byte reserved field.
The text of the OBTAIN command as it was typed by the user. This field also contains the current
settings of any valid keywords that are used for data filtering (such as DATE, TIME, BAND, RANGE,
RIF, or SIF) and the default value of the DATEFORM keyword, if applicable.
Your output file will contain one attribute record for each OBTAIN command in the command file. A sample
attribute record follows.
0089216F 0083413F A 03 SDATE P 4 0 00A STIME T 4 0 00E SMFID C 4 0 012
The record contains the following fields:
v Four-byte packed decimal date field containing the Julian date on which the data was extracted.
v Four-byte time stamp containing the time of day that the data was extracted.
Chapter 11. Exporting EPILOG data
195
v One-byte record-type indicator of A, indicating that this is an attribute record.
v One-byte reserved field.
v Two-byte integer field containing the number of elements being OBTAINed, that is, the number of value
fields that will appear on each observation record.
v For each element requested, plus SDATE and STIME, an element definition consisting of the following
fields:
– Eight-byte character field containing the element name.
– One-byte character field containing the data type (I=numeric signed integer, U=unsigned integer,
P=packed decimal, C=character, D=floating-point, X=hexadecimal, and T=time).
– One-byte integer field containing the element length in bytes.
– One-byte integer field containing the decimal point displacement (number of positions to the right of
the decimal).
– Three-byte integer field containing the field offset (location of the field from the beginning of the
observation record, in bytes).
The number of observation records for each recording interval depends on the report name keyword, as
noted in Chapter 13, “Report element tables,” on page 233. A sample observation record follows.
0089200F 0083413F O
0089214F 084500
SYSA AA AA RMF 4.1.1
The record contains the following fields:
v Four-byte packed decimal date field containing the Julian date on which the record was extracted.
v Four-byte time stamp containing the time of day that the record was extracted.
v One-byte record-type indicator of O.
v One-byte reserved field.
v Four-byte packed decimal date field (SDATE) containing the Julian date on which the data was
recorded.
v Four-byte time field (STIME) containing the start time of the interval for which the data was recorded.
v Value field for each element selected by the user, in the order listed on the attribute record.
An example of a typical OBTAIN output file generated by FORMAT(INT) is shown in Figure 124 on page
198.
Output files generated by FORMAT(COL)
This output file contains two record types: the attribute record and the observation record. For each
OBTAIN command issued using FORMAT(COL), there is one attribute record defining the elements being
selected and multiple observation records containing element values selected from the datastore.
Your output file will contain one attribute record for each OBTAIN command in the command file. A sample
attribute record is shown below.
*SDATE
00 008 P *STIME
08 008 T *SMFID
10 004 C *IPS
14 002 C
The record contains the following fields:
v One-byte character field containing an asterisk (*).
v Eight-byte character field containing the element name.
v Three-byte integer field containing the offset (position of the field from the beginning of the record).
v Three-byte integer field containing the length of the element in bytes.
196
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
v One-byte character field containing the data type (N=numeric signed integer, C=character, and
H=hexadecimal).
Your output file will contain one observation record for each recording interval. A sample observation
record is shown below.
008920008:45:00SYSAAAAARMF 4.1.1
The record contains the following fields:
v Eight-byte packed decimal date field (SDATE) containing the Julian date on which the data was
recorded.
v Eight-byte time field (STIME) containing the start time of the interval for which the data was recorded.
v Value field for each element selected by the user, in the order listed on the attribute record.
The contents of a typical OBTAIN output file for FORMAT(COL) are shown in Figure 125 on page 199.
Output files generated by FORMAT(PC)
This output file contains two record types: the attribute record and the observation record. For each
OBTAIN command issued with FORMAT(PC), there is one attribute record defining the elements being
selected and multiple observation records containing element values selected from the datastore.
Your output file will contain one attribute record. A sample attribute record is shown below.
"SDATE
" "STIME
" "SMFID
" "IPS
" "ICS
" "RMF
"
The record contains the element names that have been selected, delimited by double quotes.
Your output file will contain one observation record for each recording interval. A sample observation
record is shown below.
"0089214" "08:00:01" "SYSA" "AA" "AA" "RMF 4.1.1
"
The record contains the following fields:
v Eight-byte packed decimal date field (SDATE) containing the Julian date on which the data was
recorded.
v Eight-byte time field (STIME) containing the start time of the interval for which the data was recorded.
v Value field for each element selected by the user, in the order listed on the attribute record. If the value
is in character or hexadecimal format, it will be enclosed in double quotes.
An example of a typical OBTAIN output file for FORMAT(PC) is shown in Figure 126 on page 199.
Output file examples
This section presents three examples of OBTAIN/SET command sequences followed by the output files
they create. One example is given for each of the three possible file layouts (INT, COL, and PC).
An output file generated by FORMAT(INT)
The following sample command sequence generated five comment records, one attribute record, and eight
observation records.
Chapter 11. Exporting EPILOG data
197
SET
SET
SET
OBT
FORMAT(INT)
OUTFILE(RINF) REPLACE
ELEMENTS(SMFID)
RINF STIME(9) ETIME(11) SDATE(8/2) EDATE(8/2) ELEMENTS(IPS,ICS,RMF)
END
The first two SET commands in this example specify the default format, name, and disposition of all output
files created by subsequent OBTAIN commands in this job. The first SET command causes all output files
to be written in INT format, unless specified otherwise. The second SET command specifies that the
output file will be written to the data set that is defined in the DD or FILE statement RINF. The third SET
command requests that all subsequent OBTAIN commands in the command sequence extract the element
SMFID along with any other elements specified.
A sample of the output file appears in Figure 124. The records displayed in the figure have been
interpreted into readable format. If you were to list the contents of an OBTAIN output file generated with
FORMAT(INT), much of the data would be non-readable, binary data.
0089216F
0089216F
0089216F
0089216F
0083413F
0083413F
0083413F
0083413F
*
*
*
A
0089216F
0089216F
0089216F
0089216F
0089216F
0089216F
0089216F
0089216F
0083413F
0083413F
0083413F
0083413F
0083413F
0083413F
0083413F
0083413F
O
O
O
O
O
O
O
O
OBT RINF STIME(9) ETIME(11) SDATE(8/2) EDATE(8/2) DATEFORM(MMDDYY)
ELEMENTS(IPS,ICS,RMF)
06 SDATE P 4 0 00A STIME
T 4 0 00E SMFID C 4 0 012
IPS
C 2 0 016 ICS
C 2 0 018 RMF
C20 0 01A
0089214F 090004
SYSA AA AA RMF 4.1.1
0089214F 091502
SYSA AA AA RMF 4.1.1
0089214F 093004
SYSA AA AA RMF 4.1.1
0089214F 094501
SYSA AA AA RMF 4.1.1
0089214F 100000
SYSA AA AA RMF 4.1.1
0089214F 101501
SYSA AA AA RMF 4.1.1
0089214F 103002
SYSA AA AA RMF 4.1.1
0089214F 104501
SYSA AA AA RMF 4.1.1
Figure 124. Output file for OBTAIN using FORMAT(INT)
Since the default sampling interval (15 minutes) is in effect, and the requested reporting interval is 2 hours
(9 to 11), eight observation records are produced.
Each observation record contains the following information:
v
v
v
v
date and time on which the data was extracted from the datastore
record-type indicator O
start date and time (SDATE and STIME) of the interval for which the data was recorded
SMFID, IPS suffix, ICS suffix and RMF level that were active during the recording interval
An output file generated by FORMAT(COL)
The following command generates one attribute record and eight observation records.
SET
SET
SET
OBT
FORMAT(COL)
OUTFILE(RINF) REPLACE
ELEMENTS(SMFID)
RINF STIME(9) ETIME(11) SDATE(8/2) EDATE(8/2) ELEMENTS(IPS,ICS,RMF)
END
A sample of the output file is shown in Figure 125 on page 199. The records displayed in the figure are in
the same format as they would appear if you listed the file at a terminal or printer.
198
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
*SDATE
00 008 N *STIME
*ICS
008921409:00:04SYSAAAAARMF
008921409:15:02SYSAAAAARMF
008921409:30:04SYSAAAAARMF
008921409:45:01SYSAAAAARMF
008921410:00:00SYSAAAAARMF
008921410:15:01SYSAAAAARMF
008921410:30:02SYSAAAAARMF
008921410:45:01SYSAAAAARMF
08 008 N *SMFID
16 002 C *RMF
10 004 C *IPS
18 020 C
14 002 C
4.1.1
4.1.1
4.1.1
4.1.1
4.1.1
4.1.1
4.1.1
4.1.1
Figure 125. Output file for OBTAIN using FORMAT(COL)
The output file pictured above will be written to the data set defined by the RINF DD statement, as
specified on the OUTFILE keyword.
Note that since the default sampling interval is 15 minutes and the requested reporting interval is 2 hours
(9 to 11), eight observation records are produced.
Output files generated by FORMAT(PC)
The following command generates one attribute record and eight observation records.
SET
SET
SET
OBT
FORMAT(PC)
OUTFILE(RINF) REPLACE
ELEMENTS(SMFID)
RINF STIME(8) ETIME(17) SDATE(8/2) EDATE(8/2) CMB(1H) ELEMENTS(IPS,ICS,RMF)
END
A sample of the output file is shown in Figure 126. The records displayed in the figure are in the same
readable format as they would appear if you listed the file at a terminal or printer.
"SDATE " "STIME
" "SMFID
"
"0089214" "08:00:01" "SYSA" "AA"
"0089214" "09:00:01" "SYSA" "AA"
"0089214" "10:00:01" "SYSA" "AA"
"0089214" "11:00:01" "SYSA" "AA"
"0089214" "12:00:04" "SYSA" "AA"
"0089214" "13:00:02" "SYSA" "AA"
"0089214" "14:04:28" "SYSA" "AA"
"0089214" "15:00:01" "SYSA" "AA"
"0089214" "16:00:01" "SYSA" "AA"
"IPS
"AA"
"AA"
"AA"
"AA"
"AA"
"AA"
"AA"
"AA"
"AA"
" "ICS
"RMF 4.1.1
"RMF 4.1.1
"RMF 4.1.1
"RMF 4.1.1
"RMF 4.1.1
"RMF 4.1.1
"RMF 4.1.1
"RMF 4.1.1
"RMF 4.1.1
" "RMF
"
"
"
"
"
"
"
"
"
"
Figure 126. Output file for OBTAIN using FORMAT(PC)
Since the requested combine period was one hour on this OBTAIN command and the requested reporting
interval is nine hours, nine observation records are produced.
Graphic reporting with sequential EPILOG data
Once you have extracted sequential EPILOG data, you can use other software tools (such as GDDM/ICU)
to manipulate the data and create customized statistical or graphic reports. For instructions concerning the
use of sequential EPILOG data in customized reports, see Chapter 13, “Report element tables,” on page
233.
Chapter 11. Exporting EPILOG data
199
200
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 12. Reporting with SAS graphics
The SAS interface allows you to extend the unique performance, degradation, and profile data produced
by EPILOG for MVS to graphic reports using SAS. It extracts the data from both the EPILOG datastore
and the Profile datastore and presents it to SAS in a format suitable for SAS statistical reduction and
graphic presentation. This data can then be formatted into a variety of reports using standard SAS
programming. Alternatively, the data can be added to an existing SAS performance database to enhance
your current performance analysis system.
A starter set of reports is included with EPILOG for MVS. These ready-to-use reports serve two functions:
1. They allow you to start producing meaningful graphs about your system resources and workloads
without having to write SAS code.
2. They serve as models for customized SAS programs that you may want to write for your installation.
Note: In previous releases of EPILOG for MVS, the SAS interface was referred to as the PRO¬ for MVS
reporting feature. If you used PRO in previous releases, you can continue to use it with no
conversion or recustomization. The product code PM has been retained in some of the sample JCL
and CLISTs on the product tape. It now refers to the SAS interface.
How the SAS interface can help you
The SAS interface helps you in each of the following areas:
v Performance Problem Resolution
v Management Reporting
v Capacity Planning and Trending
Performance problem resolution
EPILOG helps you identify the main wait reasons responsible for performance degradation. The SAS
interface makes this data available for SAS statistical analysis and graphic reporting. You can design
custom reports that clearly identify the factors that account for batch job elapsed time or performance
group response time at your installation. These reports also give you a clear picture of how various system
resources are being utilized, and which resources are constraining system performance.
In addition, the SAS interface also makes workload profile data available for SAS formatting. With this
data, you can generate SAS reports that compare workloads’ current degradation to profiles of their past
performance.
Management reporting
The statistical reduction and graphics capabilities available with SAS, combined with the data available
through EPILOG, provide you with a powerful management reporting tool. Because the data is statistically
reduced and graphically formatted, technically complex information can be presented in concise,
easy-to-read reports that communicate at a glance. For example, the reports in the starter kit allow you to
easily see such things as:
v How much CPU is being utilized by the different performance groups during peak and off-peak hours.
v How various channels are utilized throughout the day.
v What system resources contribute to TSO response time when it exceeds two seconds, and how much
they contribute.
v Why the last run of a particular batch job ran longer than the average run for that job.
Capacity planning and trending
Because you can clearly see the performance impact of constrained resources, you can identify the
resources that should be upgraded. Additionally, you can estimate how much improvement you can expect
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
201
from a particular upgrade. For example, if your performance group response time reports indicate that 30%
of the TSO response time is spent waiting for CPU, and tuning the IPS has not solved the problem, you
will probably want to arrange for an upgrade for your CPU. The trending reports can show you what the
pattern has been over the last few months, i.e., if the delay due to CPU waiting has increased or remained
constant. Because you can see what the trend has been in the recent past, you can make a reasonable
estimation of what it will be in the future. This information allows you to schedule the upgrade
appropriately.
Reporting overview
Processing takes place in two steps:
1. Degradation and resource data is written by the EPILOG collector to the EPILOG datastore.
Degradation profile records created with the Workload Profiling Facility are written to the Profile
datastore. Both datastores can be used as input for the SAS interface.
2. A SAS program that includes a procedure called KEPILGPM (for SAS Version 5) or KEPSEPIL (for
SAS Version 6.07) extracts the data for a given report from the datastores, and writes it to a SAS
database. The data is then available for SAS reporting using standard SAS programs.
The SAS database that is created by the procedure KEPILGPM is used as input to a SAS program. The
graphic output of the program can then be displayed on a color terminal or stored in a graphics replay file.
The contents of the file can subsequently be displayed online or printed on a color plotter.
Designing your own reports
The SAS interface writes historical data to SAS data sets that correspond to the resource and workload
display reports in EPILOG for MVS. You can use this data to design your own SAS reports. For further
information, see Chapter 13, “Report element tables,” on page 233.
Getting started
This section is designed for people who have never used the SAS graphical reporting feature. It shows
you how to supply parameters to the SAS programs that come with EPILOG, run the programs, and
display the results. There are two things you need to use this chapter:
1. A TSO terminal to display the graphs. In order to take full advantage of the SAS interface, you should
use a 3279 color terminal. If you do not have one available, you will have to run the non-graphic
versions of the report programs described in this chapter. See “Starter kit reports” on page 215 for
details on how to do this.
2. An EPILOG datastore to supply the data. The EPILOG for MVS collector should have been running for
at least a few days to gather enough data to generate graphs.
Let’s start with a brief overview of the product and what you have to do to produce reports.
Overview
The starter kit comes with two main components:
1. A set of SAS programs. Each of these programs produces a separate report, although a single report
may have more than one graph. For example, one of the reports we are going to run shows you two
things about the wait reasons that affect the response time of a given TSO performance group:
v A vertical bar chart that shows you how the average response time for each day was broken down
by summary wait reasons.
v A smooth trend of the response time over the month (or week). Different colored bands in the graph
show you how each wait reason contributed to the overall response time.
Each SAS program contains some user overrides and an imbedded procedure called KEPILGPM,
which you will use to supply parameters to the program. When these programs are run, output can be
202
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
directed to your terminal or to a graphics replay file. At first, we’ll just display the output on your
terminal. The instructions to run the last report show you how to store the graphs in a graphic replay
file.
2. A graphics replay procedure, which directs the contents of the graphics replay file to a terminal or to
another output device, such as a color plotter. If you are producing the reports interactively (using the
CLIST provided with the product), the graphs are displayed on the terminal you are working on. The
graphics replay procedure is used either to re-display graphs or to print out the graphs on another
output device.
Running reports
To run the reports, you need to do three things:
1. Allocate the graphics replay file. This step is only necessary if you are using the replay procedure. If
you are not interested in keeping the report output, the CLIST and batch job distributed with the
product dynamically allocate and free a SAS output file for you. A sample job is provided for you to
allocate the file.
The CLIST and batch job for SAS 5.18 are distributed in rhilev.REDDATA(KEPPMTSO) and
(KEPPMJCL), respectively, and rhilev.REDDATA(KEDPMTSO) and (KEDPMJCL) if DELTAMON is
installed at your site.
For SAS 6.07 and above, use the CLIST and batch job provided in rhilev.REDDATA(KPMTSO6) and
(KPMJCL6), respectively, for both EPILOG and DELTAMON.
2. Supply parameters to the SAS program. You do this by editing the members of the thilev.TEDSAS data
set, which is included on the product tape.
3. Run the SAS program. You can do this interactively (by using the CLIST provided with the product), or
in batch mode (by using the sample job provided with the product). In this chapter, we will be using the
CLIST to run the programs interactively.
That’s all there is to it. Since we are running the programs interactively, we don’t need to use the graphics
replay procedure to see the output displays. However, we’ll use it to re-display a graph later on just to
show you how it works.
Now, before we start on the first graph, you should allocate the graphics replay file.
Allocating the graphics replay file
The job to allocate a graphics replay file is contained in rhilev.REDDATA(KEPGALLC), and is shown
below.
// ... JOB CARD ...
//*
//* THIS JOB ALLOCATES THE 'GRAFBASE' FILE USED BY KEPPMTSO and
//* KEPPROFJ.
//*
//* TO RUN THIS JOB:
//*
1. ADD A JOB CARD
//*
2. CHANGE THE 'SPACE' PARAMETER TO REFLECT YOUR SPACE REQMNTS
//*
3. CHANGE 'VVVVVV' TO THE VOLSER YOU WISH TO ALLOC ON
//*
4. CHANGE 'ZZZZZZ' TO THE HIGH LEVEL INDEX OF THE DATASET TO
//*
BE USED AS GRAFBASE
//*
//ALLOC
EXEC PGM=IEFBR14
//GRAPH
DD SPACE=(UUU,(P,S)),
ENTER UNIT OF ALLOC, PRIM, SEC
//
UNIT=SYSDA,
//
VOL=SER=VVVVVV,
//
DISP=(,CATLG,DELETE),
//
DSN=ZZZZZZ.PRO.GRAFBASE,
//
DCB=(DSORG=DA,RECFM=U,LRECL=6156,BLKSIZE=6160)
Figure 127. Sample Job to Allocate a Graphics Replay File
Chapter 12. Reporting with SAS graphics
203
As you can see, you need to specify a space parameter, a volser, and a high-level index name. On
average, about four SAS graphs fit on one cylinder of 3380 disk space. This figure varies widely,
depending on the complexity of the graphs. We suggest that you start with about five cylinders; you can
adjust this figure later when you get a better feel for how much space you actually use.
Set the VOLSER to an appropriate volume serial number at your installation.
The high-level index name should be set to match your TSO user ID. (If no USERID parameter is
specified when using the CLIST, your TSO user ID is used as a default. The batch job always requires that
you enter the high-level qualifier.)
When you have set these parameters, run the job to allocate the file.
Now you must define the file to the CLIST that you will use to run the report. The CLISTs distributed with
the product are shown in “Sample CLIST and batch job for running the SAS interface” on page 214. Set
the GRAPH parameter in the CLIST to correspond to the name of the file you just allocated.
The Batch Job Trending and Forecasting report
The Batch Job Trending report was described in the beginning of the chapter, and produces three graphs.
To run the report, you must first modify some parameters in the SAS program. Edit
thilev.TEDSAS(KEPBJOBT). Copy this member into another member, so that the original remains
unchanged. The beginning of this program is shown below.
+-------------------------------------------------------------------+
|COPYRIGHT (C) 1985, AN UNPUBLISHED WORK BY CANDLE CORPORATION. ALL |
|RIGHTS RESERVED. THIS PROGRAM IS THE PROPERTY OF CANDLE CORPORATION|
|AND CONTAINS PROPRIETARY CONFIDENTIAL INFORMATION AND TRADE
|
|SECRETS. IT IS PROVIDED ONLY FOR INTERNAL USE UNDER LICENSE FROM |
|CANDLE CORPORATION. IT MAY NOT BE USED COPIED OR DISTRIBUTED EXCEPT|
|AS AUTHORIZED UNDER SUCH LICENSE.
|
+-------------------------------------------------------------------+
%INC PROMAC(KEPPROVE)
/ SOURCE2 ;
+------------------------------------------------------------------+
|
BATCH JOB ELAPSED TIME, TREND, AND FORECAST REPORTS
|
|
|
| DESCRIPTION:
|
|
THIS PROGRAM DEVELOPS THREE GRAPHS FOR A PARTICULAR BATCH JOB.|
|
THE FIRST GRAPH IS A VERTICAL BAR CHART. THERE IS ONE BAR FOR |
|
EACH EXECUTION OF THE JOB. EACH BAR IS COMPOSED OF COLOR
|
|
PATTERNS WHICH INDICATE THE CONTRIBUTION OF THE MAJOR WAIT
|
|
REASONS. THE HEIGHT OF THE BAR INDICATES THE JOB'S ELAPSED
|
|
RUN TIME. THE SECOND GRAPH IS A SMOOTH TREND OF THE SAME JOB. |
|
THE COLOR BANDS INDICATE THE GENERAL BEHAVIOR OF THAT WAIT
|
|
REASON'S CONTRIBUTION TO THE ELAPSED TIME. THE LAST GRAPH IS |
|
AN ATTEMPT TO FORECAST THE BEHAVIOR OF THE JOB FOR THE NEXT |
|
FEW EXECUTIONS OF THE JOB.
|
|
|
| REQUIRED EXTRACT KEYWORDS:
|
|
JOB(________)
|
|
|
| OPTIONAL SUGGESTED EXTRACT KEYWORDS:
|
|
{TIME SELECTIONS}
|
|
|
| NOTES:
|
|
SAS MUST BE EXECUTED USING THE MACRO OPTION.
|
|
|
|
THIS SAMPLE IS DESIGNED TO WORK WITH SAS VERSION 5.08
|
|
|
|
THIS SAMPLE IS DESIGNED TO USE SAS/GRAPH. IF SAS/GRAPH IS
|
|
NOT AVAILABLE, THIS PROGRAM WILL NOT RUN.
|
+------------------------------------------------------------------+
204
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+------------------------------------------------------------------+
|
MACROS
|
+------------------------------------------------------------------+
%INCLUDE
%INCLUDE
%INCLUDE
%INCLUDE
%INCLUDE
%INCLUDE
PROMAC(KEPCOLOR)
PROMAC(KEPBTCHM
PROMAC(KEPVERSI)
PROMAC(KEPCOLLS
PROMAC(KEPCPOLLC)
PROMAC(KEPDELGR)
/
/
/
/
/
/
SOURCE2
SOURCE2
SOURCE2
SOURCE2
SOURCE2
SOURCE2
;
;
;
;
;
;
+------------------------------------------------------------------+
|
USER OVERRIDES - CHANGE JJJJJJJ TO DESIRED JOBNAME
|
+------------------------------------------------------------------+
%LET
%LET
%LET
%LET
%LET
%LET
%LET
BATNAME=JJJJJJJJ ;
MAXJOB=30;
DDNAME=EPPRO ;
FORECAST=NO ;
LEAD=5 ;
DEVICE=IBM32793;
TERMINAL=TERMINAL ;
%KEPCOLOR (CRT )
%KEPVERSI
/* BATCH JOB NAME TO BE DISPLAYED
*/
/* MAXIMUM NUMBER OF JOBS TO PROCESS */
/* DDNAME USED FOR PROC KEPILGPM OUTPUT*/
/* FORECAST PLOT WANTED? 'YES' OR 'NO'*/
/* NUMBER OF FORECAST JOB RUNS
*/
/* GRAPHIC DEVICE DRIVER TO BE USED */
/*'TERMINAL'=SHOW GRAPHS IMMEDIATELY */
/*'NOTERMINAL'=DONT SHOW GRAPHS NOW */
/* 'HARD'=HARD COPY, 'CRT'=CRT SCREEN */
+------------------------------------------------------------------+
|
CALL PROC KEPILGPM - CHANGE JJJJJJJ TO DESIRED JOBNAME
|
+------------------------------------------------------------------+
PROC KEPILGPM DDNAME=&DDNAME ;
PARMCARDS ;
EXT JOB(JJJJJJJ)
;
(The SAS program to process the data extracted above follows here.)
Notice that the program contains six sections:
v A copyright section
v A description of the report, some EXTRACT keyword guidelines, and processing notes
v Macros used by the program
v User overrides
v The procedure KEPILGPM
v The body of the SAS code (not shown)
You pass parameters to the program through the user overrides and the EXTRACT command in the
procedure KEPILGPM. The first three sections are purely informational, and the body of the SAS code is
already set to run the standard report. First, set the user overrides:
BATNAME
Defines the batch job which is going to be the subject of the report. If the EPILOG
collector has been running on your system for more than a month, enter the name of a
batch job that is run several times a week. If the collector has been running for less than a
month, enter the name of a batch job that is run several times a day.
MAXJOB
Defines a limit on the number of batch jobs that are used as input to the report. Since the
first graph is a bar chart, and each bar represents a run of the job, if you try to report on
more than about 60 jobs, the bars become so thin that SAS has difficulty constructing the
graph. If there are more than 60 runs of the job during the time period, the SAS interface
selects the first 60 and ignores the rest. (You will set the time period in the EXTRACT
command.)
Chapter 12. Reporting with SAS graphics
205
DDNAME
Refers to the ddname defined in the CLIST or batch job that runs the program. This file
holds the output from the procedure KEPILGPM. The CLIST and batch job distributed with
the product use the ddname EPPRO.
FORECAST
If you have SAS ETS; software installed, you can run an additional forecasting section for
this report. To do so, set this override to YES. If you do not have SAS ETS, you must
leave this set to NO.
LEAD
If you are running the forecasting section of the report, this override allows you to set the
number of job runs to be forecasted. Generally, this should not exceed 10% of the number
of runs in the report.
DEVICE
Since you are directing the output to your 3279 terminal, leave this set to IBM32793.
TERMINAL
Leave this set to TERMINAL.
KEPCOLOR
Leave this set to CRT.
In the procedure KEPILGPM, you have to set some EXTRACT keywords. (The EXTRACT command is
described in “The EXTRACT command” on page 227. Notice that EXTRACT can be abbreviated to EXT).
This report requires that you enter the job name and perhaps a time period. (The time period is not
specifically required; it defaults to the first and last records in your EPILOG datastore. However, you
usually want to be more specific about the time period of this report.) Enter the same job name you
defined in the user override BATNAME. If you entered a job that is run weekly, leave the date specification
set at SDATE(-30). This sets the start of the time period to 30 days prior to the current date. (Since no end
date is specified, it defaults to the current date.) If you entered a job that is run several times a day, enter
LASTWEEK instead of SDATE(-30) as the time period.
Running the program
You are now ready to run the program. You can submit it in batch or run it interactively using the CLIST
provided with the product. Since you want to see the output on your terminal, you should run it using the
CLIST. When you execute the CLIST, include the PROG(nnn) parameter, where nnn = the name of
member into which you copied KEPBJOBT. For example, if you copied KEPBJOBT into a member called
MYCOPY, and you are using SAS 5.18, you would enter:
%KEPPMTSO PROG(MYCOPY)
If both EPILOG and DELTAMON are installed, and you are using SAS 5.18, enter:
%KEDPMTSO PROG(MYCOPY)
If you are using SAS 6.07 and above, enter:
%KPMTSO6 PROG(MYCOPY)
You will get a series of messages and page prompts (three asterisks) on the screen as the program
executes. When the first graph has been produced, it is displayed on your terminal. Press PF3 or PF15 to
continue processing and see the next graph.
The Graphics Replay File
When you finish looking at the graphs, they are written to the graphics replay file that you allocated earlier.
Each time you run a program, this file is overwritten with the output of the new program. If you want to
save a set of graphs, allocate another file and copy the graphics replay file into it. Later on, we’ll use the
graphics replay procedure to re-display graphs from the replay file. Right now, let’s run another program.
The Channel Utilization report
The Channel Utilization report shows the average channel utilization by hour over a specified period of
time. You can specify:
206
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
v The channel number you want to report on. If you select all channels, a separate graph is constructed
for each channel.
v The period of time to be averaged.
We follow essentially the same steps to run this report as before.
1. Copy the contents of thilev.TEDSAS(KEPRCHBH) to another member. The user overrides and the
procedure KEPILGPM sections follow.
+------------------------------------------------------------------+
|
USER OVERRIDES
|
+------------------------------------------------------------------+
%LET
%LET
%LET
%LET
%LET
CHANNEL=999 ;
THRESHLD=30 ;
DDNAME=EPPRO ;
DEVICE=IBM32793 ;
TERMINAL=TERMINAL ;
%KEPCOLOR (CRT )
/*
/*
/*
/*
/*
/*
/*
SPECIFY THE CHANNEL NUMBER (999=ALL)
SPECIFY THE WARNING LEVEL %BUSY
DDNAME FOR PROC KEPILGPM OUTPUT
GRAPHICS DEVICE DRIVER TO BE USED
'TERMINAL'=SHOW GRAPHS IMMEDIATELY
'NOTERMINAL'=DON'T SHOW GRAPHS NOW
'HARD'=HARD COPY, 'CRT'=CRT SCREEN
*/
*/
*/
*/
*/
*/
*/
+------------------------------------------------------------------+
|
PROC KEPILGPM CALL
|
+------------------------------------------------------------------+
/*
WE WILL AVERAGE ONLY WEEK DAYS
*/
PROC KEPILGPM DDNAME=&DDNAME ;
PARMCARDS ;
EXT RCHN SDATE(-30) DAY(WKDAY) CMB(1H)
;
2. Modify the user overrides and EXTRACT command keywords in the procedure KEPILGPM. As you
can see in the figure above, there are six user overrides:
CHANNEL
Specifies the channel you would like to report on. You can enter a single channel
number or 999 to see all channels.
THRESHLD
A horizontal line is drawn across the graph at the value specified as THRESHLD. You
should set this to whatever value you consider to be the maximum acceptable %BUSY
value for a channel.
DDNAME
Same as before; refers to the ddname defined in the CLIST or batch job that runs the
program. The CLIST and batch job distributed with the product use the ddname
EPPRO.
DEVICE
Same as before; since you are directing the output to your terminal, leave this set to
IBM32793.
TERMINAL
Same as before; leave this set to TERMINAL.
KEPCOLOR
Same as before; leave this set to CRT.
The procedure KEPILGPM contains an EXTRACT command which uses three types of keywords:
a. The RCHN keyword extracts information about channels. Since this report uses channel
information exclusively, this keyword should not be modified.
b. The SDATE and DAY keywords specify the time period over which the information is extracted, and
can be set to any time period you like. The SDATE(-30) expression specifies a start date 30 days
prior to the current date, and the DAY(WKDAY) expression limits the data to weekdays.
c. The CMB(1H) expression specifies that data should be combined into intervals of one hour.
Remember, the EPILOG collection interval is synchronized with the RMF interval at your
installation, and records are collected every RMF interval. The RMF interval is usually too short a
period for this kind of report. The COMBINE keyword (CMB for short) combines RMF-based
Chapter 12. Reporting with SAS graphics
207
intervals into longer intervals for reporting purposes. Since the SAS program is set to expect
intervals of one hour, this key should not be modified when the standard report is run.
3. After you have modified the user overrides and the procedure KEPILGPM, run the report interactively
using the CLIST. When you execute the CLIST, specify PROG(nnn), where nnn = the name of the
member you copied KEPRCHBH into.
4. Again, you will see messages and page prompts on your screen as the program executes. When the
first graph has been produced, it is displayed on the screen. If you specified CHANNEL=999 in the
user overrides, a graph is produced for each channel in your system. Press PF3 or PF15 to display the
next graph.
The Performance Group Response Time Degradation report
The Performance Group Response Time Degradation Report shows you the major wait reasons that
contributed to a specified performance group’s response time. It only looks at intervals in which the
response time was greater than a specified value (such as two seconds). For example, you might be
interested in what wait reasons were affecting performance group 2 during intervals in which the response
time was greater than two seconds. The report generates two pie charts:
1. The first pie chart shows you the major wait reasons that contributed to the response time of the
specified performance group. The total response time is represented as a pie, with each wait reason
comprising a piece of that pie. In the first pie chart, the wait reasons are actually summary categories
of more detailed wait reasons.
2. The second pie chart takes the largest wait reason from the first chart and breaks it down into detailed
wait reasons. The summary wait reason is represented as the pie, and the detailed wait reasons
comprise the pieces of the pie.
This time, we are going to run the report using the batch job and then use the graphic replay procedure
under TSO to display it on the terminal. Follow these steps:
1. Copy the thilev.TEDSAS(KEPPPGNP) member into another member. The user overrides and the
procedure KEPILGPM follow.
+------------------------------------------------------------------+
| USER OVERRIDES - CHANGE PGN TO DESIRED PGN NUMBER
|
+------------------------------------------------------------------+
%LET
%LET
%LET
%LET
%LET
SEC=2 ;
PGN=30;
DDNAME=EPPRO ;
DEVICE=IBM32793 ;
TERMINAL=TERMINAL
%KEPCOLOR (CRT )
%PERIOD (ALL)
/* SELECT INTERVALS RESPONSE TIME THRESHOLD*/
/* SELECT PERFORMANCE GROUP TO BE CHARTED */
/* SELECT DDNAME FOR PROC KEPILGPM OUTPUT */
/* SELECT GRAPHICS DEVICE DRIVER NAME
*/
;/*'TERMINAL'=SHOW GRAPHS IMMEDIATELY
*/
/*'NOTERMINAL'=DONT SHOW GRAPHS NOW
*/
/* HARD = HARD COPY, CRT = CRT SCREEN KEPCOLOR*/
/* ALL = ALL PERIODS, ONE = PERIOD ONE ONLY */
+------------------------------------------------------------------+
|
CALL PROC KEPILGPM - CHANGE PGN TO DESIRED PGN NUMBER
|
+------------------------------------------------------------------+
PROC KEPILGPM DDNAME=&DDNAME ;
PARMCARDS;
EXT PGN(30) LASTMONTH DAY(WKDAY) STIME(9) ETIME(17) SIF(RESP(>2S)) CMB
;
2. Set the user overrides as follows:
SEC
Defines the response time exception criteria; only intervals in which the response time
exceeded this value will be reported on.
PGN
Defines the performance group number to be reported on.
208
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
DDNAME
Same as before; refers to the ddname defined in the CLIST or batch job that runs the
program. The CLIST and batch job distributed with the product use the ddname
EPPRO.
DEVICE
Same as before; since you are directing the output to your terminal, leave this set to
IBM32793.
TERMINAL
Same as before; leave this set to TERMINAL.
KEPCOLOR
Same as before; leave this set to CRT.
3. Set the EXTRACT keywords in the procedure KEPILGPM. The EXTRACT command used in this report
uses four kinds of keywords:
a. The PGN keyword specifies the performance group number. It should specify the performance
group you set in the user overrides.
b. The time/date keywords (LASTMONTH, DAY, STIME, and ETIME) set the time period of the report.
LASTMONTH specifies that the report will use data from last month. DAY(WKDAY) specifies that
only data collected during weekdays will be used. STIME(9) and ETIME(17) specify that only data
collected between 9:00 AM and 5:00PM will be used. These keywords can be modified to specify
whatever time period you want.
c. The expression SIF(RESP(>2S)) means “select if response time is greater than two seconds.” SIF
is the short form of SELECTIF, and RESP is the short form of RESPONSE. This expression
determines the response time exception criteria. You can modify the value entered in the
expression (in this case, 2), but it must agree with the SEC value you entered in the user
overrides.
d. The CMB keyword is specified without an operand in this report. It therefore combines all
RMF-based intervals into one larger interval specified by the time/date keywords (in this case, one
month).
4. When you have modified the user overrides, submit the program in batch. You can use the sample
batch proc provided with the product for this. For example, if you copied the KEPPPGNP member into
a member called MYCOPY, and you are using SAS 5.18, submit the job as follows:
//JOB...
// EXEC KEPPMJCL, PROG=MYCOPY
5. When the job finishes, copy member thilev.TEDSAS(KEPGREPL) into another member of the same
data set and edit the following user overrides:
DEVICE
Set to IBM32793.
TERMINAL
Set to TERMINAL.
DDNAME
Set to the ddname you used when you allocated the graphics replay file at the
beginning of this session.
6. Execute the CLIST. This time, specify PROG(nnn) when you execute the CLIST, where nnn = the
name of the member into which you copied KEPGREPL.
7. A menu containing each graph generated by the report is displayed. Select the graphs you want to see
by placing an S (or any other letter) next to the graph name on the menu.
Using the starter kit
A starter kit of reports is included with EPILOG. Detailed descriptions of these reports are provided in
“Starter kit reports” on page 215. This chapter describes how to run these reports and direct output either
to an online terminal or to a color plotter. You produce reports from the starter kit in the following steps:
1. The CLIST and batch job provided with the SAS interface dynamically allocate and free a temporary
SAS file to hold the output of the SAS program.
For SAS 5.18 the CLIST and batch job are distributed in rhilev.REDDATA(KEPPMTSO) and
(KEPPMJCL), respectively, and rhilev.REDDATA(KEDPMTSO) and (KEDPMJCL) if DELTAMON is
installed at your site.
Chapter 12. Reporting with SAS graphics
209
For SAS 6.07 and above, use the CLIST and batch job provided in rhilev.REDDATA(KPMTSO6) and
(KPMJCL6), respectively, for both EPILOG and DELTAMON.
The SAS output is deleted after the program is run. If you wish to save the output, you must allocate a
permanent SAS file and then define it to the CLIST or batch job. To define it to the CLIST, include the
following parameter when you invoke the CLIST:
GRAPH(dataset name)
To define the output file to the batch job, simply modify the JCL to override the GRAPH DD card with
your data set name.
A sample job to allocate the permanent SAS output file is provided with the SAS interface. This is
described in “Allocating a permanent SAS file” on page 210.
2. Locate the PDS member name that contains the SAS program for the report you wish to produce. The
member name for each report is provided in the description of that report. You should copy this
member into another member so that the original remains unchanged.
3. Using an editor, look at the member into which you just copied the program. Each SAS program
contains a set of user overrides and an imbedded procedure called KEPILGPM. The user overrides
are used for SAS formatting, report titling, and, in some cases, data selection from the extracted data.
(This type of data selection is used only for consistency checking when the starter kit reports are run,
since each SAS program has it’s own KEPILGPM. When you design your own reports, however, you
may want to use one procedure KEPILGPM with several EXTRACT commands to extract data for
more than one report. Under these circumstances, you can use user overrides in the SAS programs to
select the type of data you want for the report.)
KEPILGPM contains one or more EXTRACT commands, which specify the types of data to be
extracted from the EPILOG datastore. The EXTRACT command syntax is described in “The EXTRACT
command” on page 227. You must modify the user overrides and KEPILGPM before running the
report. Guidelines for doing so are provided in the report descriptions in “Starter kit reports” on page
215 and in the internal program documentation.
4. Run the SAS program.
5. Run the graphic replay procedure (optional). If you allocated a permanent SAS output file and defined
it to the batch job or CLIST, the graphic output is written out to that file. A graphic replay procedure is
provided with the SAS interface so you can re-display the graphs on a color terminal or direct the
output to another device, such as a color plotter.
The following sections describe how to perform each of these steps:
v allocate a permanent SAS file
v modify the user overrides
v modify the procedure KEPILGPM
v run the graphic replay procedure
Allocating a permanent SAS file
A sample job to allocate a permanent SAS output file is provided in rhilev.REDDATA(KEPGALLC), and
follows.
210
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
// ... JOB CARD ...
//*
//* THIS JOB ALLOCATES THE 'GRAFBASE' FILE USED BY KEPPMTSO and
//* KEPPMJCL.
//*
//* TO RUN THIS JOB:
//*
1. ADD A JOB CARD
//*
2. CHANGE THE 'SPACE' PARAMETER TO REFLECT YOUR SPACE REQMNTS
//*
3. CHANGE 'VVVVVV' TO THE VOLSER YOU WISH TO ALLOC ON
//*
4. CHANGE 'ZZZZZZ' TO THE HIGH LEVEL INDEX OF THE DATASET TO
//*
BE USED AS GRAFBASE
//*
//ALLOC
EXEC PGM=IEFBR14
//GRAPH
DD SPACE=(UUU,(P,S)),
ENTER UNIT OF ALLOC, PRIM, SEC
//
UNIT=SYSDA,
//
VOL=SER=VVVVVV,
//
DISP=(,CATLG,DELETE),
//
DSN=ZZZZZZ.PRO.GRAFBASE
//
DCB=(DSORG=DA,RECFM=U,LRECL=6156,BLKSIZE=6160)
Figure 128. Sample Job to Allocate SAS File
As you can see, you need to specify a space parameter, a volser, and a high-level index name. On
average, about four SAS graphs fit on one cylinder of 3380 disk space. This figure varies widely,
depending on the complexity of the graphs. We suggest that you start with about five cylinders; you can
adjust this figure later when you get a better feel for how much space you actually use.
Set the VOLSER to an appropriate volume serial number at your installation.
The high-level index name should be set to match the the data set name that you define to the CLIST or
batch job used to run the programs. It is probably good practice to use your TSO user ID for this.
When you have set these parameters, run the job to allocate the file.
Modifying the user overrides
Each program in the starter kit contains a set of user overrides. A sample set of user overrides follows.
+------------------------------------------------------------------+
|
USER OVERRIDES - CHANGE JJJJJJJJJ TO DESIRED JOBNAME
|
+------------------------------------------------------------------+
%LET BATNAME=JJJJJJJJJ; /* SELECT THE JOB TO BE CHARTED
*/
%LET DEVICE=IBM32793 ;
/* SELECT GRAPHICS DEVICE DRIVER NAME
*/
%LET TERMINAL=TERMINAL ; /*'TERMINAL'=SHOW GRAPHS IMMEDIATELY
*/
/*'NOTERMINAL'=DON'T SHOW GRAPHS NOW
*/
%LET DDNAME=EPPRO ;
/* DDNAME OF KEPILGPM OUTPUT
*/
%KEPCOLOR (CRT )
/* HARD= HARD COPY, CRT= CRT SCREEN KEPCOLOR*/
These overrides control such things as:
v Setting SAS formatting parameters such as color specifications, output device characteristics, etc.
v Setting report titling specifications.
v Controlling data selection from the extracted data. This is useful when multiple EXTRACT commands
extract data records for several reports.
Modifying the procedure KEPILGPM
A sample procedure KEPILGPM follows.
Chapter 12. Reporting with SAS graphics
211
+------------------------------------------------------------------+
|
CALL PROC KEPILGPM - CHANGE JJJJJJJJ TO DESIRED JOBNAME
|
+------------------------------------------------------------------+
PROC KEPILGPM DDNAME=&DDNAME ;
PARMCARDS ;
EXT JOB(JJJJJJJJ)
;
The procedure KEPILGPM consists of four components:
1. The procedure KEPILGPM statement, which includes the ddname parameter. This ddname is set to
the SAS output file name defined in the user overrides. (&DDNAME does this automatically.)
2. The PARMCARDS line signals the beginning of the input cards for the procedure. This line must end
with a semicolon.
3. The input cards for the procedure are one or more EXTRACT commands. These commands perform
the actual data extraction from the EPILOG datastore. EXTRACT commands can not be written on the
same line as the PARMCARDS statement, nor can they be terminated with a semicolon. Guidelines for
entering EXTRACT commands are provided with each report description in Chapter 11, “Exporting
EPILOG data,” on page 193. A description of the EXTRACT command syntax appears in “The
EXTRACT command” on page 227.
4. The terminating semicolon for the input cards must be written on the line following the last input card.
The graphic replay procedure
Once you have run the SAS program specifying a permanent graphics output file and it has successfully
executed, you can use the graphic replay procedure to direct the output either to a color terminal or to
another output device, such as a color plotter. The SAS program to run the replay procedure is shown in
Figure 129 on page 213, and is distributed with the product in thilev.TEDSAS(KEPGREPL). You should
copy this member to another member so that the original remains unchanged. To replay the graphs on a
terminal, invoke the copy of KEPGREPL when you run the CLIST. To replay the graphs on a hardcopy
device, define the copy of KEPGREPL in the batch job.
212
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
+-------------------------------------------------------------------+
|COPYRIGHT (C) 1985, AN UNPUBLISHED WORK BY CANDLE CORPORATION. ALL |
|RIGHTS RESERVED. THIS PROGRAM IS THE PROPERTY OF CANDLE CORPORATION|
|AND CONTAINS PROPRIETARY CONFIDENTIAL INFORMATION AND TRADE
|
|SECRETS. IT IS PROVIDED ONLY FOR INTERNAL USE UNDER LICENSE FROM |
|CANDLE CORPORATION. IT MAY NOT BE USED COPIED OR DISTRIBUTED EXCEPT|
|AS AUTHORIZED UNDER SUCH LICENSE.
|
+-------------------------------------------------------------------+
%INC PROMAC(KEPPROVE) / SOURCE2 ;
+------------------------------------------------------------------+
|
REPLAY STORED GRAPHS
|
|
|
| THIS PROCEDURE MAY BE USED TO REPLAY GRAPHIC CHARTS FOR EITHER |
| A CRT DEVICE OR PLOTTER. FOR BEST RESULTS THE REPLAY DEVICE AND |
| THE DEVICE SPECIFIED IN THE GRAPH CREATION SHOULD BE THE SAME. |
|
|
| NOTE :
|
|
SAS MUST BE EXECUTED USING THE MACRO OPTION.
|
|
|
|
THIS PROGRAM IS DESIGNED TO WORK WITH SAS VERSION 5.08
|
|
AS WELL AS SAS VERSION 82.4
|
+------------------------------------------------------------------+
+------------------------------------------------------------------+
|
MACROS
|
+------------------------------------------------------------------+
%INCLUDE PROMAC(KEPVERSI) / SOURCE2 ;
%INCLUDE PROMAC(KEPMREPL) / SOURCE2 ;
+------------------------------------------------------------------+
|
USER OVERRIDES
|
+------------------------------------------------------------------+
%LET DEVICE=IBM32793;
/* SPECIFY THE REPLAY DEVICE DRIVER
*/
%LET TERMINAL=TERMINAL ;
/* 'TERMINAL'=SHOW GRAPHS IMMEDIATELY */
/* 'NOTERMINAL'=DONT SHOW GRAPHS NOW
*/
%LET DDNAME=GRAPH;
/* DDNAME OF STORED GRAPHIC DISPLAY FILE*/
%LET DSNAME=JOBDATA ;
/* SAS DATA SET NAME OF GRAPH FILE
*/
+------------------------------------------------------------------+
|
KEPGREPL
|
+------------------------------------------------------------------+
%KEPVERSI
/* DETERMINE SAS VERSION BEING EXECUTED */
RUN;
GOPTIONS DEVICE=&DEVICE &TERMINAL ;
OPTIONS &TEXT82 DQUOTE ;
%KEPMREPL
/* RUN THE CORRECT KEPGREPL FOR SAS VERSION*/
ENDSAS;
Figure 129. The graphic replay procedure
There are four user overrides in the replay procedure:
DEVICE
The output device identifier. This should correspond to the identifier entered as the
DEVICE override in the SAS program. If different identifiers are entered, incompatibilities
may cause such problems as visual distortion or abnormal session termination.
TERMINAL
If a terminal was specified as the output device identifier, enter TERMINAL in this override.
If another identifier was specified, enter NOTERMINAL as the override.
DDNAME
The graphic replay data set name, as defined by the ddname of the JCL used to run the
procedure. If you are running reports from the starter kit, you should leave the DDNAME
override set to GRAPH.
DSNAME
The second word of the data set name used during graph generation. In the starter kit
programs, this is set to JOBDATA.
After you set the overrides, run the appropriate CLIST or batch job. If you replay the output to a hardcopy
device, the output will be printed when the program completes. If you replay the output to a terminal, run
Chapter 12. Reporting with SAS graphics
213
the CLIST with the name of your copy of KEPGREPL as the PROG parameter. For example, if you copied
KEPGREPL into a member named MYREPLAY, enter:
%KEPPMTSO PROG(MYREPLAY)
Macros
Nine macros were used to write the programs for the starter kit reports. You may want to use or modify
these macros when you write your own programs.
Table 20. User Macros
Macro
Definition
KEPBTCHM
Used to condense and simplify data in the batch job trending report. The macro groups several
related wait reasons into a summary group. The summary group represents a major wait reason.
KEPCOLLS,
KEPCOLLC
When a starter-kit program generates more than one graph with SAS 82.4, KEPCOLLS starts the
process of collecting them from a temporary data set by writing the first graph to a permanent data
set (GRAPH.JOBDATA). KEPCOLLC continues the process by writing the subsequent graphs to
the same permanent data set.
KEPCOLOR
Sets eight global macro variables to define the color scheme on the output device. The version that
is distributed with the product has an input variable that can be set to CRT or HARD. CRT sets the
color scheme for an IBM 3279 terminal. HARD sets the color scheme for a color plotter. You should
modify the color settings for the HARD option to correspond with your hard copy output device, if
necessary.
KEPDELGR
When a starter-kit program generates a new graph(s) with SAS 5.08, this macro deletes the old
version(s) from the graphics replay file.
KEPMREPL
Conditionally executes the SAS 82.4 or the SAS 5.08 version of the KEPGREPL procedure. It is
used only in the starter-kit program KEPGREPL.
KEPPGNMA
Used to condense and simplify data in the performance group trending and response time
degradation reports. The macro groups several related wait reasons into a summary group that
represents a major wait reason.
KEPSUMMG
Simplifies defining a summary report group. In the starter-kit programs KEPRPGNC and
KEPRPGNS, you can use this option to combine the data for several performance groups and
define them as a single reporting group, such as TSO, CICS, BATCH, and the like. You must first
modify this macro to correspond with performance group definitions at your installation.
KEPVERSI
Uses the SAS automatic variable &SYSVER to determine which version of SAS is being executed.
It then modifies release-dependent macro variables accordingly.
SAS considerations
All starter kit programs are written in SAS and are subject to SAS language requirements. Statement
syntax rules, procedure parameters, and the like must conform to the rules of the SAS level you are using.
Sample CLIST and batch job for running the SAS interface
A sample CLIST and batch job are provided with the product to run the SAS interface online and in batch
mode. Both are included in the rhilev.REDDATA data set. If you are using SAS 5.18, the CLIST is in
member KEPPMTSO and the batch job is in KEPPMJCL. If both EPILOG and DELTAMON are installed,
and you are using SAS 5.18, the CLIST is in member KEDPMTSO and the batch job is in KEDPMJCL.
For SAS 6.07 and above, the CLIST is in member KPMTSO6 and the batch job is in KPMJCL6 for both
EPILOG and DELTAMON.
Both the sample CLIST and the sample batch job invoke the reporter to process the EXTRACT command,
and require that the EPILOG datastore has been allocated. If you also have DELTAMON for MVS
installed, the sample CLIST also allocates the DELTAMON datastore because of the reporter requirement,
even though DELTAMON data is not used in SAS reporting.
214
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Like the CLIST, the batch job invokes the reporter to process the EXTRACT command, and requires that
the EPILOG datastore be allocated. If you also have DELTAMON installed, the sample CLIST also
allocates the DELTAMON datastore because of the reporter requirement, even though DELTAMON data is
not used in SAS reporting.
Starter kit reports
This section describes the starter kit of SAS reports that comes with EPILOG for MVS. Each report
description includes:
v The names of the PDS members that contain the programs to run the report. Most reports can be
produced in both graphic and standard SYSOUT form. The first member name listed in the report
description contains the program that produces the graphic form of the report. The second member
name contains the program that produces the standard SYSOUT form of the report. (Members that
contain standard SYSOUT versions of reports have an L (for listing) as the second letter of the member
name.)
v A report description.
v Required EXTRACT command keywords.
v Recommended EXTRACT command keywords.
The reports are grouped into four categories:
1. Reports in the first set describe system resource utilization. You will notice that the PDS member
names that contain programs to produce resource reports begin with an R.
2. Reports in the next set are performance group reports. These member names begin with a P.
3. Reports in the next set are batch job reports. These member names begin with a B.
4. Reports in the final set are non-graphic and use standard SYSOUT.
User overrides
Each of these programs contains user overrides used for SAS formatting and titling. The complete list of
overrides is described below. The overrides used by individual reports are identified in the report
descriptions. In most cases, these overrides are SAS global macro variables.
BATNAME
The name of the batch job.
CHANNEL
Specifies the channel number for the channel utilization and I/O rate reports. To produce
graphs for all channels in your system, enter 999.
KEPCOLOR
Sets the color formatting for the output device. If the output device is a 3279 terminal,
enter CRT. If you are using another output device to produce hard copy, enter HARD.
(This override is actually a macro call. You can supply your own version of the
KEPCOLOR macro to match your hardware configuration.)
DDNAME
Defines the output SAS data library name specified in the CLIST or batch job that runs the
report.
DEVICE
Defines the name of the output graphic device driver. If you are directing the output to a
3279 terminal, enter IBM32793.
FORECAST
If you have SAS ETS software installed, you can run additional forecasting sections for the
KEPBJOBT and KEPPPGNT reports. To do so, set this override to YES. If you do not
have SAS ETS, you must leave this set to NO.
LEAD
If you are running the forecasting section of the KEPBJOBT or KEPPPGNT report, LEAD
allows you to set the number of job runs to be forecasted. Generally, this should not
exceed 10% of the number of runs in the report.
MAXJOB
Defines a limit on the number of batch jobs that are used as input to the batch job
trending report. The first graph of the report is a bar chart, and each bar represents a run
Chapter 12. Reporting with SAS graphics
215
of the job. If you try to report on more than about 60 runs, the bars become so thin that
SAS has difficulty constructing the graph. If there are more than 60 runs of the job during
the time period, the SAS interface selects the first 60 and ignores the rest. If you wish to
limit the graph to fewer than 60 runs of the job, lower this override accordingly.
PGN
Defines the performance group number in performance group reports.
SEC
Sets the exception criteria for the report title for the performance group trending report. For
example, if this override is set to two, only intervals in which the response time exceeded
two seconds are expected in the report. If you are including all intervals, set this override
to zero (0).
TERMINAL
When reports are run using the CLIST provided with the SAS interface, by default the
graphic output is displayed on the screen. You can override this default by specifying
NOTERMINAL. (This override is ignored when the report is run in batch.)
THRESHLD
Used in the channel utilization report. If you specify a threshold, a dotted line is drawn
across the graph at that level. In the sample report, the threshold was set to 30%.
TRAP
Sets the trapezoidal aspect of the graph. This override is used with graphs that
occasionally experience SAS area fill routine failures. If this error occurs, increase the
TRAP override from the default value of 0.1 by 0.05 increments until the error is corrected.
Do not set TRAP higher than 0.5.
Channel Utilization report
PDS Member Name: thilev.TEDSAS(KEPRCHBH) or (KEPRLCBH)
Description
This report shows the average percent utilization of physical channels (for MVS/370) or CHPIDs (for
MVS/XA) for each hour of the day. The command syntax allows you to report on a single channel or on all
channels in your system. If you select all channels, a separate graph is produced for each channel.
User overrides
CHANNEL
THRESHLD
DDNAME
DEVICE
TERMINAL
KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v RCHN (To extract data about channel resources)
v CMB(1H) (To combine information into hourly intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Channel I/O Rate report
PDS Member Name: thilev.TEDSAS(KEPRCHIH) or (KEPRLCIH)
Description
This report highlights periods of peak channel activity by showing the average number of I/Os per second
for each hour of the day. You can report on a single channel or on all channels in your system. If you
select all channels, a separate graph is generated for each channel.
216
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
This report applies to MVS/370 systems only.
User overrides
CHANNEL
DDNAME
DEVICE
TERMINAL
KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v RCHN (To extract information about channel resources)
v CMB(1H) (To combine information into hourly intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Channel Peak Utilization report
PDS Member Name: thilev.TEDSAS(KEPRCHBP) or (KEPRLCBP)
Description
This report shows the peak utilization for physical channels (MVS/370) or CHPIDs (MVS/XA) for each day
in the specified time period. The average utilization for each hour during the day is examined, and the
highest of these figures is selected as the peak utilization. Each bar on the SAS chart represents the peak
hour of a particular day, and the height of the bar represents the percent utilization during that hour. You
can specify a particular channel, or all channels in your system. If you select all channels, a separate
graph is generated for each channel.
User overrides
CHANNEL
THRESHLD
DDNAME
DEVICE
TERMINAL
KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v RCHN (To extract information about channel resources)
v CMB(1H) (To combine information into hourly intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Channel Peak I/O Rate report
PDS Member Name: thilev.TEDSAS(KEPRCHIP) or (KEPRLCIP)
Description
This report shows the peak I/O rate for each day in the specified time period. The average I/O rate for
each hour during the day is examined, and the highest of these figures is selected as the peak I/O rate.
Each bar on the SAS bar chart represents the peak hour of a particular day, and the height of the bar
Chapter 12. Reporting with SAS graphics
217
represents the I/O rate during that hour. You can specify a particular channel, or all channels in your
system. If you select all channels, a separate graph is generated for each channel.
This report applies to MVS/370 systems only.
User overrides
CHANNEL
DDNAME
DEVICE
TERMINAL
KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v RCHN (To extract information about channel resources)
v CMB(1H) (To combine information into hourly intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Demand Paging Activity report
PDS Member Name: thilev.TEDSAS(KEPRPAGP) or (KEPRLPAP)
Description
This report shows average demand paging activity for each hour of the day. Each point on the
two-dimensional SAS graph represents the average demand paging rate for that hour. Separate plots are
shown for page-in and page-out rates.
User overrides
DDNAME
DEVICE
TERMINAL
KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v RPAG (To extract information about page data sets)
v COMB(1H) (To combine information into hourly intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Swap Paging Activity report
PDS Member Name: thilev.TEDSAS(KEPRPAGS) or (KEPRLPAS)
Description
This report shows average swap paging activity for each hour of the day. Each point on the
two-dimensional SAS graph represents the average swap paging rate for that hour. Separate plots are
shown for page-in and page-out rates.
218
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
User overrides
DDNAME
DEVICE
TERMINAL
KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v RPAG (To extract information about page data sets)
v CMB(1H) (To combine information into hourly intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
CPU Utilization by Performance Group report
PDS Member Name: thilev.TEDSAS(KEPRPGNC)
Description
This report shows you the total CPU utilization broken down by performance groups, and distributed over
the hours of the day. If you have a large number of performance groups at your installation, you can
combine performance groups into summary groups to simplify the graphic output. A model for this is
included in the source, and can be found by locating the word COMMENT.
There is no standard SYSOUT version of this report.
User overrides
DDNAME
TRAP
DEVICE
TERMINAL
KEPCOLOR
Required EXTRACT keywords
The program requires separate EXTRACT commands for each of the 24 hourly intervals. Each command
must include:
v RPGN (To extract resource information about performance groups)
v RCPU (To calculate the difference between total CPU used and CPU used by performance groups. This
difference is plotted as overhead.)
v STARTTIME and ENDTIME (To define the hourly intervals, set these an hour apart)
v CMB (To combine the data into hourly intervals. This keyword should be entered without an operand.)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Real Storage Utilization by Performance Group report
PDS Member Name: thilev.TEDSAS(KEPRPGNS)
Description
This report shows you the total real storage utilization broken down by performance groups, and
distributed over the hours of the day. If you have a large number of performance groups at your
installation, you can combine performance groups into summary groups to simplify the graphic output. A
model for doing this is included in the source, and can be found by locating the word COMMENT.
Chapter 12. Reporting with SAS graphics
219
A horizontal line is automatically plotted to indicate the maximum storage available for private working
sets.
There is no standard SYSOUT version of this report.
User overrides
DDNAME
TRAP
DEVICE
TERMINAL
KEPCOLOR
Required EXTRACT keywords
The program requires separate EXTRACT commands for each of the 24 hourly intervals. Each command
must include:
v RPGN (To extract resource information about performance groups)
v RPAG (To calculate the maximum storage available)
v STARTTIME and ENDTIME (To define hourly intervals, set an hour apart.)
v CMB(1H) (To combine the data into hourly intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Performance Group Response Time Degradation report
PDS Member Name: thilev.TEDSAS(KEPPPGNP) or (KEPPLPGP)
Description
This report analyzes response time degradation for a performance group by showing the main wait
reasons that contributed to the response time. The report produces two pie charts:
1. The first pie chart shows the major wait reasons that contributed to the response time of the
performance group. The total response time is represented as a pie, with each wait reason comprising
a piece of that pie. The wait reasons in this chart are summary categories of more detailed wait
reasons.
2. The second pie chart breaks down the largest wait reason from the first chart into detailed wait
reasons.
User overrides
SEC DDNAME TERMINAL
PGN DEVICE KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v PGN(nn) (To define the performance group number. The number entered in this keyword must agree
with the performance group number specified in the user overrides.)
v SIF(RESP(>nS)) (To define the exception criteria. The value entered in this keyword must agree with
the SEC user override.)
v CMB (To combine the data using the time/date keywords as the interval bounds. This keyword should
be entered without an operand.)
Recommended EXTRACT keywords
v PGP (To specify one or more individual performance group periods)
v Date and day of week selections.
220
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Performance Group Trending and Forecasting report
PDS Member Name: thilev.TEDSAS(KEPPPGNT) or (KEPPLPGT)
Description
This report examines the major wait reasons that degrade response time for a specified performance
group. By default, the report generates two graphs:
v A vertical bar chart that shows you how the average response time for each day was broken down by
summary wait reasons.
v A smooth trend of the response time over the selected time period. Different colored bands in the graph
show you how each wait reason contributed to the overall response time.
If you have SAS ETS software, you can generate a third graph that shows a forecast. To do so, set the
FORECAST override to YES. If you do not have ETS, you cannot generate the graph and must leave
FORECAST set to NO.
The standard SYSOUT version generates only the first bar chart.
User overrides
PGN DEVICE
LEAD TERMINAL
FORECAST KEPCOLOR
DDNAME
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v PGN(nn) (To define the performance group number. The number entered in this keyword must agree
with the performance group number specified in the user overrides.)
v CMB (To combine the data into daily groups)
Recommended EXTRACT keywords
v PGP(nn) (To specify one or more individual performance group periods.)
v Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week. Also, you
may want to use STARTTIME and ENDTIME keywords to select a period of interest, such as prime shift
or peak hours. If you do so, you must enter a corresponding operand with the COMBINE keyword. For
example, if you specify STIME(9) and ETIME(17), you should set CMB to (8H).
Because the performance group selection keywords interact with the user overrides, you must be careful
when using them. Here are some guidelines:
1. You can use the SYMBOLIC keyword in the EXTRACT command, but the performance group number
must be specified in the user overrides. If you use the SYMBOLIC keyword, make sure that the
symbolic name you enter corresponds to the performance group number you specify in the user
overrides.
2. You should not specify multiple performance groups with the PGN keyword.
Batch Job Comparison report
PDS Member Name: thilev.TEDSAS(KEPBJOBC) or (KEPBLJBC)
Description
This report compares a profile of the last run of a batch job with a profile of an average run of that job.
Each profile consists of a breakdown of the main wait reasons that contributed to the run time of the job.
The profile of the average run is obtained by examining all runs of the job over a specified period of time.
The output of the report is a bar graph, and the bands in the bars represent summary wait reasons.
Chapter 12. Reporting with SAS graphics
221
Note: This program uses only information from the EPILOG datastore. A more comprehensive form of
profile comparisons using the Workload Profiling Facility is discussed in Chapter 10, “Workload
Profiling Facility (WPF),” on page 167.
User overrides
BATNAME
DEVICE
TERMINAL
DDNAME
KEPCOLOR
The DEVICE, TERMINAL, and KEPCOLOR overrides apply to the graphic version only.
Required EXTRACT keywords
v JOB(JJJ) (To specify the name of the batch job)
Recommended EXTRACT keywords
You determine the average profile of the job by including date/time keywords. If you select a time period
when the job was running well, then the graphs compare the last run of the job to the target or ideal run of
the job. This would require a second EXTRACT command to obtain data about the last run of the job. The
program always selects the most recent run as the last run, and all other runs are averaged into the
average run. If you use one EXTRACT command to extract information about an earlier good period, and
a second EXTRACT for the later bad run, the bad run is compared to earlier good runs.
Batch Job Trending and Forecasting report
PDS Member Name: thilev.TEDSAS(KEPBJOBT) or (KEPBLJBT)
Description
This report examines the major wait reasons that affect the run time for a given batch job. By default, the
report generates two graphs:
v A vertical bar chart that shows how the average run time for each run of the job was broken down by
summary wait reasons.
v A smooth trend of the run time over the selected time period. Different colored bands in the graph show
how each wait reason contributed to overall run time.
If you have SAS/ETS software, you can generate a third graph that shows a forecast. To do so, set the
FORECAST override to YES. If you do not have ETS, you cannot generate the graph and must leave
FORECAST set to NO.
The standard SYSOUT version generates only the first bar chart.
User overrides
BATNAME
MAXJOB
DDNAME
FORECAST
LEAD
DEVICE
TERMINAL
KEPCOLOR
LEAD, DEVICE, TERMINAL, and KEPCOLOR apply to the graphic version only.
Required EXTRACT keywords
v JOB(JJJ) (To specify the name of the batch job)
222
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Channel Service Rate report
PDS Member Name: thilev.TEDSAS(KEPRCHNS)
Description
This report produces a schematic plot of the channel service time (in milliseconds per I/O) versus average
I/O rate. It can be run for a single channel, or for all channels in the system. The “box and whisker” graph
is produced by the SPLOT procedure, which is described in the SAS SUGI Supplemental User’s Guide.
The output is standard SYSOUT, and should be output to a printer.
User overrides
CHANNEL
DDNAME
Required EXTRACT keywords
v RCHN (To extract information about channel resources)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Chapter 12. Reporting with SAS graphics
223
S
C
H
E
M
A
T
I
C
P
L
O
T
S
F
O
R
V
A
R
I
A
B
L
E
C
S
R
V
T
I
M
E
CHANNEL AVERAGE I/O SERVICE TIME VS. NUMBER OF I/O-S PER SECOND
CHANNEL 999 AVERAGED FOR DAYS 11MAR99 TO 15MAR99
8:02 FRIDAY, MARCH 22, 1999
CHANNEL NUMBER=1
41.0
+---------------------------------------------------------+
|
*
|
|
|
|
|
|
|
|
|
|
*
|
32.7
+
+
|
|
|
|
|
|
|
*
|
|
|
|
|
|
|
|
|
24.5
+
+
|
|
|
|
|
|
|
*
|
|
|
|
|
|
|
|
|
16.3
+
+
|
|
|
|
|
|
|
*
|
|
*
*
|
|
*
|
|
*
*
|
8.17
+
0
+
|
0
|
*
|
|
|
|
|
|
|
+---+
|
|
+---+
*
|
|
+
| + |
+
|
|
|
|
|
|
+---+
*---*
+---+
*-+-*
*-+-*
*-+-*
|
|
*---*
+---+
*---*
+---+
|
|
|
|
+---+
|
|
|
*
*
|
|
|
*
*
|
.0
+-----------|---------------------------------------------+
CIOPRSEC
5.00
10.0
15.0
20.0
25.0
30.0
Figure 130. The Channel Service Rate report
224
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
DASD AVERAGE BUSIEST HOUR
7:30 FRIDAY,
OBS PERCENT DEV VOLUME
BUSY
ADR NAME
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
44.5500
42.5000
40.5200
39.7333
39.5600
39.3800
38.4600
29.6000
28.8333
28.6333
24.6800
21.5200
19.8600
18.3200
16.6400
15.6500
15.4400
13.0600
12.4400
11.8200
11.0000
10.4400
10.1400
9.8500
9.2200
6.7000
6.6000
6.3600
6.2400
5.6200
5.1000
5.0000
4.6600
4.6500
3.5500
3.1000
142
333
337
140
962
323
960
325
155
143
145
963
154
146
157
326
151
150
141
153
322
751
324
336
152
335
754
331
144
961
334
964
330
321
327
967
CIMS03
SYS025
SYS026
CIMS01
SYS024
HSM001
SYS027
MSA031
CIMS02
CIMS05
MVS134
SYS023
CAN031
SPOOL1
SYS021
CIMS04
TSO022
MVSA21
TSO021
OMON22
OMON24
WORK22
SYS022
OMON23
OMON25
TSO025
WORK21
TSO023
OMON26
TSO024
OMON21
OMON27
NEWRES
MSA033
MSA032
CIMS07
FOR DAYS 11MAR99 TO
MARCH 22, 1999
BSY NUMBER
TOT I/O
HR
I/O PER TIME
SECOND
(MSEC)
23
21.8500 26.5000
14
11.3800 42.8000
11
6.3800 67.4000
1
22.8333 12.0000
11
5.9000 66.2000
11
5.7600 67.4000
11
5.6800 66.4000
20
15.3500 22.0000
0
10.9333 25.3333
0
9.8333 31.6667
11
6.1000 65.4000
10
7.1400 36.4000
12
6.8000 27.8000
10
6.1400 40.8000
11
8.0400 35.2000
23
7.4500 35.0000
11
6.3200 28.2000
11
8.4800 22.2000
11
5.6600 25.6000
15
4.2800 33.6000
5.2000 38.3333
17
2.9800 33.4000
10
3.6600 31.2000
21
5.6500 19.5000
10
3.5400 30.0000
16
2.6200 28.6000
1
0.8333 63.6667
10
2.8200 24.8000
11
1.7800 41.0000
10
1.9200 35.2000
23
0.5000 97.5000
23
0.4000 65.0000
16
2.2600 17.4000
20
1.9500 25.0000
19
1.5500 23.5000
2
1.2667 23.0000
18MAR99
AVG #
OPEN
DSNS
0.000
0.000
0.000
3.167
0.040
1.000
0.000
0.550
4.200
1.100
5.780
0.080
2.020
32.660
44.980
0.000
21.840
0.000
45.880
3.220
0.267
1.980
41.080
3.350
39.940
24.740
1.367
26.580
27.900
114.860
0.100
4.150
1.020
0.150
2.200
0.833
Figure 131. The DASD Utilization report
DASD Utilization report
PDS Member Name: thilev.TEDSAS(KEPRDASD)
Description
This report lists DASD devices in decreasing order of sustained peak utilization. The sustained peak
utilization is determined in the following way:
1. Data is extracted for the DASD devices during the time period you specify in the EXTRACT command,
combined at 1 hour intervals.
2. For each device, an average hourly busy is calculated for each hour of the day over the days in the
time period.
3. The program selects the averaged hour that had the highest percent busy value for each device. This
is the sustained peak utilization.
4. The program sorts the devices in descending order of percent busy values.
The output is standard SYSOUT, and should be output to a printer.
Chapter 12. Reporting with SAS graphics
225
User overrides
DDNAME
Required EXTRACT keywords
v RDAS (To extract information about DASD devices)
v CMB(1H) (To combine the data into 1 hour intervals)
Recommended EXTRACT keywords
Date and day of week selections. The date/time keywords select the time period of the report. You may
want to use the DAYOFWK keyword to select only weekdays or certain days of the week.
Batch Program Resource Consumption report
PDS Member Name: thilev.TEDSAS(KEPBPGML)
Description
This report describes the resources consumed by your batch programs. The report is produced in three
parts:
1. The first part of the report lists batch programs that ran during the specified time period by frequency
of execution. Programs are listed in decreasing order, with the most frequently run program at the top
of the list.
2. The second part of the report lists programs in decreasing order by CPU consumption. The program
that used the most aggregate CPU time appears at the top of the list.
3. The third part of the report lists programs in decreasing order by active I/O time. The program that
used the most active I/O time appears at the top of the list.
The output is standard SYSOUT, and should be output to a printer.
User overrides
DDNAME
Required EXTRACT keywords
v PGM(*) (To extract data for all batch programs)
Recommended EXTRACT keywords
To avoid extracting very large quantities of data, the time/date keywords can be used to limit the extraction
to a day or two.
226
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
OBS
1
2
3
4
5
6
OBS
1
2
3
4
5
6
PROGRAM ACTIVITY REPORT BY NUMBER OF PROGRAM EXECUTIONS
7:01 FRIDAY, MARCH 22, 1999
1
NUMBER
PROGRAM
AVERAGE
AVERAGE
OF EXEC.
NAME
I/O SEC.
CPU SEC.
94
41
39
22
20
19
ENQOBJ
IKJEFT01
SASLPA
IEBGENER
REFORM
CFRS1
30.04
23.80
31.75
1.97
4.39
14.86
AVERAGE
ELAPSED
SEC.
270.14
77.77
134.67
106.40
495.90
51.98
9.297
3.297
5.846
0.450
118.800
1.305
PROGRAM ACTIVITY REPORT BY TOTAL CPU TIME USED
7:01 FRIDAY, MARCH 22, 1999
1
TOTAL
PROGRAM
NUMBER
AVERAGE
CPU SEC.
NAME
OF EXEC.
I/O SEC.
2376.00
1142.13
1018.97
873.93
478.79
228.00
REFORM
BCMRPT
IDCAMS
ENQOBJ
SORT
SASLPA
20
2
18
94
12
39
AVERAGE
ELAPSED
SEC.
495.90
876.24
465.12
270.14
178.85
134.67
4.39
50.05
134.79
30.04
110.58
31.75
PROGRAM ACTIVITY REPORT BY TOTAL I/O ACTIVE TIME USED
7:01 FRIDAY, MARCH 22, 1999
1
OBS
TOTAL
PROGRAM
NUMBER
AVERAGE
I/O SEC.
NAME
OF EXEC.
CPU SEC.
1
2
3
4
5
6
2823.74
2536.79
2426.19
1759.05
1695.88
1373.60
ENQOBJ
FDRDSF
IDCAMS
DFHFTAP
BCMDMSG
BCMSAXGN
94
6
18
1
1
1
AVERAGE
ELAPSED
SEC.
270.14
856.98
465.12
8143.77
2527.39
1795.55
9.297
9.407
56.609
32.580
187.030
168.780
Figure 132. The Batch Program Resource Consumption report
The EXTRACT command
The EXTRACT command selects data from the EPILOG datastore and converts it into a format that is
compatible with SAS processing. The data is written out to SAS data sets, which are described in
Chapter 13, “Report element tables,” on page 233. The EXTRACT command is very similar to the
DISPLAY command, which is described in theEPILOG for MVS Basic User’s Guide.
Figure 133 on page 228 shows the syntax of the EXTRACT command.
Chapter 12. Reporting with SAS graphics
227
EXTRACT
{workload | resource} [time period] [combining factor] [SYSID(cccc, ...)] ┌
┐
│ REPORTIF (filter) │
│
or
│
│ SELECTIF (filter) │
└
┘
Figure 133. EXTRACT command syntax
As you can see in Figure 16 on page 26, the only required keywords are those that specify a workload or
resource. However, you will generally want to include a time period in the command. If you omit the time
period, the reporter displays the entire span of data from the EPILOG datastore, that is, from the time
when collection first began to the time when the last interval was collected. This usually results in long
displays that do not focus on useful information.
The keyword types used with the EXTRACT command are summarized in Table 2 on page 26.
Table 21. EXTRACT command keyword summary
Keyword Type
Description
workload
The workload keywords display degradation data for a performance group, batch job,
started task, or TSO user session. These keywords are described in “Workload keywords”
on page 171.
resource
The resource keywords display information about any of the thirteen types of system
resources that are monitored by EPILOG. These keywords are described in “Using
resource data” on page 233.
time period
Since the EPILOG datastore is a historical database, you can select data from any time
period that the EPILOG collector was running. The date/time keywords for doing so are
described in “Date and time keywords” on page 171.
combining factor
Typically, resource and performance group collection is synchronized with RMF intervals at
your installation. Thus, by default, each display panel usually corresponds to an RMF
interval. However, this interval is not always a convenient or desirable unit for reporting
data; it may provide more detail than is really useful. For reporting purposes, you can
combine RMF-based intervals into a more reasonable interval size, such as one hour or
one day.
Batch jobs, started tasks, and TSO user sessions can either be collected at RMF-based
intervals, or at termination time (end-of-job, end-of-step, end-of-session, or task
termination). The way you combine this data depends on how it was collected. If data was
collected at RMF-based intervals, the data is automatically combined unless you specify
the INTERVAL keyword. If data was collected at termination time, you can used the
COMBINE keyword to combine steps together, or combine selected job runs, TSO
sessions, or started tasks together.
The COMBINE keyword is described in the EPILOG for MVS Basic User’s Guide.
SYSID
228
If you run more than one system on the same hardware, you can report on each system
individually or on both systems together. To restrictthe data display to a particular system,
include the SYSID(cccc) keyword in the DISPLAY command, where cccc = the SMF ID of
the system data you want to display.
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 21. EXTRACT command keyword summary (continued)
Keyword Type
Description
exception threshold
There are times when you are only interested in those time periods when certain criteria
were met. For example, you might only want to display degradation data for those periods
when TSO response time was greater than two seconds. Exception thresholds allow you to
select periods using such criteria as response time, elapsed time, or a particular wait
reason.
Setting exception criteria is described in “Workload exception filters” on page 143.
The KEPILGPM procedure accepts both EXTRACT and COMPEXT commands. Keywords are free-format,
and do not have to be entered in any particular order. For a discussion of general rules for entering
commands, keywords, and operands, see the EPILOG for MVS Basic User’s Guide.
The COMPEXT command
The COMPEXT (CMX) command is similar to the EXTRACT command. However, rather than selecting
data from only the EPILOG datastore, it selects data from both the EPILOG datastore and the Profile
datastore. Then, like the COMPARE command in EPILOG for MVS, it compares degradation data for a
given workload and time period with a profile. Unless preceded by a SETP command directing otherwise,
COMPEXT extracts the most current profile for the workload and profile name (PROFNAME) you have
specified.
The results of the comparison are converted into a format that is compatible with SAS processing and are
written out to SAS data sets. These data sets are described in “The COMPLETE keyword” on page 103.
COMPEXT command syntax
The COMPEXT command follows the same syntax conventions that are outlined for the COMPARE
command, as described in theEPILOG for MVS Basic User’s Guide. The COMPEXT command and its
keywords are illustrated below:
COMPEXT {workload} {PROFNAME(cc...cc)} [time period] [SYSID(cccc,...)] [REPORTIF(filter)]
Figure 134. COMPEXT command format
As you can see in Figure 134, the only required keywords are a workload identifier and a profile name. A
description of the keywords follows.
workload
The workload identifier used to select the workloads records you want to
extract and compare with a profile. Workload records can be performance
groups, batch jobs, started tasks, or TSO user sessions. You can use any
of the workload-type keywords, except STEP and INTERVAL.
PROFNAME(cc...cc)
The 1- to 16-character name of the profile with which you want to
compare the EPILOG datastore records. Typically, this is the name you
assigned to the specified workload when you created its profile. If you
created the profile with the PROFNAME (or PNAME(cc...cc)) default,
simply specify PNAME without any operand. Note that you cannot use the
asterisk as a substitute for a profile name or for any part of a profile name
that is, TSO*).
time period
The date and/or time interval for which you are extracting the specified
Chapter 12. Reporting with SAS graphics
229
workload’s EPILOG datastore records. Use any of the date/time keywords.
If no time period is given, all EPILOG datastore records for the specified
workload are selected.
Note that your time period specification applies only to the selection of
EPILOG datastore data. Unless you have indicated otherwise on a prior
SETP command, the profile selected is the most current profile for the
workload, PROFNAME, time period, and SYSID that you have specified
on the command.
SYSID(cccc,...)
The SMF System ID for the workload you are comparing. Required only
when the records in yourdatastores contain different SYSIDs and you want
to extract a workload and profile with a specific SYSID. If you specify
multiple SYSIDs, for example (SYSA,SYSB), you will compare each
workload-profile combination with a corresponding SYSID that meets your
other comparison criteria. If no SYSID is specified, the specified workload
and the most recent profile that meet your other comparison criteria will be
compared regardless of SYSID. If you want to compare a profile that has
a SYSID different from that specified on the COMPEXT command, you
may do so by setting the SYSID of the profile with the SETP command.
REPORTIF(filter)
Specifies the criteria that workload records must satisfy to be extracted.
The REPORTIF keyword must be followed by a valid wait reason and
appropriate reporting criteria. Use any of the exception-threshold keywords
described in “Workload exception filters” on page 143. In addition, you
may apply the filter (ELT(>nn%) or RESP(>nn%)) to indicate that you want
to compare only those workload records with an elapsed (ELT) or
response (RESP) time nn% greater than the profile’s. (REPORTIF(filter)
may also be expressed as RIF(filter).)
Notes:
1. The COMPEXT command does not support multiple types of workloads.
2. The REPORTIF filter applies only to the EPILOG datastore records for the specified workload and not
to the profiles.
3. When the PROFNAME is used to display a profile record, the COMBINE and SELECTIF keywords are
disallowed (REPORTIF is used to specify exception criteria).
4. Only certain display options can be used with PROFNAME: COMPLETE, LIMIT, LOGOFF, LOGON,
PLOTMIN, TWAITOFF, AND TWAITON. All others are disallowed.
Examples of using the COMPEXT command
Comparing a workload with a profile
COMPEXT PGN(2) PNAME('TSO DEV') TODAY STIME(0900) ETIME(1700) RIF(RESP(>15%))
This command extracts EPILOG datastore records for performance group 2 for the current day between
the hours 9 a.m. to 5 p.m., and the profile named TSO DEV that was created for performance group 2. In
addition, the RIF keyword indicates that only those EPILOG datastore records with response times that
exceed the profile’s by 15% should be extracted. Assuming that no SETP command was issued prior to
this command, the EPILOG datastore records are compared with the most current profile record created
for class F with the name PRIME SHIFT.
230
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Comparing all batch jobs to their corresponding profiles
COMPEXT JOB(*) PNAME -
This command compares all batch job workloads with their most recent profiles. Because the PNAME
keyword is specified without any operand, the profiles must have been created with the PNAME default
(that is, automatic).
Comparing workloads and profiles for different SYSIDs
COMPEXT JOB(SORT) PNAME('NITE BATCH') SYSID(SYSA,SYSB) LASTWEEK STIME(1700) ETIME(0700)
Because multiple SYSIDs are specified, qualifying workload-profile combinations for either SYSID (SYSA
or SYSB) are extracted and compared. Thus, for SYSA and SYSB, this command extracts the EPILOG
datastore records for jobs named SORT that ran last week between the hours of 5 P.M. and 7 A.M. It
compares them with the most recent profile for the corresponding SYSID, that has the name NITE BATCH
and that was created for SORT jobs.
The SETP command: setting default values for the COMPEXT
command
You can use the SETP command to establish a default workload, time period, and/or SYSID value for
subsequent COMPEXT commands. The COMPEXT command then uses this value to select the profile to
be extracted from the Profile datastore (PRDS). These defaults remain in effect until changed by a
subsequent SETP command or cancelled by SETP CLEAR.
For information concerning the syntax and use of the SETP command, see “The SETP command: setting
default values for the COMPARE command” on page 189.
Chapter 12. Reporting with SAS graphics
231
232
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Chapter 13. Report element tables
This chapter describes the data elements that can be extracted from the EPILOG and Profile datastores,
and provides the report information that you need to select data fields and design your own customized
reports and graphs. EPILOG for MVS provides two methods for extracting historical data from the EPILOG
and Profile datastores:
v You can use the OBTAIN command to export EPILOG data to sequential files for processing by GDDM®
or other software tools, as described in Chapter 11, “Exporting EPILOG data,” on page 193.
v You can use the SAS interface commands, EXTRACT and COMPEXT, to extract EPILOG data for
processing by SAS programs, as described in Chapter 12, “Reporting with SAS graphics,” on page 201.
The two methods of exporting historical data, OBTAIN and the SAS interface, differ in the way that they
generate output files. When you use the SAS interface, the EXTRACT and COMPEXT commands write
data to SAS data sets, which have been defined with names that correspond to EPILOG workload and
resource display reports. The OBTAIN command, however, allows you to name the data sets to be used
as the output files.
Both methods of exporting historical data require that you specify at least one resource or workload
keyword. (In Chapter 11, “Exporting EPILOG data,” on page 193, these keywords are referred to as
“report-name keywords”). In other words, both methods extract historical data based on a report type,
which corresponds to a resource or workload keyword.
Report table organization
The data element definitions in this chapter are organized into three sections. In the first section, “Using
resource data” on page 233, data elements are organized into tables sorted by resource keyword. These
table names are identical with the SAS data set names that are used as output files by the SAS interface
commands. In “Using workload degradation data” on page 279, data elements are organized into workload
tables that correspond to the SAS data set names. In “The COMPLETE keyword” on page 103, data
elements are organized by SAS data set name; the data elements in these tables are not available to the
OBTAIN command.
In some cases, you may want to look up reporting information based on the data element name. To meet
this need, the OBTAIN and SAS data elements are also defined (in alphabetical order) in Appendix A,
“Data dictionary,” on page 301.
Using resource data
This section describes the tables of data elements that can be generated by using the resource keywords
with the OBTAIN or SAS interface commands. Table 22 shows the resource table names and the report
type associated with each resource table.
Table 22. Resource Tables for the OBTAIN and EXTRACT Commands
Resource Table
Type of Data
RALL
All resource types
RCCH
Cache subsystem statistics
RCHN
Channel resource
RCPU
CPU resource
RDAS
DASD resource
RDOM
SRM domain activity
RINF
General information
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
233
Table 22. Resource Tables for the OBTAIN and EXTRACT Commands (continued)
Resource Table
Type of Data
RLCU
I/O queuing information
RPAG
Paging and storage information
RPDS
Page data set information
RPGN
Performance group resource
RSDS
Swap data set information
RSRM
SRM MPL adjustment values
RSWA
Swap activity statistics
RSWR
Swap reasons
RVLF
VLF class statistics
Each table name shown in Table 9 on page 63 corresponds to a resource keyword that can be used to
generate the data in the table. The SAS interface writes the extracted data to SAS data sets using the
same names. For example, if you use the RDAS keyword with the EXTRACT command to generate
device activity information, the extracted data is written to a SAS data set that is named RDAS, as
explained in Chapter 12, “Reporting with SAS graphics,” on page 201. If you use the RDAS keyword with
the OBTAIN command to generate device activity information, the extracted data is written to a
user-specified data set, as explained in Chapter 11, “Exporting EPILOG data,” on page 193.
The SAS data sets
In most instances, EXTRACT and COMPEXT commands write to only one SAS data set. For example, if
you specify:
EXTRACT RCHN LASTMONTH
the SAS interface invokes the KEPILGPM procedure, extracts information about channel resources, and
writes that information out to the RCHN data set.
The exceptions to this occur with the PROGRAM and ACCOUNT keywords. Since these keywords may
select batch jobs, started tasks, and TSO user sessions, they may update more than one SAS data set
with a single EXTRACT or COMPEXT command.
Each invocation of the KEPILGPM procedure resets the 16 SAS data sets. If either the EXTRACT or the
COMPEXT command(s) in the procedure does not write any data to a particular data set, then that data
set contains no records after the procedure is executed.
Table contents
The following sections describe the data elements in each of the resource report tables. For each data
element, the table shows the corresponding field on the EPILOG display screen, a description, the
dictionary name, and the data type and length, and the filters available.
EPILOG field
This column shows the column header or display label of the data element on the EPILOG display panel.
A value of ---- in this column indicates that the data element does not appear on the EPILOG report, but
is available for OBTAIN or SAS processing.
Description
The description field includes information concerning value ranges, environmental restrictions, and product
restrictions. For example, the term (SAS only) indicates that a data element can be exported by the
EXTRACT command in the SAS interface, but not by the OBTAIN command.
234
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Environmental considerations are documented on an exception basis. Data elements are available on all
MVS operating systems and environments unless specifically restricted. For example, the environmental
restriction (MVS/XA and MVS/ESA only) indicates that a data element is not available in the MVS/370
environment. The following environmental restrictions are documented: MVS/370, MVS/XA, MVS/ESA, the
PR/SM¬ machine, the 3090¬ processor, and expanded storage.
Dictionary name
The name of the data element as it appears in Appendix A, “Data dictionary,” on page 301. This is the
same as the element name in the internal data dictionary, the element name used as a keyword in the
OBTAIN command, and the variable name that is written by the SAS interface commands to SAS data
sets.
Short name
The short form of the dictionary name. (The data elements in the workload tables do not have short
names.)
Data type and length
For SAS data, the data type indicates the internal SAS representation type and length (in bytes). For
example, a data type of “F, 4” means that the variable is a floating point number 4 bytes long. The date
and time variables contain standard SAS date/time values.
For OBTAIN data, the INT column shows the internal OBTAIN representation and length (in bytes). The
COL,PC column shows the data type and length if the OBTAIN command uses the COL or PC operand in
the FORMAT keyword. The COL and PC formats use the same data type and length. For further
information, see Chapter 11, “Exporting EPILOG data,” on page 193.
The following symbols are used in the Data Type columns:
Symbol
Definition
C
Character (all formats)
D
Floating point number (INT format)
F
Floating point number (SAS format)
H
Hexadecimal (in EBCDIC for COL and PC formats)
I
Numeric signed integer (INT format)
N
General numeric format. Can be integer or decimal. If positive, will be unsigned; if
negative, will be signed. (COL and PC formats)
P
Packed decimal (INT format)
T
Time in hundredths of a second (for INT format)
U
Unsigned binary integer (INT format)
X
Hexadecimal (INT format)
Filters
Indicates whether exception filtering is supported for the data element (R=RIF, S=SIF). For further
information about exception filters, see “Workload exception filters” on page 143 and “Resource exception
filters” on page 155.
Using cache resource data (RCCH)
The RCCH keyword generates information about cache subsystem activity on your system.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Chapter 13. Report element tables
235
Table 23. Cache resource activity
Data Type and Length
EPILOG
Field
Description
Volume
Serial
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
The cache device volume
serial number.
VOLSER
VOLS
C,6
C,6
C,6
R
Dev #
The cache device address or
number.
DEVADD
DEVA
C,4
C,4
C,4
R
CU type
The controller model.
CNTRLMOD
CMOD
C,8
C,7
C,8
R
SSID
The cache device subsystem
id on the 3990.
SUBSYSID
SSYS
C,4
C,4
C,4
R
Stg Dir
The first storage director on
the 3880.
STGDIR1
STD1
C,2
C,2
C,2
R
Stg Dir
The second storage director on STGDIR2
the 3880.
STD2
C,2
C,2
C,2
R
Cache State The caching status of the 3880 CACHSTT1
first storage director or 3990
subsystem.
CST1
C,8
C,8
C,8
R
Cache State The caching status of the 3880 CACHSTT2
second storage director.
CST2
C,8
C,8
C,8
R
Read Hit %
The total read hit percentage
for the cache device,
expressed in tenths of a
percent.
RDHITTOT
RHTT
F,4
I,4
N,11
R
----
The sequential read hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3880 first storage
director or the 3990
subsystem.
RDHITSQ1
RHS1
F,4
I,4
N,11
R
----
The sequential read hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3880 second storage
director.
RDHITSQ2
RHS2
F,4
I,4
N,11
R
----
The normal read hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3880 first storage
director or the 3990
subsystem.
RDHITNM1
RHN1
F,4
I,4
N,11
R
----
The normal read hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3880 second storage
director.
RDHITNM2
RHN2
F,4
I,4
N,11
R
236
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 23. Cache resource activity (continued)
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Write Hit %
The total write hit percentage
for the cache device,
expressed in tenths of a
percent.
WRHITTOT
WHTT
F,4
I,4
N,11
R
----
The sequential write hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3990 subsystem.
WRHITSEQ
WHSQ
F,4
I,4
N,11
R
----
The normal write hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3990 subsystem or
the 3880 first storage director.
WRHITNM1
WHN1
F,4
I,4
N,11
R
----
The normal write hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3880 second storage
director.
WRHITNM2
WHN2
F,4
I,4
N,11
R
----
The cache fast write write hit
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3990 subsystem.
WRHITCFW
WHCF
F,4
I,4
N,11
R
----
The DASD fast write status.
This statistic comes from the
3990 subsystem.
DASDFWST
DFWS
C,8
C,8
C,8
R
----
The DASD fast write sequential WRHITDFS
write hit percentage for the
cache device, expressed in
tenths of a percent. This
statistic comes from the 3990
subsystem.
WHDS
F,4
I,4
N,11
R
----
The DASD fast write normal
write hit percentage for the
cache device, expressed in
tenths of a percent. This
statistic comes from the 3990
subsystem.
WRHITDFN
WHDN
F,4
I,4
N,11
R
----
The total cache fast write read
hit percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3990 subsystem.
RDHITCFW
RHCF
F,4
I,4
N,11
R
RDWRTRT
The reads to writes percentage RDWRTRT
for the cache device,
expressed in tenths of a
percent.
RDWR
F,4
I,4
N,11
R
Chapter 13. Report element tables
237
Table 23. Cache resource activity (continued)
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Inhibit
Cache %
The inhibit cache percentage
for the cache device,
expressed in tenths of a
percent. This statistic comes
from the 3990 subsystem or
the 3880 first storage director.
INHBCCH1
IHC1
F,4
I,4
N,11
R
Inhibit
Cache %
The inhibit cache percentage
for the cache device,
expressed in tenths of a
percent. This statistic comes
from the 3880 second storage
director.
INHBCCH2
IHC2
F,4
I,4
N,11
R
Bypss
Cache %
The bypass cache percentage
for the cache device,
expressed in tenths of a
percent. This statistic comes
from the 3990 subsystem or
the 3880 first storage director.
BPSSCCH1
BPC1
F,4
I,4
N,11
R
Bypss
Cache %
The bypass cache percentage
for the cache device,
expressed in tenths of a
percent. This statistic comes
from the 3880 second storage
director.
BPSSCCH2
BPC2
F,4
I,4
N,11
R
DFW Bypss The DASD Fast Write retry
%
percentage for the cache
device, expressed in tenths of
a percent. This statistic comes
from the 3990 subsystem.
DFWBYPSS
DFWR
F,4
I,4
N,11
R
----
The sequential track transfers
from DASD to cache for cache
device, expressed in tenths of
a track per second. This
statistic comes from the 3990
subsystem.
DASDCCHS
TDCS
F,4
I,4
N,11
R
----
The normal track transfers
from DASD to cache for cache
device, expressed in tenths of
a track per second. This
statistic comes from the 3990
subsystem.
DASDCCHN
TDCN
F,4
I,4
N,11
R
----
The normal track transfers
from cache to DASD for cache
device, expressed in tenths of
a track per second. This
statistic comes from the 3990
subsystem.
CCHDASDN
TCDN
F,4
I,4
N,11
R
238
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 23. Cache resource activity (continued)
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
----
The normal write requests for
cache device, expressed in
tenths of a write per second.
This statistic comes from the
3880 first storage director.
WRITENM1
WRN1
F,4
I,4
N,11
R
----
The normal write requests for
cache device, expressed in
tenths of a write per second.
This statistic comes from the
3880 second storage director.
WRITENM2
WRN2
F,4
I,4
N,11
R
----
The sequential write requests
for cache device, expressed in
tenths of a write per second.
This statistic comes from the
3880 first storage director.
WRITESQ1
WRS1
F,4
I,4
N,11
R
----
The sequential write requests
for cache device, expressed in
tenths of a write per second.
This statistic comes from the
3880 second storage director.
WRITESQ2
WRS2
F,4
I,4
N,11
R
DFW Hit %
The total DASD fast write write WRHITDFT
hit percentage for the cache
device, expressed in tenths of
a percent. (3990 subsystem)
WHDT
F,4
I,4
N,11
R
Using channel resource data (RCHN)
The RCHN keyword generates information about physical channel activity on your system. One record (or
SAS observation) is generated for each channel on your system. For example, if you have eight channels
on your system, each interval generates eight records (or observations).
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Table 24. Channel resource activity
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
CPU/CSID
Channel set ID to which the
channel is attached. (MVS/370
only)
CHNSETID
CHNS
F,4
I,4
N,10
R
CHPID
Channel path ID in hex.
(MVS/XA and MVS/ESA only)
CHPID
CPID
C,2
C,2
C,2
R
Chn
Channel number. (MVS/370
only)
CHNNUM
CHNM
F,4
U,1
N,3
R
Typ
Channel type, such as byte
multiplexor, selector, or block
multiplexor.
CHNTYPE
CTYP
C,8
C,6
C,6
R
Chapter 13. Report element tables
239
Table 24. Channel resource activity (continued)
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Mode
The active environment for the
CPU complex.
CPENVIRN
CPEN
C,8
C,8
C,8
R
CPMF
The Channel Path
Measurement Facility status.
CPMFSTAT
CPMF
C,12
C,12
C,12
R
%Busy
Percentage of the time during CPCTBUSY
the interval that the channel
was busy. The total or complex
percent utilization of the
channel.
BUSY
F,4
I,4
N,11
R
%Utilization
Percentage of the RMF interval CPLTBUSY
time the channel was busy to
the LPAR.
CPLU
F,4
I,4
N,11
R
State
Channel path status with
respect to the monitored
partition.
CPLPSTAT
CPST
C,4
C,4
C,4
R
SHR
Channel path LPAR sharable
status.
CPLSHARE
CPSH
C,2
C,2
C,2
R
Cnfg
Dynamic I/O update status of
the channel.
CPCONFIG
CPCF
C,4
C,4
C,4
R
I/Os /sec
Average number of successful CIOPRSEC
START and RESUME I/O
instructions issued to the
channel per second. (MVS/370
only)
CIOS
F,4
I,4
N,11
R
Avg Srv.
Time(ms)
Average time (in milliseconds)
required to complete each I/O
operation to the channel.
(MVS/370 only)
CSRVTIME
SERV
F,4
I,4
N,10
R
----
Architecture Flag. XA»=
MVS/XA and MVS/ESA, 370=
MVS/370.
DSXAFLAG
C,3
C,3
C,3
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
P,4
N,8
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
T,4
C,8
240
EDATTIME
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Table 24. Channel resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
Short
Name
OBTAIN
SAS
INT
COL,PC
----
Date and time of extraction
from the EPILOG datastore.
(SAS only)
EXTDTIME
F,8
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
F,4
I,4
N,10
----
Indicates the level of MVS/ESA OSLEVEL
which you are running. A value
of 1 indicates that you are
running under ESA 3.1.0; a
value of 2 indicates that you
are running under ESA 3.1.3; a
value of 3 indicates that you
are running under SP 4.1.0; a
value of 4 indicates that you
are running under SP 4.2.0.
F,4
U,1
N,10
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
P,4
N,8
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
T,4
C,8
----
SMF ID of system.
SMFID
C,4
C,4
OSLV
Filters
F,8
C,4
Using CPU resource data (RCPU)
The RCPU keyword generates information about CPU activity on your system. For multi-CPU systems,
there is one record for each CPU, and the common system-wide variables such as BATMINAS and
READYIN are repeated.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type. In addition to the fields described in the table, a display will
Chapter 13. Report element tables
241
contain warning messages if one or more intervals had missing, invalid, or changed data, such as a
missing CPU speed factor or a processor configuration change.
Table 25. CPU resource activity
Data Type and Length
EPILOG
Field
OBTAIN
Description
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Batch Min
Batch Max
Batch Avg
The minimum, maximum, and
average number of batch jobs
running during the time period.
BATMINAS
BATMAXAS
BATAVGAS
BMIN
BMAX
BAVG
F,4
F,4
F,4
I,4
I,4
I,4
N,10
N,10
N,11
R
R
R
STC Min
STC Max
STC Avg
The minimum, maximum, and
average number of started
tasks running during the time
period.
STCMINAS
STCMAXAS
STCAVGAS
SMIN
SMAX
SAVG
F,4
F,4
F,4
I,4
I,4
I,4
N,10
N,10
N,11
R
R
R
TSO Min
TSO Max
TSO Avg
The minimum, maximum, and
average number of TSO users
during the time period.
TSOMINAS
TSOMAXAS
TSOAVGAS
TMIN
TMAX
TAVG
F,4
F,4
F,4
I,4
I,4
I,4
N,10
N,10
N,11
R
R
R
Ready
Users In
Average number of ready
users swapped in.
READYIN
RIN
F,4
I,4
N,11
R
Ready
Users Out
Average number of ready
users swapped out.
READYOUT
ROUT
F,4
I,4
N,11
R
----
Indicates the level of MVS/ESA OSLEVEL
which you are running. A value
of 1 indicates that you are
running under SP 3.1.0; a
value of 2 indicates that you
are running under SP 3.1.3; a
value of 3 indicates that you
are running under SP 4.1.0; a
value of 4 indicates that you
are running under SP 4.2.0.
OSLV
F,4
U,1
N,10
CPU
Utilization
SRB TCB
MVS
PCTCPSRB
Percent CPU used by SRBs,
PCTCPTCB
TCBs, and MVS for all
processors. (In processors that PCTCPMVS
support PR/SM, these values
pertain to the logical CPU, that
is, the partition under which
EPILOG is running.) If MVS is
running as a guest under VM,
these values pertain only to the
CPU percentage provided by
VM.
PSRB
PTCB
PMVS
F,4
F,4
F,4
I,4
I,4
I,4
N,11
N,11
N,11
CPUn
n = the CPU ID.
CPUID
F,4
I,4
N,10
nn%
Total percent CPU utilization
for CPUn.
PCTCPU
PCPU
F,4
I,4
N,11
R
Interrupt
Rate
Number of I/O interrupts per
second handled by the
processor. (MVS/XA and
MVS/ESA only)
IOINTRPT
IONT
F,4
I,4
N,11
R
%TPI
Percent of the total I/O
interrupts that were
circumvented with the TPI
instruction. (MVS/XA and
MVS/ESA only)
PCTTPI
PTPI
F,4
I,4
N,11
R
242
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
R
R
R
Table 25. CPU resource activity (continued)
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Average
Logical
Processor
Utilization
The average percentage of
total CPU cycles assigned to
this partition that this partition
used during the time period.
(PR/SM machine only)
PCTPARTA
PRTA
S,4
I,4
N,11
R
Physical
CPU
Utilization
The average percentage of
complex-wide CPU cycles that
this partition used during the
time period. (PR/SM machine
only)
PCTPRTCP
PRTC
S,4
I,4
N,11
R
LPAR
Management
The percentage of LPAR
management overhead for the
partition in which the EPILOG
collector is running. (PR/SM
machine only)
PCTPRMCP
PRMC
F,4
I,4
N,10
R
----
CPUFLAG
Indicates any change to the
status of the CPU configuration
that may have affected RCPU
values during the interval. The
CPU flags are represented as
one decimal value. The
decimal value is equivalent to
a 32 bit unsigned binary field.
Each bit represents a status
flag. Bit 0 represents the most
significant bit.
F,4
X,4
H,8
Bit
Description
0-7
Unused.
8
Number of physical
processors assigned
for use by PR/SM
changed during the
combined reporting
interval.
9
Number of logical
processors assigned
to this PR/SM partition
changed during the
combined reporting
interval.
10
Wait state assist
status of the PR/SM
partition changed
during the combined
reporting interval.
Chapter 13. Report element tables
243
Table 25. CPU resource activity (continued)
Data Type and Length
EPILOG
Field
----
Description
Bit
Description
11
The combined record
includes interval
records with different
PR/SM partition
names or numbers.
12
LPAR Management
data indicator.
Dictionary
Name
Short
Name
CPUFLAG
13 - 14 Unused.
244
15
Indicates the
availability of LPAR
management data
during the
user-requested
combined reporting
interval.
16
The number of CPUs
in the configuration
changed during the
combined reporting
interval.
17
Reserved.
18
The record contains
queuing I/O
information (XA and
ESA only).
19
Some combined
interval records lack
queuing information.
20
SRB service definition
coefficient changed
during the combined
interval.
21
Indicates that the
system was running
as PR/SM partition.
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
OBTAIN
SAS
INT
COL,PC
F,4
X,4
H,8
Filters
Table 25. CPU resource activity (continued)
Data Type and Length
EPILOG
Field
----
----
----
Description
Bit
Description
22
PR/SM information
missing.
23
PR/SM information
missing (combined
interval records).
24
CPU speed factor is
missing from the
record. The CPU
speed factor is used
to calculate the SRB,
TCB, and MVS
percent utilization
figure. If it is missing,
EPILOG derives a
CPU speed factor
from the current
processor
configuration and
uses it in calculating
the percent utilization
figures.
25
The number of CPUs
in the configuration
changed during the
combined reporting
interval
26
Reserved
27
An error was
encountered in
calculating the MVS
overhead
Bit
Description
28
Indicates dedicated
processors in the
PR/SM partition.
29
Indicates the wait
state assist status for
the PR/SM logical
processor
30
Unused.
31
Unused.
Architecture Flag. XA»=
MVS/XA and MVS/ESA, 370=
MVS/370.
Dictionary
Name
Short
Name
OBTAIN
SAS
INT
COL,PC
CPUFLAG
F,4
X,4
H,8
CPUFLAG
F,4
X,4
H,8
DSXAFLAG
C,3
C,3
C,3
Filters
Chapter 13. Report element tables
245
Table 25. CPU resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
Short
Name
OBTAIN
SAS
INT
COL,PC
P,4
N,8
T,4
C,8
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
(SAS only)
EXTDTIME
F,8
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
F,4
I,4
N,10
----
Indicates the level of MVS/ESA OSLEVEL
which you are running. A value
of 1 indicates that you are
running under ESA 3.1.0; a
value of 2 indicates that you
are running under ESA 3.1.3; a
value of 3 indicates that you
are running under SP 4.1.0; a
value of 4 indicates that you
are running under SP 4.2.0.
F,4
U,1
N,10
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
P,4
N,8
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
246
EDATTIME
F,8
OSLV
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Filters
Table 25. CPU resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
----
SMF ID of system.
SMFID
Short
Name
OBTAIN
SAS
C,4
INT
COL,PC
T,4
C,8
C,4
C,4
Filters
Using DASD resource data (RDAS)
The RDAS keyword generates information about DASD device activity on your system. One record is
generated for each DASD device on your system. For example, if you have 16 DASD devices, each
interval generates 16 records.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Table 26. DASD device resource activity
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Dev #
Device address (for MVS/370)
or device number (for MVS/XA
and MVS/ESA).
DEVADD
DEVA
C,4
C,3
C,3
R
Volume
Serial
Volume serial identifier.
VOLSER
VOLS
C,6
C,6
C,6
R
LCU
Logical control unit associated
with the DASD device.
(MVS/XA and MVS/ESA only)
DLCUID
DLCU
C,3
C,3
C,3
R
I/O Rate
Rate per second at which I/O
requests were issued to the
device.
DIOPRSEC
DIOP
F,4
I,4
N,11
R
Total
The average end-to-end
service time per I/O to the
device (in milliseconds).
TOTIOTME
TTME
F,4
I,4
N,10
R
Path Q
The average time that an I/O
spent queued on the path (in
milliseconds). (MVS/370 only)
PATHQTME
PQTM
F,4
I,4
N,10
R
Dev Q
The average time that an I/O
DEVQTIME
spent queued on the device (in
milliseconds). (MVS/370 only)
DQTM
F,4
I,4
N,10
R
Srv
The average time that an I/O
spent transferring data to the
device (in milliseconds).
(MVS/370 only)
SVTM
F,4
I,4
N,10
R
PBSY
F,4
I,4
N,11
R
SRVTIME
% Dev Busy Percentage of time during the DPCTBUSY
time period that the device was
found busy.
Chapter 13. Report element tables
247
Table 26. DASD device resource activity (continued)
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
PCTRSVDL
PRSV
F,4
I,4
N,11
R
% RSV
Delay
Percentage of time during the
time period that a shared
device was reserved by
another processor or a control
unit was busy with another
path. (MVS/370 only)
IOSQ
IOSQTIME
The average time that an I/O
spent on the IOS queue for the
device (in milliseconds).
(MVS/XA and MVS/ESA only)
QTIM
F,4
I,4
N,10
R
Pend
The average time that an I/O
PENDTIME
spent in the pending condition
(in milliseconds). (MVS/XA and
MVS/ESA only)
PTIM
F,4
I,4
N,10
R
Conn
The average connect time for
an I/O to the device (in
milliseconds). (MVS/XA and
MVS/ESA only)
CONNTIME
CTIM
F,4
I,4
N,10
R
Disc
The average disconnect time
for an I/O to the device (in
milliseconds). (MVS/XA and
MVS/ESA only)
DISCTIME
DTIM
F,4
I,4
N,10
R
CUB
Average delay to an I/O
request caused by the control
unit busy condition (in
milliseconds). (3090 only)
DCUBSY
CUBY
F,4
I,4
N,11
R
DB
Average delay to an I/O
request caused by the device
busy condition (in
milliseconds). (3090 only)
DDEVBSY
DVBY
F,4
I,4
N,11
R
% Dev Util
See % Dev Busy.
Open Dsns
or Avg Allcs
Average number of DCBs and
ACBs concurrently open on a
volume. (On MVS/ESA
systems, the column header is
Avg Allcs, and this field
contains the average number
of allocations against a
device.)
OPENDSN
ODSN
F,4
I,4
N,11
R
----
Architecture Flag. XA»=
MVS/XA and MVS/ESA, 370=
MVS/370.
DSXAFLAG
C,3
C,3
C,3
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
P,4
N,8
248
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 26. DASD device resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
Short
Name
EDATTIME
OBTAIN
SAS
INT
COL,PC
T,4
C,8
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
(SAS only)
EXTDTIME
F,8
----
Machine flag. (“309X” for 3090
processors, “»»»»” for
non-3090 processors)
MACHFLAG
C,4
C,4
C,4
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records have passed the
exception criteria and were
included in the observation.
F,4
I,4
N,10
----
Indicates the level of MVS/ESA OSLEVEL
which you are running. A value
of 1 indicates that you are
running under ESA 3.1.0; a
value of 2 indicates that you
are running under ESA 3.1.3; a
value of 3 indicates that you
are running under SP 4.1.0; a
value of 4 indicates that you
are running under SP 4.2.0.
F,4
U,1
N,10
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
P,4
N,8
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
Filters
F,8
OSLV
F,8
Chapter 13. Report element tables
249
Table 26. DASD device resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
----
SMF ID of system.
SMFID
Short
Name
OBTAIN
SAS
C,4
INT
COL,PC
T,4
C,8
C,4
C,4
Filters
Using SRM domain activity data (RDOM)
The RDOM keyword generates information about SRM domain activity on your system. One record is
generated for each domain defined in your system. For example, if you have 10 domains, each interval
generates 10 records. The following table shows a description of each EPILOG field that is extracted,
along with the corresponding dictionary name and data type.
Table 27. SRM domain resource activity
Data Type and Length
EPILOG
Field
OBTAIN
Description
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Dmn
Domain number.
DOMNUM
DMN
F,4
U,1
N,3
R
Min MPL
Max MPL
Minimum and maximum MPL
MINMPL
values specified for the domain MAXMPL
in the IPS.
MNM
MXM
F,4
I,4
N,10
R
Tar MPL
Target MPL computed by SRM. TARGMPL
TDM
F,4
I,4
N,10
R
Cur MPL
Actual MPL for the interval.
CURMPL
CUM
F,4
I,4
N,10
R
Contn Index Contention index computed by
SRM for the interval.
CONTIDX
CIDX
F,4
I,4
N,11
R
Avg U
Ready
Average number of ready
users.
AVGURDY
ARDY
F,4
I,4
N,11
R
Users In
Average number of swappable
users from the domain that
were in-storage during the
interval.
AVGUIN
AUIN
F,4
I,4
N,11
R
Users Out
Average number of
swapped-out users from the
domain during the interval.
AVGUOUT
AUOT
F,4
I,4
N,11
R
Going Out
Average number of users from
the domain that were marked
for swap-out.
AVGUGOUT
AGOT
F,4
I,4
N,11
R
Avg NSwap
Average number of
non-swappable in-storage
users during the interval.
AVGNSWAP
ANSP
F,4
I,4
N,10
R
Intvl Srvce
Average number of SRM
service units used by all
members of the domain during
the interval.
INTVLSRV
SRVU
F,4
I,4
N,10
R
250
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 27. SRM domain resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
Short
Name
OBTAIN
SAS
INT
COL,PC
P,4
N,8
T,4
C,8
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
(SAS only)
EXTDTIME
F,8
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records have passed the
exception criteria and were
included in the observation.
F,4
I,4
N,10
----
Indicates the level of MVS/ESA OSLEVEL
which you are running. A value
of 1 indicates that you are
running under ESA 3.1.0; a
value of 2 indicates that you
are running under ESA 3.1.3; a
value of 3 indicates that you
are running under SP 4.1.0; a
value of 4 indicates that you
are running under SP 4.2.0.
F,4
U,1
N,10
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
P,4
N,8
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
EDATTIME
Filters
F,8
OSLV
F,8
Chapter 13. Report element tables
251
Table 27. SRM domain resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
----
SMF ID of system.
SMFID
Short
Name
OBTAIN
SAS
C,4
INT
COL,PC
T,4
C,8
C,4
C,4
Filters
Using general system resource data (RINF)
The RINF keyword generates information about the status of your general system resources. One record
is generated for each interval. The following table shows a description of each EPILOG field that is
extracted, along with the corresponding dictionary name and data type.
Table 28. General system resource activity
Data Type and Length
EPILOG
Field
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
Operating
Operating system and level
System and (e.g., MVS SP3.1.0).
Level
SYSLEVEL
SYS
C,8
C,8
C,8
R
Model
CPU model.
CPUMODEL
CPM
C,6
C,6
C,6
R
Serial
CPU serial number.
CPUSERL
CPS
C,6
C,6
C,6
R
Mode
Processor mode during the
time period:
Native
Under VM
Under VM with PMA
PARTITIONED
MODE
C,17
C,17
C,17
R
Description
(Note: If you are searching with
RIF for: Under VM or Under
VM with PMA, you must
enclose the character string in
single quotes, because it
contains blanks.)
IPS
The current IPS suffix.
IPS
C,2
C,2
C,2
R
ICS
The current ICS suffix.
ICS
C,2
C,2
C,2
R
OPT
The current OPT suffix.
OPT
C,2
C,2
C,2
R
RMF
RMF version number during
the time period.
RMF
C,20
C,20
C,20
R
Partition
Name
PARTNAME
The name assigned to the
partition in which EPILOG is
running. (PR/SM machine only)
PTN
C,8
C,8
C,8
R
Partition
Number
The partition number in which
EPILOG is running. (PR/SM
machine only)
PNUM
F,4
U,1
N,3
R
252
PARTNUM
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 28. General system resource activity (continued)
Data Type and Length
EPILOG
Field
Description
OBTAIN
Dictionary
Name
Short
Name
SAS
INT
COL,PC
Filters
NUMPHYS
NPRO
C,1
C,1
C,1
R
Physical
Processors
The number of processors
available to the PR/SM
processor complex. (PR/SM
machine only)
Processor
Weight
PARTWGT
Relative importance of the
logical partition with respect to
the other partitions that share
the same CPU (PR/SM
machine only). A value of MIX
indicates a logical partition with
both shared and dedicated
processors.
PWT
C,3
C,3
C,3
R
Processor
Complex
CPU
Utilization
Percentage of total CPU cycles PCTCMPLX
used by all the partitions in the
processor complex. (PR/SM
machine only)
PCTX
F,4
I,4
N,11
R
Total LPAR
Management
Overhead
Percent of total LPAR
management overhead for the
processor complex. (PR/SM
machine only)
PCTMGPLX
PCMX
F,4
I,4
N,10
R
----
INFFLAG
Bits 8-15 indicate changes
made during a combined
interval (not used unless the
interval is combined). The flags
are represented as one
decimal value. The decimal
value is equivalent to a 32 bit
unsigned binary field. Each bit
represents a status change. Bit
0 represents the most
significant bit.
F,4
X,4
H,8
Bit
Description
0-7
Unused.
8
Partition number
changed.
9
Partition name
changed.
Chapter 13. Report element tables
253
Table 28. General system resource activity (continued)
Data Type and Length
EPILOG
Field
----
Description
Bit
Description
10
Number of physical
processors changed.
11
Relative share
changed.
12
Global time slice
changed.
13
CPU capping
changed.
14 -15
Unused.
16
VM is running in
native mode.
17
PR/SM partition.
18
Dedicated processors.
19
Indicates the wait
state assist status for
the PR/SM logical
processor.
20
Indicates CPU
capping in effect for
the partition in which
the EPILOG collector
is running.
21
Indicates LPAR
management data is
present.
Dictionary
Name
Short
Name
OBTAIN
SAS
INT
COL,PC
INFFLAG
F,4
X,4
H,8
INFFLAG
F,4
X,4
H,8
P,4
N,8
22 - 23 Unused.
----
----
254
Bit
Description
24
MVS level.
25
IPS.
26
ICS.
27
OPT.
28
CPU model.
29
Serial.
30
Mode.
31
RMF version.
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Filters
Table 28. General system resource activity (continued)
Data Type and Length
EPILOG
Field
Description
Dictionary
Name
EDATTIME
Short
Name
OBTAIN
SAS
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
(SAS only)
EXTDTIME
F,8
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
F,4
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
----
SMF ID of system.
SMFID
INT
COL,PC
T,4
C,8
I,4
N,10
P,4
N,8
T,4
C,8
C,4
C,4
Filters
F,8
F,8
C,4
Using I/O queuing data (RLCU)
The RLCU keyword generates information about I/O queuing on your system.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Chapter 13. Report element tables
255
Table 29. I/O queuing activity
Data type and length
EPILOG
field
Description
I/O
processor
The I/O processor identifier.
(3090 only)
Dictionary
name
Short
name
IOPROCID
OBTAIN
SAS
INT
COL,PC
C,2
C,2
C,2
Filters
I/Os per sec For non-3090 processors, this
field refers to the rate at which
I/O requests were successfully
selected for initiation by the
associated LCU.
LCUIOSEC
LIOS
F,4
I,4
N,11
R
I/Os per sec For 3090 processors, this field
shows the rate per second at
which I/O requests are placed
on the IOP initiation queue.
IOPIOSEC
IOPS
F,4
I,4
N,11
R
Avg # I/O Q This field indicates the average LCUIOQLN
number of I/O requests queued
for this LCU.
LOQL
F,4
I,4
N,11
R
IOPIOQLN
IOQL
F,4
I,4
N,11
R
DLCU
C,3
C,3
C,3
R
Avg # I/O Q For 3090 processors, this field
indicates the average number
of I/O requests queued for the
I/O processor.
LCU
The logical control unit number. DLCUID
I/Os per sec The rate at which the IOP
Deferred
placed deferred I/O requests
on the control unit header
queue. (3090 only)
IOPTLDEF
IDEF
F,4
I,4
N,11
R
% All Ch
Path Busy
CPPCTBSY
CBSY
F,4
I,4
N,11
R
The percentage of the time
that all channel paths for this
LCU were busy at the same
time.
% Requests Shows the percentage of I/O
Deferred:
requests deferred, broken
down by the reason (device
busy or control unit busy).
(non-3090 only)
Total
IODEFSEC
Total percent of I/O requests
on the LCU that were deferred.
(non-3090 only)
IDFS
F,4
I,4
N,11
R
Dev Busy
Percent of requests deferred
due to a DEVICE BUSY
condition. (non-3090 only)
DPCTBSY
DBSY
F,4
I,4
N,11
R
CU Busy
Percent of requests deferred
due to a CONTROL UNIT
BUSY condition. (non-3090
only)
CUPCTBSY
UBSY
F,4
I,4
N,11
R
256
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 29. I/O queuing activity (continued)
Data type and length
EPILOG
field
% CU Busy
Description
Dictionary
name
The percentage of I/O requests CUDEFSEC
deferred for that control unit.
(3090 only)
OBTAIN
Short
name
SAS
INT
COL,PC
Filters
UDEF
F,4
I,4
N,11
R
C,3
C,3
C,3
R
This value is reported even if
the channel path status was
changed during the interval.
(You can check the LCUCPST
variable to see if any changes
were made to the channel path
during the interval.) In the
event of such a change, this
data should be interpreted with
caution.
CU
CUID
Identifies a control unit in the
logical control unit (LCU). If
there is more than one channel
path per control unit, a value is
generated for each channel
path.
CHN
The channel path ID
associated with this control
unit. (non-3090 only)
CHPID
CPID
C,2
C,2
C,2
R
CHPID
The channel path ID(s)
associated with this control
unit. (3090 only)
CHPID
CPID
C,2
C,2
C,2
R
----
LCU record subtype.
LCUTYPE
LTYP
C,1
U,1
N,3
CHPID Ints
per sec
The rate at which I/O requests
for devices attached to this
control unit are completed by
this channel path. (3090 only)
CHPINSEC
CINS
F,4
I,4
N,11
R
If the channel path was varied
online or offline during the
interval, this variable contains
zeros.
Chapter 13. Report element tables
257
Table 29. I/O queuing activity (continued)
Data type and length
EPILOG
field
Comment
Description
Indicates the status of the
channel path during the
supporting interval. The status
flags are represented as one
decimal value. The decimal
value is equivalent to a 32-bit
unsigned binary field. Each bit
represents a status flag. Bit 0
represents the most significant
bit.
Bit
Description
0
Channel path installed
1
Channel path online
2
Channel path varied
3
Channel path offline to
all devices
4
Channel path to all
devices altered
5
Path data invalid
Dictionary
name
Short
name
LCUCPST
OBTAIN
SAS
INT
COL,PC
F,4
X,1
H,2
P,4
N,8
T,4
C,8
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
(SAS only)
EXTDTIME
F,8
----
Error flag. If this field is equal
to any value other than
hexadecimal zero (x’0’), it
indicates that some error took
place in data collection, and
that the data in this
observation may not be
reliable.
LCUERROR
C,1
X,1
H,2
----
Machine flag. (“309X” for 3090
processors, “»»»»” for
non-3090 processors)
MACHFLAG
C,4
C,4
C,4
258
EDATTIME
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Filters
Table 29. I/O queuing activity (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
----
Indicates the level of MVS/ESA OSLEVEL
which you are running. A value
of 1 indicates that you are
running under ESA 3.1.0; a
value of 2 indicates that you
are running under ESA 3.1.3; a
value of 3 indicates that you
are running under SP 4.1.0; a
value of 4 indicates that you
are running under SP 4.2.0.
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
SMFID
SMF ID of system.
SMFID
Short
name
OSLV
OBTAIN
SAS
INT
COL,PC
F,4
I,4
N,10
F,4
U,1
N,10
P,4
N,8
T,4
C,8
C,4
C,4
Filters
F,8
C,4
Note: To correctly match the RMF field, you must include “RMF” in the character string. For example:
RIF(RMF= “RMF»4.1.0”)
Using paging and storage resource data (RPAG)
The RPAG keyword generates information about paging and storage resource activity on your system. As
you can see in the table below, there are 8 storage areas for MVS/370 systems, and 12 storage areas for
MVS/XA and MVS/ESA systems. Since one record is generated for each storage area, each interval
generates 8 or 12 records. If you want to select a specific storage area, use the OBTAIN command with
one of the following syntaxes:
Chapter 13. Report element tables
259
v REPORTIF (STORNAME (=ccc»»»»»»»»)), where ccc is the storage name whose character string must
be 12 characters long, padded with blanks on the right.
v REPORTIF (STORNAME (=ccc*)), where ccc is the storage name, and is padded with * (wild card) on
the right.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Table 30. Paging and Storage Resource Activity
Data type and length
EPILOG
field
Short
name
OBTAIN
SAS
INT
COL,PC
STORNAME
C,6
C,12
C,12
All SQA (above and below the
line) is reported simply as
SQA.
STORNAME
C,6
C,12
C,12
Virtual
Storage
Virtual storage (in kilobytes).
VIRT
F,4
I,4
N,10
Real
Average real frame counts (in
4K frames).
REAL
F,4
I,4
N,10
Fixed
Average fixed frame counts (in
4K frames).
FIXED
F,4
I,4
N,10
Non-Fix
Average non-fixed frame
counts (in 4K frames).
NONFIXED
F,4
I,4
N,10
storage
areas
Description
Dictionary
name
These areas are mapped for
MVS/370, MVS/XA and
MVS/ESA:
SQA
CSA
LPA
LSQA
Pvt
Free
Nuc
Total
On MVS/XA and MVS/ESA
systems, the designation Nuc
refers to the MVS nucleus
code, both above and below
the 16Mb line.
The following storage areas
above the 16Mb line are
reported for MVS/XA and
MVS/ESA systems:
EPvt
ECSA
ELPA
ESQA
EXP. STOR.
These extended areas do not
show any frame count, paging,
or swapping statistics; this is
because MVS itself does not
report them separately.
260
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Filters
Table 30. Paging and Storage Resource Activity (continued)
Data type and length
EPILOG
field
Description
Page/Sec In Average page-ins per second.
Dictionary
name
Short
name
PAGEIN
OBTAIN
SAS
INT
COL,PC
F,4
I,4
N,11
F,4
I,4
N,11
I,4
N,11
Filters
Page/Sec
Out
Average page-outs per second. PAGEOUT
----
Average page operations per
second (page-ins plus
page-outs).
PAGES
Swap
paging /sec
In
Average number of pages
swapped in per second.
SWAPIN
F,4
I,4
N,11
Swap
paging /sec
Out
Average number of pages
swapped out per second.
SWAPOUT
F,4
I,4
N,11
----
Average pages swapped per
second (swap-ins plus
swap-outs).
SWAPS
SPS
I,4
N,11
Installed
Frames
The total amount of expanded
storage installed (in frames).
(expanded storage only)
ESINST
ESIN
F,4
I,4
N,10
R
Online
Frames
The amount of expanded
storage online during the
interval (in frames). (expanded
storage only)
ESONLINE
ESON
F,4
I,4
N,10
R
Pages to
ES per Sec
ESFRREAL
The number of pages per
second sent to expanded
storage from real storage for
paging and swapping requests.
(expanded storage only)
ESRL
F,4
I,4
N,11
R
Pages
Migrated
per Sec
The number of pages per
second migrated from
expanded storage to auxiliary
storage. (expanded storage
only)
ESTOAUX
ESAX
F,4
I,4
N,11
R
Available
Frames:
The number of expanded
storage frames available during
the interval. (expanded storage
only)
Min
Minimum number of available
frames. (expanded storage
only)
ESMIN
ESMN
F,4
I,4
N,10
R
Max
Maximum number of available
frames. (expanded storage
only)
ESMAX
ESMX
F,4
I,4
N,10
R
Avg
Average number of available
frames. (expanded storage
only)
ESAVG
ESAV
F,4
I,4
N,10
R
PGS
R
Chapter 13. Report element tables
261
Table 30. Paging and Storage Resource Activity (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
Short
name
OBTAIN
SAS
INT
COL,PC
P,4
N,8
T,4
C,8
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
----
Machine flag. (EXP for systems MACHFLAG
with expanded storage, “»»»”
for systems without expanded
storage)
C,4
C,4
C,4
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
F,4
I,4
N,10
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
P,4
N,8
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
T,4
C,8
----
SMF ID of system.
SMFID
C,4
C,4
262
EDATTIME
EXTDTIME
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
F,8
F,8
C,4
Filters
Using page data set resource data (RPDS)
The RPDS keyword generates information about page data set resource activity on your system. One
record is generated for each page data set in your system. For example, if you have defined four page
data sets, each interval generates four records.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Table 31. Page data set Resource Activity
Data type and length
EPILOG
field
Description
OBTAIN
Dictionary
name
Short
name
SAS
INT
COL,PC
Filters
Space Type
Page space type. Valid types
are PLPA, common, duplex,
local, or swap.
SPCTYPE
STYP
C,6
C,6
C,6
R
Dev Adr
Device address (for MVS 370)
or device number (for MVS/XA
and MVS/ESA).
DEVADD
DEVA
C,4
C,3
C,3
R
VOLSER
Volume serial number.
VOLSER
VOLS
C,6
C,6
C,6
R
Dev Type
Device type.
DEVTYPE
DTYP
C,4
C,4
C,4
R
% Full
The average percent of slots
used on the data set.
PCTFULL
PFUL
F,4
I,4
N,11
R
% In Use
Percent of the time during the
interval when the data set was
considered busy by the ASM.
PCTINUSE
PUSE
F,4
I,4
N,11
R
I/O’s per
sec
Average number of I/O
requests for the data set per
second during the interval.
IOPERSEC
IOSC
F,4
I,4
N,11
R
Pages per
IO
Average number of pages per
I/O request.
PAGPERIO
PGIO
F,4
I,4
N,11
R
Trnsfr Time
Average time required to
complete a page transfer (in
seconds).
TRNSTIME
TRTM
F,4
I,4
N,11
R
----
If the page data set is found to PDSFLAG
be unusable, bit “0” of this field
is set to “1.”
F,4
X,4
H,8
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
P,4
N,8
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
T,4
C,8
EDATTIME
F,8
Chapter 13. Report element tables
263
Table 31. Page data set Resource Activity (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
Short
name
OBTAIN
SAS
----
Date and time of extraction
from the EPILOG datastore.
EXTDTIME
F,8
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
F,4
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
----
SMF ID of system.
SMFID
INT
COL,PC
I,4
N,10
P,4
N,8
T,4
C,8
C,4
C,4
Filters
F,8
C,4
Using performance group resource data (RPGN)
The RPGN keyword generates information about performance group resource activity on your system.
One record is generated for each performance group on your system. If a performance group is defined
with multiple periods in the IPS, then a separate record is generated for each performance period.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Table 32. Performance group resource activity
Data type and length
EPILOG
field
PGN
264
Description
Performance group number.
(For the EXTRACT command,
the dictionary name is PGN.)
Dictionary
name
Short
name
PERF
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
OBTAIN
SAS
INT
COL,PC
Filters
F,4
I,2
N,10
R
Table 32. Performance group resource activity (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
OBTAIN
Short
name
SAS
INT
COL,PC
Filters
PGP
Performance group period. If
PERIOD
the performance group is not
defined in the IPS with multiple
periods, then all activity is
considered to be period 1. (For
the EXTRACT command, the
dictionary name is PGP.)
PRD
F,4
I,4
N,10
R
SUs/Sec
Average SRM service units
used per second.
SUPERSEC
SUPS
F,4
I,4
N,10
R
Avg Resp
Average response time for a
transaction (in seconds). (For
the EXTRACT command, the
dictionary name is RESP.)
AVGRESP
ARSP
F,4
I,4
N,11
R
Avg Trx
AVGCTRX
Average number of
transactions concurrently active
during the time period.
AVTX
F,4
I,4
N,11
R
Swaps/Trx
Average number of swaps per
transaction during the interval.
SWAPERTX
SWTX
F,4
I,4
N,11
R
%Res
Percent of the time that at
least one transaction was in
storage.
PCTRES
PRES
F,4
I,4
N,11
R
Avg WKST
Average working set size for
an address space in the
performance group (displayed
in kilobytes or megabytes).
AVGWKST
WKST
F,4
I,4
N,10
R
Avg In
Average number of resident
address spaces from the
performance group.
AVGTRXIN
TXIN
F,4
I,4
N,11
R
Total Stg
Average amount of working set TOTSTGK
storage used by the
performance group over the
time period (displayed in
kilobytes or megabytes).
TSTG
F,4
I,4
N,10
R
SRB%
Percent of CPU used by SRBs SRBPCT
in the performance group over
the time period.
SRBP
F,4
I,4
N,11
R
TCB%
Percent of CPU used by TCBs
in the performance group over
the time period.
TCBP
F,4
I,4
N,11
R
PIOS
F,4
I,4
N,11
R
C,1
C,1
C,1
R
TCBPCT
EXCPs /sec Average EXCPs per second for PIOPRSEC
the performance group (as
counted by SMF).
----
Report or Control Flag.
REPTCNTL
Indicates whether the
performance group is a report
performance group or a control
performance group (R or C).
Chapter 13. Report element tables
265
Table 32. Performance group resource activity (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
Short
name
OBTAIN
SAS
INT
COL,PC
F,4
X,4
H,8
P,4
N,8
T,4
C,8
I,4
N,10
P,4
N,8
----
Indicates whether the CPU
PGNFLAG
speed was recorded for the
interval (0 = recorded, 64 = not
recorded). Since OBTAIN
values are formatted as
hexadecimal, the “not
recorded” indicator will appear
as 40.
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
EXTDTIME
F,8
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
F,4
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
266
EDATTIME
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
F,8
Filters
Table 32. Performance group resource activity (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
----
Time of interval start If intervals STIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
SMF ID of system.
Short
name
SMFID
OBTAIN
SAS
C,4
INT
COL,PC
T,4
C,8
C,4
C,4
Filters
Using swap data set resource data (RSDS)
The RSDS keyword generates information about swap data set resource activity on your system. One
record is generated for each swap data set in your system. For example, if you have defined two swap
data sets, each interval generates two records.
The following table shows a description of each EPILOG field that is extracted, along with the
corresponding dictionary name and data type.
Table 33. Swap data set resource activity
Data type and length
Short
Dictionary name name
EPILOG field
Description
Dev Adr
Device address (for
DEVADD
MVS/370) or device number
(for MVS/XA and
MVS/ESA).
VOLSER
Volume serial number.
Dev Type
OBTAIN
SAS
INT
COL,PC
Filters
DEVA
C,4
C,3
C,3
R
VOLSER
VOLS
C,6
C,6
C,6
R
Device type.
DEVTYPE
DTYP
C,4
C,4
C,4
R
% Full
Average percent of swap
sets in use during the time
period.
PCTFULL
PFUL
F,4
I,4
N,11
R
% In Use
Percent of the time during
PCTINUSE
the interval when the data
set was considered busy by
the ASM.
PUSE
F,4
I,4
N,11
R
IO per second Average number of I/O
requests for the data set
per second during the
interval.
IOPERSEC
IOSC
F,4
I,4
N,11
R
Avg Srv Time
Average service time (in
seconds per page).
TRNSTIME
TRTM
F,4
I,4
N,11
R
----
Indicates the swap data set SDSFLAG
condition. If the X’80’ bit is
on, ASM has found the
data set to be unusable.
F,4
X,4
H,8
Chapter 13. Report element tables
267
Table 33. Swap data set resource activity (continued)
Data type and length
Short
Dictionary name name
OBTAIN
EPILOG field
Description
----
Date of interval end. If
EDATE
intervals have been
combined, this figure
reflects the entire combined
interval. (OBTAIN only)
----
EDATTIME
Date and time of interval
end. If intervals have been
combined, this figure
reflects the entire combined
interval. (SAS only)
----
Date of interval end. If
ETIME
intervals have been
combined, this figure
reflects the entire combined
interval. (OBTAIN only)
----
Date and time of extraction EXTDTIME
from the EPILOG datastore.
F,8
----
NUMICMB
Number of EDS records
extracted for the
observation. This value is
particularly useful when you
are combining records and
also applying exception
criteria with the SELECTIF
keyword. Under those
circumstances, this value
indicates how many EDS
records passed the
exception criteria and were
included in the observation.
F,4
----
Date of interval end. If
SDATE
intervals have been
combined, this figure
reflects the entire combined
interval. (OBTAIN only)
----
Date and time of interval
SDATTIME
start. If intervals have been
combined, this figure
reflects the entire combined
interval. (SAS only)
F,8
----
SMF ID of system.
C,4
----
Time of interval start. If
STIME
intervals have been
combined, this figure
reflects the entire combined
interval. (OBTAIN only)
268
SMFID
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
SAS
INT
COL,PC
P,4
N,8
T,4
C,8
I,4
N,10
P,4
N,8
C,4
C,4
T,4
C,8
F,8
Filters
Using SRM MPL statistics (RSRM)
The RSRM keyword generates information about MPL adjustment values on your system. One record is
generated for each value of SRMVALS in the interval. The following table shows a description of each
EPILOG field that is extracted, along with the corresponding dictionary name and data type.
Table 34. SRM happy value statistics
Data type and length
EPILOG
field
Description
OBTAIN
Dictionary
name
Short
name
SAS
INT
COL,PC
Filters
SRM
parameters
and
observed
values
Possible values are:
HIGH THRESHOLD
OBSERVED MAX
OBSERVED AVG
OBSERVED MIN
LOW THRESHOLD
SRMVALS
SRVS
C,14
C,14
C,14
R
CPU
Average percent CPU
utilization, as calculated by
SRM.
SRCPU
ACPU
F,4
I,4
I,10
R
UIC
Average unreferenced interval
count, as calculated by SRM.
SRUIC
AUIC
F,4
I,4
N,10
R
ASMQ
Average ASM queue length, as SRASMQ
calculated by SRM.
ASMQ
F,4
I,4
I,10
R
Page Fault
Page fault rate (non-swap,
non-VIO page-ins plus
reclaims) (in pages per
second).
SRPAGFLT
PFLT
F,4
I,4
N,10
R
Dmand
Pging
Demand paging rate
(non-swap, non-VIO page-ins
plus page-outs) (in pages per
second).
SRDMDPAG
DPGR
F,4
I,4
N,10
R
Page Delay
Page delay time (excluding
swap paging) in milliseconds,
as calculated by SRM.
SRPAGDLY
PGDL
F,4
I,4
N,10
R
SRFIXB16
FXB
F,4
I,4
N,10
R
F,4
I,4
N,10
R
I,4
I,10
P,4
N,8
% of
Percent of the first 16M of real
storage that was fixed (i.e.,
Frames
Fixed Below non-pageable).
16M
% of
Frames
Fixed Total
Percent of total real storage
that was fixed (i.e.,
non-pageable).
SRFIXTOT
FXT
----
Expanded storage migration
age (for ESA environments).
(OBTAIN only)
ESMIGRAT
EMIG
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
EDATTIME
F,8
Chapter 13. Report element tables
269
Table 34. SRM happy value statistics (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
Short
name
OBTAIN
SAS
INT
COL,PC
T,4
C,8
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
----
Date and time of extraction
from the EPILOG datastore.
EXTDTIME
F,8
----
Machine flag. (309X for 3090
processors, »»»» for non-3090
processors)
MACHFLAG
C,4
C,4
C,4
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
F,4
I,4
N,10
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
P,4
N,8
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
T,4
C,8
----
SMF ID of system.
SMFID
C,4
C,4
Filters
F,8
C,4
Using swap activity data (RSWA)
The RSWA keyword generates information about swapping activity on your system. One record is
generated for each interval. The following table shows a description of each EPILOG field that is extracted,
along with the corresponding dictionary name and data type.
270
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 35. Swapping activity
Data type and length
EPILOG
field
Description
Logical
See Logical Swaps/Sec.
Physical
Dictionary
name
OBTAIN
Short
name
SAS
INT
COL,PC
Filters
Rate of physical swap-outs per PSWPRSEC
second (non-3090 only).
PSW
F,4
I,4
N,11
R
Percent
logical that
became
Physical
Percent of logical swaps that
became physical swaps.
(non-3090 only)
LTPPRSEC
LSPS
F,4
I,4
N,11
R
Average
pages per
swap:
Average number of pages per
swap-out and swap-in.
(non-expanded storage only)
Out
Average pages per swap-out.
PGPRSWPO
PPSO
F,4
I,4
N,11
R
In
Average pages per swap-in.
PGPRSWPI
PPSI
F,4
I,4
N,11
R
Physical
Swaps Per
Sec:
Physical swaps are those
made directly from central
storage to auxiliary storage,
without the involvement of
expanded storage. The
difference between direct and
transition physical swaps
depends on whether the
address space was logically
swapped before being
physically swapped. (3090
only)
Direct
DSWPPHSC
Direct swaps indicate the
number of swap-outs that were
not marked as logically
swapped when the physical
swap was made.
DSWP
F,4
I,4
N,11
R
Transition
TSWPSEC
Transition swaps indicate the
number of swap-outs that were
already marked as logically
swapped when the physical
swap was made.
TSWP
F,4
I,4
N,11
R
Logical
Swaps/Sec
LSWPRSEC
The total number of address
spaces that were marked as
logically swapped during the
interval, divided by the number
of seconds in the interval.
(Some of these may have
subsequently transitioned to
auxiliary or expanded storage
during the interval.)
LSW
F,4
I,4
N,11
R
Expanded
Storage:
Shows the swap-out rate from
central storage to expanded
storage, and from expanded
storage to auxiliary storage.
(3090 only)
Chapter 13. Report element tables
271
Table 35. Swapping activity (continued)
Data type and length
EPILOG
field
Description
OBTAIN
Dictionary
name
Short
name
SAS
INT
COL,PC
Filters
Direct
Direct swap-outs are those
physically swapped from
central storage to expanded
storage without first being
marked as logical swaps.
ESDSWPSC
DESW
F,4
I,4
N,11
R
Trans
ESTSWPSC
Transition swaps are those
which were already marked as
logical swaps when the
physical swap from central
storage to expanded storage is
done.
TESW
F,4
I,4
N,11
R
Migrated
ESASWPSC
Migrated swaps are those in
which address spaces in
expanded storage are migrated
to auxiliary storage. (An
address space may have
arrived in expanded storage
either through a direct or
transition swap.)
ESMG
F,4
I,4
N,11
R
----
Architecture Flag. XA»=
MVS/XA and MVS/ESA, 370=
MVS/370.
C,3
C,3
C,3
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
P,4
N,8
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
T,4
C,8
----
Date and time of extraction
from the EPILOG datastore.
EXTDTIME
F,8
----
Machine flag. (309X for 3090
processors, »»»» for non-3090
processors)
MACHFLAG
C,4
C,4
C,4
272
DSXAFLAG
EDATTIME
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Table 35. Swapping activity (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
----
SMF ID of system.
SMFID
Short
name
OBTAIN
SAS
INT
COL,PC
F,4
I,4
N,10
P,4
N,8
T,4
C,8
C,4
C,4
Filters
F,8
C,4
Using swap reason data (RSWR)
The RSWR keyword generates information about swap reasons on your system. The following table
shows a description of each EPILOG field that is extracted, along with the corresponding dictionary name
and data type.
Chapter 13. Report element tables
273
Table 36. Swap reasons
Data type and length
EPILOG
field
Swap
Category
274
Description
Dictionary
name
SWPCATG
This field may include any of
the swap reasons described
below:
Terminal Wait - Terminal
waiting for input or output
buffers.
Detected Long Think - An
address space that had been
logically swapped out had to
be physically swapped out
before it was ready to execute
again.
Long Wait - User-requested
long waits.
Detected Wait - User address
spaces that have not executed
for 8 SRM seconds or 2
seconds, whichever is less (in
current versions of MVS), and
have not issued a long WAIT.
Exchange - A user in a domain
that must be swapped out to
allow another user in that
domain to be swapped in.
Unilateral - The MPL of a
domain exceeded the target
MPL.
OBTAIN
Short
name
SAS
INT
COL,PC
Filters
CAT
C,21
C,21
C,21
R
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 36. Swap reasons (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
Request - Because a user had SWPCATG
to be swapped because of an
operator command or system
action, such as VARY
STORAGE OFFLINE.
Enq Exchange - To make room
for users enqueued on a
resource required by other
users.
Aux Storage Shortage Because of a shortage of local
page slots.
Real Storage Shortage Because of a shortage of real
pageable frames.
Transition to Nonswap - An
address space has been made
non-swappable.
Central Storage - An address
space was swapped out to
improve central storage
availability. (MVS/SP 4.2 and
above)
System Paging - An address
space was swapped out to
reduce the system page fault
rate. (MVS/SP 4.2 and above)
OBTAIN
Short
name
SAS
INT
COL,PC
Filters
CAT
C,21
C,21
C,21
R
Chapter 13. Report element tables
275
Table 36. Swap reasons (continued)
Data type and length
EPILOG
field
----
Description
Direct swaps indicate the
number of swap-outs of
address spaces that were not
marked as logically swapped
when the physical swap was
made.
OBTAIN
Short
name
SAS
INT
COL,PC
Filters
CAT
C,24
C,21
C,21
R
PSWPRSEC
PSW
F,4
I,4
N,11
R
DSWPPHSC
DSWP
F,4
I,4
N,11
R
SWPCATG
Out Too Long - An address
space was swapped out to
enable the swap-in of an
address space that was
swapped out too long.
(MVS/SP 4.2 and above)
APPC Wait - An address space
was swapped out of an idle
APPC address space.
(MVS/SP 4.2 and above)
Total - Total of all the above.
For non-3090 processors, a
separate observation is
generated for each swap
category, even if there were no
swaps of that type during the
interval.
On 3090 displays, if there were
no swaps for a particular
category during the interval, no
observation is generated. If an
observation is generated for a
category and there are zero
values for all swap rates, this
indicates that there were
swaps of that type during the
interval, but that the rate was
less than .01 when the data
was averaged over the interval.
Phys Swaps For systems without expanded
per sec:
storage, this value indicates
the number of physical swaps
caused by each of the above
reasons.
For systems using expanded
storage, physical swaps are
those made directly from real
storage to auxiliary storage,
without the involvement of
expanded storage. Physical
swaps are broken down into
two categories: Direct and
Transition.
276
Dictionary
name
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 36. Swap reasons (continued)
Data type and length
EPILOG
field
Description
Transition swaps indicate the
swap-outs of address spaces
that were already marked as
logically swapped when the
physical swap was made.
OBTAIN
Dictionary
name
Short
name
SAS
INT
COL,PC
Filters
TSWPSEC
TSWP
F,4
I,4
N,11
R
LSW
F,4
I,4
N,11
R
Logical
Swaps per
Second
LSWPRSEC
The total number of address
spaces that were marked as
logically swapped during the
interval, divided by the number
of seconds in the interval.
(Some of these may have
subsequently transitioned to
auxiliary or expanded storage
during the interval.) (expanded
storage only)
Expanded
Storage:
Indicates the rate of three
types of swaps (in swaps per
second). (expanded storage
only)
Direct
Direct swaps are those which
RSM sends directly from real
storage to expanded storage,
without first being marked as
logical swaps.
ESDSWPSC
DESW
F,4
I,4
N,11
R
Trans
Transition swaps are those
which RSM first made logical
swaps, and then became
physical swaps to expanded
storage.
ESTSWPSC
TESW
F,4
I,4
N,11
R
Migrated
Migrated swaps are those
ESASWPSC
which migrated from expanded
storage to auxiliary storage.
ESMG
F,4
I,4
N,11
R
----
Date of interval end. If intervals EDATE
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
P,4
N,8
----
Date and time of interval end.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
----
Time of interval end. If intervals ETIME
have been combined, this
figure reflects the entire
combined interval. (OBTAIN
only)
T,4
C,8
----
Date and time of extraction
from the EPILOG datastore.
EDATTIME
EXTDTIME
F,8
F,8
Chapter 13. Report element tables
277
Table 36. Swap reasons (continued)
Data type and length
EPILOG
field
Description
Dictionary
name
----
Machine flag. (309X for 3090
processors, »»»» for non-3090
processors)
----
NUMICMB
Number of EDS records
extracted for the observation.
This value is particularly useful
when you are combining
records and also applying
exception criteria with the
SELECTIF keyword. Under
those circumstances, this value
indicates how many EDS
records passed the exception
criteria and were included in
the observation.
----
Date of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
SDATE
----
Date and time of interval start.
If intervals have been
combined, this figure reflects
the entire combined interval.
(SAS only)
SDATTIME
----
Time of interval start. If
intervals have been combined,
this figure reflects the entire
combined interval. (OBTAIN
only)
STIME
----
SMF ID of system.
SMFID
Short
name
MACHFLAG
OBTAIN
SAS
INT
COL,PC
C,4
C,4
C,4
F,4
I,4
N,10
P,4
N,8
T,4
C,8
C,4
C,4
Filters
F,8
C,4
Using virtual lookaside facility statistics (RVLF)
The RVLF keyword generates information about VLF class activity on your system. One record is
generated for each interval. The following table shows a description of each EPILOG field that is extracted,
along with the corresponding dictionary name and data type.
Table 37. VLF class statistics
Data type and length
EPILOG
field
OBTAIN
Description
Dictionary
name
Short
name
SAS
INT
COL,PC
Filters
VLF Class
The name of this class.
VLFCLASS
VLFC
F,4
I,4
C,10
R
Used
The number of pages of virtual VLFVSUSD
storage used by this class.
VLFU
F,4
I,4
N,10
R
Maximum
The maximum number of
virtual storage pages that this
class can use.
VLFM
F,4
I,4
N,10
R
278
VLFMXSTG
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 37. VLF class statistics (continued)
Data type and length
EPILOG
field
Description
OBTAIN
Dictionary
name
Short
name
SAS
INT
COL,PC
Filters
VLFLRGST
VLFL
F,4
I,4
N,10
R
VLFA
F,4
I,4
N,10
R
Object
The size of the largest object
which VLF attempted to load
into virtual storage in this
class.
Add
The number of objects in this
VLFADDS
class that were added to virtual
storage during the interval.
Delete
The number of objects in this
class that were deleted from
virtual storage during the
interval.
VLFDELES
VLFD
F,4
I,4
N,10
R
Trim
The number of objects in this
class that were trimmed from
virtual storage during the
interval.
VLFTRIMS
VLFT
F,4
I,4
N,10
R
Reads/sec
The number of times cache
was searched for an object in
this class during the interval,
divided by the number of
seconds in the interval.
VLFRATE
VLFR
F,4
I,4
N,10
R
Hit %
The number of times an object
in this class was found in
cache divided by the number
of times cache was searched
for objects in this class,
expressed as a percentage in
tenths of a percent.
VLFHITS
VLFH
F,4
I,4
N,10
R
----
The storage used by objects in VLFFULL
this class, divided by the
maximum storage available for
objects in this class, expressed
as a percentage in tenths of a
percent.
VLFF
F,4
I,4
N,10
R
Using workload degradation data
Workload degradation data is available for performance groups, batch jobs, started tasks, and TSO user
sessions. The keywords for selecting workloads are described in “Workload keywords” on page 171.
For each reporting interval, a variety of information is extracted for the workload, including the amount of
time that the workload spent waiting for each wait reason. (A list of wait reasons is shown in Table 14 on
page 146.) One record is generated for each wait reason, even if there was no delay due to that wait
reason. If there was more than one type of I/O or enqueue wait reason, multiple records are generated
(one for each type of I/O or enqueue wait).
The workload tables
Data extracted from the EPILOG datastore by the EXTRACT or COMPEXT commands is written out to
one or more SAS data sets, each of which corresponds to a system resource or a workload type. Although
Chapter 13. Report element tables
279
the data extracted by the OBTAIN command can be written to user-defined data sets, the tables in this
section are arranged by SAS data set name for easy reference.
There are 4 types of workloads (performance groups, batch jobs, started tasks, and TSO user sessions).
Table 38 shows the workload tables in these data sets, the type of data they hold, and the command
keywords that correspond to them. Each workload table name corresponds to a SAS data set name.
Table 38. Workload tables for the OBTAIN and EXTRACT commands
Workload table
Type of data
EPILOG keyword
BTCHDETL
Batch job degradation data
JOB, PROGRAM,
ACCOUNT, or CLASS
PGNDETL
Performance group degradation data
PGN, SYMBOLIC, or
SYSTEM
STCDETL
Started task degradation data
STC, PROGRAM, or
ACCOUNT
TSODETL
TSO user session degradation data
TSO, PROGRAM, or
ACCOUNT
The SAS data sets
In most instances, the EXTRACT command writes to only one SAS data set. For example, if you specify
EXTRACT PGN LASTMONTH
the SAS interface invokes PROC KEPILGPM, extracts information about performance groups, and writes
that information out to the PGNDETL data set.
The exceptions to this occur with the PROGRAM and ACCOUNT keywords. Since these keywords may
select batch jobs, started tasks, and TSO user sessions, they may update more than one SAS data set
with a single EXTRACT or command.
Each invocation of the KEPILGPM procedure resets the 16 SAS data sets. If the EXTRACT command in
the procedure does not write any data to a particular data set, that data set contains no records after the
procedure is executed.
Keywords
Workload degradation data can be specified on the EXTRACT and OBTAIN commands by a number of
different keywords. See “Workload keywords” on page 171 to determine which is the most appropriate
keyword for the kind of report you are writing.
Table contents
The table contents are described in “Table contents” on page 234. You will notice that the tables in this
section are slightly different than in format from the tables that describe resource data. The EPILOG
display field is not applicable in most cases, and does not appear in the workload tables.
Performance group degradation data (PGNDETL)
If you use the EXTRACT command to generate information about performance groups, data is written out
to a data set named PGNDETL. If you use the OBTAIN command to generate information about
performance groups, data is written to the user-specified data set. In both methods, one record is
generated for each wait reason, and a separate set of records is generated for reporting interval for each
performance group. If you have requested individual period data, a separate set of records is generated
for each performance period.
The following keywords are used to generate performance group degradation data:
PERFGROUP (PGN)
280
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
SYMBOLIC (SYM)
PGPERIOD (PGP)
SYSTEM (SYS)
For a detailed description of how these keywords are used, see theEPILOG for MVS Basic User’s Guide.
The data elements associated with the PGNDETL data set are described below.
Table 39. Performance group degradation data
Data type and length
Dictionary
name
Description
Short
name
OBTAIN
SAS
INT
COL,PC
C,4
C,4
C,4
P,4
N,8
T,4
C,8
C,8
C,8
DEVADD
Device address for I/O wait reasons (in cuu
format) (for MVS/370) or device number (in
nnnn format) for MVS/XA and MVS/ESA.
EDATE
Date of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
EDATTIME
Date and time of interval end. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
ETIME
Time of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
ENQNAME
Major enqueue names for enqueue waits
(see Notes following the table).
C,8
EXTDTIME
Date and time of extraction from the EPILOG
datastore.
F,8
NUMICMB
Number of EDS records extracted for the
observation. This value is particularly useful
when you are combining records and also
applying exception criteria with the
SELECTIF keyword. Under those
circumstances, this value indicates how many
EDS records passed the exception criteria
and were included in the observation.
F,4
I,4
N,10
NUMTRX
Number of transactions in the interval. (If
multiple intervals have been combined, this
figure reflects the entire combined interval.)
F,4
D,4
N,10
PGN
Performance group number.
F,4
I,2
N,10
PGNTYPE
Performance group type (see Notes following
the table).
C,11
C,1
C,1
PGP
Performance period. Indicates the
performance period(s) specified, if any.
F,4
I,2
N,10
Filters
F,8
Chapter 13. Report element tables
281
Table 39. Performance group degradation data (continued)
Data type and length
Dictionary
name
PRODIDX
Description
Short
name
Productivity index. This is a general indicator
of how severely the performance group is
being degraded. It is derived by dividing the
amount of time spent doing productive work
(such as active I/O or using CPU) by the total
amount of time spent in productive and
unproductive activities. Idle time (such as
STIMER waits) is disregarded in the
calculation, since it is not relevant to
performance degradation.
OBTAIN
SAS
INT
COL,PC
F,4
D,4
N,10
P,4
N,8
T,4
C,8
For detailed information about how execution
states are categorized as productive,
unproductive, and idle, and instructions about
how you can modify these definitions, see
“The PRODUCTS command” on page 114.
SDATE
Date of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SDATTIME
Date and time of interval start. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
STIME
Time of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SMFID
SMF ID of the system on which the workload
ran.
C,4
C,4
C,4
SYMBNAME
Symbolic name defined for the performance
group (if any).
C,8
C,8
C,8
TOTCPUTM
Total CPU time used by all transactions
during the interval (in seconds). (If multiple
intervals have been combined, this figure
reflects the entire combined interval.)
F,4
D,4
N,10
TRXELAP
Total transaction elapsed time for all
transactions that ended during the interval (in
seconds).
F,4
D,8
N,10
TRXRESP
Average transaction response time (in
seconds).
F,4
D,4
N,10
TRXTHINK
Average transaction think time (in seconds).
F,4
D,4
N,10
VOLSER
Volume serial number for I/O wait reasons.
C,6
C,6
C,6
WAITNAME
Full name of wait reason (see “Wait reasons
(WAITNAME)” on page 283).
C,10
C,20
C,20
WAITPCT
Percent of average transaction response time
spent on this wait reason.
F,4
D,4
N,10
WAITTIME
Average transaction time spent on this wait
reason (in seconds).
F,4
D,4
N,10
282
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Filters
Table 39. Performance group degradation data (continued)
Data type and length
Dictionary
name
WKLNAME
Short
name
Description
Workload name (see Notes following the
table).
OBTAIN
SAS
INT
COL,PC
C,12
C,12
C,12
Filters
Notes
The allowed values for ENQNAME, PGNTYPE, WAITNAME, and WKLNAME are as follows:
Enqueue names (ENQNAME)
EPILOG for MVS comes with a set of default major enqueue names, which may have been modified or
replaced during product installation. (For details, see the description of the ENQNAME keyword in
theEPILOG for MVS Customization Guide.)
These defaults are as follows:
^SYSDSN ^SYSIEWLP ^SYSIEA01 ^SYSVSAM ^SYSIEFSD ^SYSVTOC ^SYSIGGV1 ^SYSZJES2
^SYSIGGV2 ^SYSZVARY ^SYSIKJBC ^MISC SYS ^SYSSMF01 ^MISC USR
These major enqueue names are described in fuller detail in Table 17 on page 150.
Performance group types (PGNTYPE)
For OBTAIN, the values are :
C = Control
R = Report
For EXTRACT and COMPEXT, the values are:
CONTROL PGN
REPORT PGN
Wait reasons (WAITNAME)
ACTIVE I/O
APPCWAIT
AUX STORE
CENTSTOR
COMMON IN
CPU USED
CPU WAIT
DETECTED
DISK MOUNT
ECB WAIT
ECB+STIMER
ENQ EXCHNG
ENQUEUE
EXCHANGE
HSM BACKUP
HSM CONTRL
HSM DELETE
HSM HLIST
HSM JES CI
HSM MIGRTE
HSM RECALL
HSM RECOVR
JES CANCEL
JES DELETE
JES REQUE
JES STATUS
JES SYSOUT
LOGICAL CH
LONG WAIT
MVS LOCK
OUTLONG
PAGE IN
QUEUED I/O
REAL STORE
REQUEST
RESRVE I/O
SRM DLY (MPL)
SRM RTO
STAGING
STIMER
SWAP IN
SYSPAGE
TAPE MOUNT
TERM IN
TERM OUT
TRANSITION
UNILATERAL
WTOR WAIT
These wait reasons are described in detail in Table 14 on page 146.
Chapter 13. Report element tables
283
Workload names (WKLNAME)
ALL SYSTEM
PERIOD
SYMBOLIC PGN
PGN
Batch job data (BTCHDETL)
The following keywords are used to select batch jobs in the EXTRACT and OBTAIN commands:
JOBNAME (JOB)
ACCOUNT (ACCT)
PROGRAM (PGM)
CLASS (CLS)
In addition, you can focus on different levels of data with the following keywords:
STEP
INTERVAL
For a detailed description of how these keywords are used, see theEPILOG for MVS Basic User’s Guide.
The data elements associated with the BTCHDETL data set are described below.
Table 40. Batch job degradation data
Data type and length
Dictionary
name
Description
ACCTCDE
Short
name
OBTAIN
SAS
INT
COL,PC
Account code of job.
C,12
C,12
C,12
CPUTRXTM
Total CPU time for the job or job step (in
CPU seconds).
F,4
D,4
N,10
DEVADD
Device address for I/O wait reasons (in cuu
format) (for MVS/370) or device number (in
nnnn format) for MVS/XA and MVS/ESA.
C,4
C,4
C,4
EDATE
Date of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
P,4
N,8
EDATTIME
Date and time of interval end. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
ETIME
Time of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
T,4
C,8
ENQNAME
Major enqueue name for enqueue waits (see
“Enqueue names (ENQNAME)” on page
283).
C,8
C,8
C,8
EXTDTIME
Date and time the record was extracted from
the EPILOG datastore.
F,8
HIGHSTEP
If several job steps have been combined,
shows the highest step number for the job.
F,4
U,2
N,5
JESCLASS
Batch job execution class.
C,4
C,4
C,4
JESNUM
JES job number.
C,4
C,5
C,5
JOBNAME
Batch job name.
C,8
C,8
C,8
284
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Filters
Table 40. Batch job degradation data (continued)
Data type and length
Dictionary
name
Description
Short
name
OBTAIN
SAS
INT
COL,PC
NUMICMB
Number of EDS records extracted for the
observation. This value is particularly useful
when you are combining records and also
applying exception criteria with the
SELECTIF keyword. Under those
circumstances, this value indicates how many
EDS records passed the exception criteria
and were included in the observation.
F,4
I,4
N,10
PGMNAME
Batch job program name.
C,8
C,8
C,8
PROCNAME
Batch job procedure name.
C,8
C,8
C,8
PRODIDX
Productivity index. This is a general indicator
of how severely the performance group is
being degraded.
F,4
D,4
N,10
F,4
D,4
N,10
P,4
N,8
T,4
C,8
Filters
For detailed information about how execution
states are categorized as productive,
unproductive, and idle, and instructions about
how you can modify these definitions, see
“The PRODUCTS command” on page 114.
RDRTIME
Elapsed time that the job was on the input
queue before either the job started, or, if the
data was extracted at the step level, the step
started (in seconds).
SDATE
Date of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SDATTIME
Date and time of the interval start. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
STIME
Time of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SMFID
SMF ID of the system on which the job ran.
C,4
C,4
C,4
STEPNAME
Batch job step name.
C,8
C,8
C,8
TRXELAP
Total elapsed time for the job (in seconds).
F,4
D,8
N,10
VOLSER
Volume serial number for I/O waits.
C,6
C,6
C,6
WAITNAME
Full name of wait reason (see “Wait reasons
(WAITNAME)” on page 283).
C,10
C,20
C,20
WAITPCT
Percent of total job elapsed time spent on
this wait reason.
F,4
D,4
N,10
WAITTIME
Total job time spent on this wait reason (in
seconds).
F,4
D,4
N,10
WKLNAME
Workload name (see notes following the
table).
C,12
C,12
C,12
F,8
Chapter 13. Report element tables
285
Notes
The allowed values for WKLNAME are as follows:
Workload names (WKLNAME)
ACCOUNT NUM
PROGRAM NAME
BATCH CLASS
BATCH JOB
Started task degradation data (STCDETL)
The following keywords are used to select started tasks in the EXTRACT and OBTAIN commands:
STARTTSK (STC)
PROGRAM (PGM)
In addition, you can control the granularity of the data with the following keywords:
STEP
INTERVAL
For a detailed description of how these keywords are used, see theEPILOG for MVS Basic User’s Guide.
The data elements associated with the STCDETL data set are described below.
Table 41. Started task degradation data
Data type and length
Dictionary
name
Description
Short
name
OBTAIN
SAS
INT
COL,PC
CPUTRXTM
Total CPU time for the task or step (in CPU
seconds).
F,4
D,4
N,10
DEVADD
Device address for I/O wait reasons (in cuu
format) (for MVS/370) or device number (in
nnnn format) for MVS/XA and MVS/ESA.
C,4
C,4
C,4
EDATE
Date of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
P,4
N,8
EDATTIME
Date and time of interval end. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
ETIME
Time of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
T,4
C,8
ENQNAME
Major enqueue name for enqueue waits (see
“Enqueue names (ENQNAME)” on page
283).
C,8
C,8
C,8
EXTDTIME
Date and time the record was extracted from
the EPILOG datastore.
F,8
HIGHSTEP
If several steps in a task have been
combined, shows the highest step number for
the job.
F,4
U,2
N,5
JESNUM
JES job number.
C,4
C,5
C,5
286
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Filters
Table 41. Started task degradation data (continued)
Data type and length
Dictionary
name
Description
Short
name
OBTAIN
SAS
INT
COL,PC
NUMICMB
Number of EDS records extracted for the
observation. This value is particularly useful
when you are combining records and also
applying exception criteria with the
SELECTIF keyword. Under those
circumstances, this value indicates how many
EDS records passed the exception criteria
and were included in the observation.
F,4
I,4
N,10
PGMNAME
Started task program name.
C,8
C,8
C,8
PROCNAME
Started task procedure name.
C,8
C,8
C,8
PRODIDX
Productivity index. This is a general indicator
of how severely the performance group is
being degraded.
F,4
D,4
N,10
P,4
N,8
T,4
C,8
Filters
For detailed information about how execution
states are categorized as productive,
unproductive, and idle, and instructions about
how you can modify these definitions, see
“The PRODUCTS command” on page 114.
SDATE
Date of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SDATTIME
Date and time of the interval start. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
STIME
Time of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SMFID
SMF ID of the system on which the task was
run.
C,4
C,4
C,4
STCNAME
Name of the started task.
C,8
C,8
C,8
STEPNAME
Name of the step (for multi-step started
tasks).
C,8
C,8
C,8
TRXELAP
Total elapsed time for the started task or step
(in seconds).
F,4
D,8
N,10
VOLSER
Volume serial number for I/O waits.
C,6
C,6
C,6
WAITNAME
Full name of wait reason (see “Wait reasons
(WAITNAME)” on page 283).
C,10
C,20
C,20
WAITPCT
Percent of total elapsed time spent on this
wait reason.
F,4
D,4
N,10
WAITTIME
Total time spent on this wait reason (in
seconds).
F,4
D,4
N,10
WKLNAME
Workload name (see notes following the
table).
C,12
C,12
C,12
F,8
Chapter 13. Report element tables
287
Notes
The allowed values for WKLNAME are as follows:
Workload names (WKLNAME)
STARTED TASK
PROGRAM NAME
TSO user session degradation data (TSODETL)
The following keywords are used to select TSO sessions in the EXTRACT and OBTAIN commands:
TSOUSER (TSO)
PROGRAM (PGM)
ACCOUNT (ACCT)
In addition, you can control the granularity of the extracted data with the INTERVAL keyword. For a
detailed description of how these keywords are used in EPILOG, see the EPILOG for MVS Basic User’s
Guide.
The data elements associated with the TSODETL data set are also described.
Table 42. TSO user session degradation data
Data type and length
Dictionary
name
Description
ACCTCDE
Short
name
OBTAIN
SAS
INT
COL,PC
Account code of the started task (if any).
C,12
C,12
C,12
CPUTRXTM
Total CPU time for the TSO session (in CPU
seconds).
F,4
D,4
N,10
DEVADD
Device address for I/O wait reasons (in cuu
format) (for MVS/370) or device number (in
nnnn format) for MVS/XA and MVS/ESA.
C,4
C,4
C,4
EDATE
Date of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
P,4
N,8
EDATTIME
Date and time of interval end. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
ETIME
Time of interval end. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
T,4
C,8
ENQNAME
Major enqueue name for enqueue waits (see
“Enqueue names (ENQNAME)” on page
283).
C,8
C,8
C,8
EXTDTIME
Date and time the record was extracted from
the EPILOG datastore.
F,8
JESNUM
JES job number.
C,4
C,5
C,5
288
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
F,8
Filters
Table 42. TSO user session degradation data (continued)
Data type and length
Dictionary
name
Description
Short
name
OBTAIN
SAS
INT
COL,PC
NUMICMB
Number of EDS records extracted for the
observation. This value is particularly useful
when you are combining records and also
applying exception criteria with the
SELECTIF keyword. Under those
circumstances, this value indicates how many
EDS records passed the exception criteria
and were included in the observation.
F,4
I,4
N,10
PRODIDX
Productivity index. This is a general indicator
of how severely the TSO session was being
degraded.
F,4
D,4
N,10
P,4
N,8
T,4
C,8
Filters
For detailed information about how execution
states are categorized as productive,
unproductive, and idle, and instructions about
how you can modify these definitions, see
“The PRODUCTS command” on page 114.
SDATE
Date of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SDATTIME
Date and time of the interval start. If multiple
intervals have been combined, this figure
reflects the entire combined interval. (SAS
only)
STIME
Time of interval start. If intervals have been
combined, this figure reflects the entire
combined interval. (OBTAIN only)
SMFID
SMF ID of the system on which the task was
run.
C,4
C,4
C,4
TRXELAP
Total elapsed time for the TSO user session
(in seconds).
F,4
D,8
N,10
TSOUSER
TSO user ID.
C,7
C,7
C,7
VOLSER
Volume serial number for I/O waits.
C,6
C,6
C,6
WAITNAME
Full name of wait reason (see “Wait reasons
(WAITNAME)” on page 283).
C,10
C,20
C,20
WAITPCT
Percent of total elapsed time spent on this
wait reason.
F,4
D,4
N,10
WAITTIME
Total time spent on this wait reason (in
seconds).
F,4
D,4
N,10
WKLNAME
Workload name (see notes following the
table).
C,12
C,12
C,12
F,8
Notes
The allowed values for WKLNAME are as follows:
Workload names (WKLNAME)
TSO SESSION
ACCOUNT NUM
Chapter 13. Report element tables
289
PROGRAM NAME
Comparing workloads and profiles
The SAS interface allows you to extract and compare data from the Profile datastore and the EPILOG
datastore using the workload keywords described in “Workload keywords” on page 171. A variety of
information is extracted for each workload and profile, including the amount of time that the workload spent
waiting for each wait reason. (A list of wait reasons is shown in Table 14 on page 146.) One SAS record is
generated for each wait reason, even if the wait reason caused no delay. If there was more than one type
of I/O or enqueue wait reason, multiple records are generated (one for each type of I/O or enqueue wait).
Because a profile may have a number of I/O devices associated with the same wait reason, an additional,
summary record is generated for ACTIVE I/O, QUEUED I/O and RESERVE I/O. This record is a
composite of the device-specific data extracted for the wait reason. This record will not have a device
address (DEVADD) associated with it.
The COMPEXT tables
In the SAS interface, the COMPEXT command allows you to compare historical data to user-defined
profiles and export the data to SAS data sets. Data extracted and compared by the COMPEXT command
is written out to four SAS data sets that correspond to the four workload types. Table 43 shows workload
tables, the type of data they hold, and the COMPEXT command keywords that correspond to them. Each
workload table name corresponds to a SAS data set name.
Table 43. Workload tables for the COMPEXT command
Workload table
Type of data
COMPEXT keyword
PGNCOMP
Performance group degradation data
PGN, PGP, SYMBOLIC, or
SYSTEM
BTCHCOMP
Batch job degradation data
JOB, PROGRAM,
ACCOUNT, or CLASS
STCCOMP
Started task degradation data
STC, PROGRAM, or
ACCOUNT
TSOCOMP
TSO user session degradation data
TSO, PROGRAM, or
ACCOUNT
In most instances, the COMPEXT command writes to only one SAS data set. The exceptions to this occur
with the PROGRAM and ACCOUNT keywords. Since these keywords may select batch jobs, started tasks,
and TSO user sessions, they may update more than one SAS data set with a single COMPEXT
command.
Each invocation of the KEPILGPM procedure resets all 16 SAS data sets. If the COMPEXT command in
the proc does not write any data to a particular data set, that data set contains no records after the proc is
executed.
The following sections describe the EPILOG datastore and Profile datastore data that is extracted and
compared for performance groups, batch jobs, started tasks, and TSO user sessions. These tables are
similar in format to the previous tables for workload degradation data. However, they are not available for
use by the OBTAIN command.
Performance group and profile comparisons (PGNCOMP)
When you use COMPEXT to extract and compare information about performance groups, data is written
out to a SAS data set named PGNCOMP. The variables associated with this data set are described below.
290
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Table 44. Performance group comparison data
Dictionary name Description
Type
Length
DEVADD
Device address for I/O wait reasons (in cuu format) (for MVS/370)
or device number (in nnnn format) for MVS/XA and MVS/ESA. A
blank address for ACTIVE I/O, QUEUED I/O, or RESRVE I/O
signifies that the observation is a composite of all the extracted
data for the wait reason and is not device specific.
Char
4
DPRODIDX
Difference in productivity index between the workload and the
profile.
Float
4
DTRXRESP
Difference in average transaction response time between the
workload and the profile (in seconds).
Float
4
DWAITTME
Difference in average transaction time spent on this wait reason
between the workload and the profile (in seconds).
Float
4
EDATTIME
Date and time interval end for the workload.
Float
8
ENQNAME
Major enqueue names for enqueue waits (see “Enqueue names
(ENQNAME)” on page 283).
Char
8
PAVGTRX
Average transaction response time for the profile.
Float
4
PDATTIME
Date and time of COMPEXT command.
Float
8
PEDATTIM
Date and time of interval end for the profile.
Float
8
PGN
Performance group number for the workload.
Float
4
PGNTYPE
Performance group type for the workload (see notes following the
table).
Char
1
PGP
Performance group period for the workload.
Float
4
PPGN
Performance group number for the profile.
Float
4
PPGNTYPE
Performance group type for the profile (see notes following the
table).
Char
1
PPGP
Performance group period for the profile.
Float
4
PPRODIDX
Productivity index for the profile. This is a general indicator of how
severely performance was being degraded during the reporting
interval.
Float
4
Filters
For detailed information about how execution states are
categorized as productive, unproductive, and idle, and instructions
about how you can modify these definitions, see “The PRODUCTS
command” on page 114.
PRODIDX
Productivity index for the workload. See the description of
PPRODIDX above for a description of this field.
Float
4
PROFNAME
Profile name.
Char
16
PSDATTIM
Date and time of interval start for the profile.
Float
8
PSMFID
SMF ID for the profile; up to 5 IDs may be included.
Char
20
PSYMBNAM
Symbolic name defined for the profile performance group (if any).
Char
8
PWAITPCT
Percent of average transaction response time spent on this wait
reason for the profile.
Float
4
PWAITTME
Average transaction time spent on this wait reason for the profile
(in seconds).
Float
4
PWKLNAME
Workload name for the profile (see notes following the table).
Char
12
SDATTIME
Date and time of interval start for the workload.
Float
8
SMFID
SMF ID for the workload.
Char
4
Chapter 13. Report element tables
291
Table 44. Performance group comparison data (continued)
Dictionary name Description
Type
Length
SYMBNAME
Symbolic name defined for the workload performance group (if
any).
Char
8
TRXRESP
Average transaction response time for the workload (in seconds).
Float
4
VOLSER
Volume serial number for I/O wait reasons.
Char
6
WAITNAME
Full name of wait reason (see notes below).
Char
12
WAITTIME
Average transaction time spent on this wait reason for the
workload (in seconds).
Float
4
WAITPCT
Percent of average transaction response time spent on this wait
reason for the workload.
Float
4
WKLNAME
Workload name (see Notes).
Char
12
Filters
Notes
The allowed values for PPGNTYPE, WKLNAME, and PWKLNAME are as follows:
Performance group types (PGNTYPE, PPGNTYPE)
For OBTAIN, the values are :
C = Control
R = Report
For EXTRACT and COMPEXT, the values are:
CONTROL PGN
REPORT PGN
Wait reasons (WAITNAME)
ACTIVE I/O
APPCWAIT
AUX STORE
CENTSTOR
COMMON IN
CPU USED
CPU WAIT
DETECTED
DISK MOUNT
ECB WAIT
ECB+STIMER
ENQ EXCHNG
ENQUEUE
EXCHANGE
HSM BACKUP
HSM CONTRL
HSM DELETE
HSM HLIST
HSM JES CI
HSM MIGRTE
HSM RECALL
HSM RECOVR
JES CANCEL
JES DELETE
JES REQUE
JES STATUS
JES SYSOUT
LOGICAL CH
LONG WAIT
MVS LOCK
OUTLONG
PAGE IN
QUEUED I/O
REAL STORE
REQUEST
RESRVE I/O
SRM DLY (MPL)
SRM RTO
STAGING
STIMER
SWAP IN
SYSPAGE
TAPE MOUNT
TERM IN
TERM OUT
TRANSITION
UNILATERAL
WTOR WAIT
These wait reasons are described in detail in Table 14 on page 146. Note that a blank DEVADD for
ACTIVE I/O, QUEUED I/O, or RESRVE I/O signifies that the observation is a composite of all the
extracted data for the wait reason and is not device specific.
Workload names (WKLNAME, PWKLNAME)
ALL SYSTEM
PERIOD
292
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
SYMBOLIC PGN
PGN
Batch job and profile comparisons (BTCHCOMP)
When you use COMPEXT to extract and compare information about batch jobs, data is written out to a
SAS data set named BTCHCOMP. The variables associated with this data set are described in the
following table.
Table 45. Batch job comparison data
Dictionary name
Description
Type
Length
ACCTCDE
Account code of job for the workload.
Char
12
DEVADD
Device address for I/O wait reasons (in cuu format) (for MVS/370) Char
or device number (in nnnn format) for MVS/XA and MVS/ESA. A
blank address for ACTIVE I/O, QUEUED I/O, or RESRVE I/O
signifies that the observation is a composite of all the extracted
data for the wait reason and is not device specific.
4
DPRODIDX
Difference in productivity index between the workload and the
profile.
Float
4
DTRXELAP
Difference in average elapsed time between the workload and the Float
profile (in seconds).
4
DWAITTME
Difference in average elapsed time spent on this wait reason
between the workload and the profile (in seconds).
Float
4
EDATTIME
Date and time interval end for the workload.
Float
8
ENQNAME
Major enqueue names for enqueue waits (see “Enqueue names
(ENQNAME)” on page 283).
Char
8
JESCLASS
Batch job execution class for the workload.
Char
4
JESNUM
JES job number for the workload.
Char
4
JOBNAME
Batch job name for the workload.
Char
8
PACCTCDE
Account code of job for profile.
Char
12
PAVGELAP
Average total elapsed time for the profile.
Float
4
PDATTIME
Date and time of COMPEXT command.
Float
8
PEDATTIM
Date and time of interval end for the profile.
Float
8
PGMNAME
Batch job program name for the workload.
Char
8
PJESCLAS
Batch job execution class for the profile.
Char
4
PJOBNAME
Batch job name for the profile.
Char
8
PPGMNAME
Batch job program name for the profile.
Char
8
PPRODIDX
Productivity index for the profile. This is a general indicator of how Float
severely the batch workload performance was being degraded.
4
Filters
For detailed information about how execution states are
categorized as productive, unproductive, and idle, and instructions
about how you can modify these definitions, refer to “The
PRODUCTS command” on page 114.
PRODIDX
Productivity index for the workload. See the description of
PPRODIDX above for a description of this field.
Float
4
PROFNAME
Profile name.
Char
16
PSDATTIM
Date and time of interval start for the profile.
Float
8
PSMFID
SMF ID for the profile; up to 5 IDs may be included.
Char
20
Chapter 13. Report element tables
293
Table 45. Batch job comparison data (continued)
Dictionary name
Description
Type
Length
PWAITPCT
Percent of average job elapsed time spent on this wait reason for
the profile.
Float
4
PWAITTME
Average elapsed time spent on this wait reason for the profile (in
seconds).
Float
4
PWKLNAME
Workload name for the profile (see notes following the table).
Char
12
SDATTIME
Date and time of interval start for the workload.
Float
8
SMFID
SMF ID for the workload.
Char
4
TRXELAP
Total elapsed time for the workload (in seconds).
Float
4
VOLSER
Volume serial number for I/O wait reasons.
Char
6
WAITNAME
Full name of wait reason (see “Wait reasons (WAITNAME)” on
page 292).
Char
12
WAITPCT
Percent of average elapsed time spent on this wait reason for the Float
workload.
4
WAITTIME
Average elapsed time spent on this wait reason for the workload
(in seconds).
Float
4
WKLNAME
Workload name for the workload (see notes following the table).
Char
12
Filters
Notes
The allowed values for PWKLNAME and WKLNAME are as follows:
Workload names (PWKLNAME, WKLNAME)
ACCOUNT NUM
PROGRAM NAME
BATCH CLASS
BATCH JOB
Started task and profile comparisons (STCCOMP)
When you use COMPEXT to extract and compare information about started tasks, data is written out to a
SAS data set named STCCOMP. The variables associated with this data set are described below.
Table 46. Started task comparison data
Dictionary name Description
Type
Length
ACCTCDE
Account code of started task for the workload.
Char
12
DEVADD
Device address for I/O wait reasons (in cuu format) (for MVS/370) or Char
device number (in nnnn format) for MVS/XA and MVS/ESA. A blank
address for ACTIVE I/O, QUEUED I/O, or RESRVE I/O signifies that
the observation is a composite of all the extracted data for the wait
reason and is not device specific.
4
DPRODIDX
Difference in productivity index between the workload and the
profile.
Float
4
DTRXELAP
Difference in average elapsed time between the workload and the
profile (in seconds).
Float
4
DWAITTME
Difference in average elapsed time spent on this wait reason
between the workload and the profile (in seconds).
Float
4
EDATTIME
Date and time interval end for the workload.
Float
8
294
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Filters
Table 46. Started task comparison data (continued)
Dictionary name Description
Type
Length
ENQNAME
Major enqueue names for enqueue waits (see “Enqueue names
(ENQNAME)” on page 283).
Char
8
JESNUM
JES job number for the workload.
Char
4
PACCTCDE
Account code of started task for the profile.
Char
12
PAVGELAP
Average total elapsed time for the profile.
Float
4
PDATTIME
Date and time of COMPEXT command.
Float
8
PEDATTIM
Date and time of interval end for the profile.
Float
8
PGMNAME
Started task program name for the workload.
Char
8
PPGMNAME
Started task program name for the profile.
Char
8
PPRODIDX
Productivity index for the profile. This is a general indicator of how
severely the performance of the started task being degraded.
Float
4
Filters
For detailed information about how execution states are categorized
as productive, unproductive, and idle, and instructions about how
you can modify these definitions, refer to “The PRODUCTS
command” on page 114.
PRODIDX
Productivity index for the workload. See the description of
PPRODIDX above for a description of this field.
Float
4
PROFNAME
Profile name.
Char
16
PSDATTIM
Date and time of interval start for the profile.
Float
8
PSMFID
SMF ID for the profile; up to 5 IDs may be included.
Char
20
PSTCNAME
Name of the started task in the profile record.
Char
8
PWAITPCT
Percent of average elapsed time spent on this wait reason for the
profile.
Float
4
PWAITTME
Average elapsed time spent on this wait reason for the profile (in
seconds).
Float
4
PWKLNAME
Workload name for the profile (see notes following the table).
Char
12
SDATTIME
Date and time of interval start for the workload.
Float
8
SMFID
SMF ID for the workload.
Char
4
STCNAME
Started task name for the workload.
Char
8
TRXELAP
Total elapsed time for the workload (in seconds).
Float
4
VOLSER
Volume serial number for I/O wait reasons.
Char
6
WAITNAME
Full name of wait reason (see “Wait reasons (WAITNAME)” on page Char
292).
12
WAITPCT
Percent of average elapsed time spent on this wait reason for the
workload.
Float
4
WAITTIME
Average elapsed time spent on this wait reason for the workload (in
seconds).
Float
4
WKLNAME
Workload name for the workload (see notes following the table).
Char
12
Notes
The allowed values for PWKLNAME and WKLNAME are as follows:
Chapter 13. Report element tables
295
Workload names (PWKLNAME, WKLNAME)
STARTED TASK
PROGRAM NAME
TSO user session and profile comparisons (TSOCOMP)
When you use COMPEXT to extract and compare information about TSO user sessions, data is written
out to a SAS data set named TSOCOMP. The variables associated with this data set are described below.
Table 47. TSO user session comparison data
Dictionary name
Description
Type
Length
ACCTCDE
Account code of TSO session for the workload.
Char
12
DEVADD
Device address for I/O wait reasons (in cuu format) (for MVS/370)
or device number (in nnnn format) for MVS/XA and MVS/ESA. A
blank address for ACTIVE I/O, QUEUED I/O, or RESRVE I/O
signifies that the observation is a composite of all the extracted
data for the wait reason and is not device specific.
Char
4
DPRODIDX
Difference in productivity index between the workload and the
profile.
Float
4
DTRXELAP
Difference in session elapsed times between the workload and the
profile (in seconds).
Float
4
DWAITTME
Difference in average session elapsed time spent on this wait
reason between the workload and the profile (in seconds).
Float
4
EDATTIME
Date and time interval end for the workload.
Float
8
ENQNAME
Major enqueue names for enqueue waits (see “Enqueue names
(ENQNAME)” on page 283).
Char
8
JESNUM
JES job number for the workload.
Char
4
PACCTCDE
Account code of the TSO session for the profile.
Char
12
PAVGELAP
Average session elapsed time for the profile.
Float
4
PDATTIME
Date and time of COMPEXT command.
Float
8
PEDATTIM
Date and time of interval end for the profile.
Float
8
PPRODIDX
Productivity index for the profile. This is a general indicator of how
severely the performance of the started task being degraded.
Float
4
For detailed information about how execution states are categorized
as productive, unproductive, and idle, and instructions about how
you can modify these definitions, refer to “The PRODUCTS
command” on page 114.
PRODIDX
Productivity index for the workload. See the description of
PPRODIDX above for a description of this field.
Float
4
PROFNAME
Profile name.
Char
16
PSDATTIM
Date and time of interval start for the profile.
Float
8
PSMFID
SMF ID for the profile; up to 5 IDs may be included.
Char
20
PTSOUSER
TSO user ID in the profile record.
Char
7
PWAITPCT
Percent of average session elapsed time spent on this wait reason
for the profile (in seconds).
Float
4
PWAITTME
Average session elapsed time spent on this wait reason for the
profile (in seconds).
Float
4
PWKLNAME
Workload name for the profile (see notes following the table).
Char
12
SDATTIME
Date and time of interval start for the workload.
Float
8
296
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Filters
Table 47. TSO user session comparison data (continued)
Dictionary name
Description
Type
Length
SMFID
SMF ID for the workload.
Char
4
TRXELAP
Total session elapsed time for the workload (in seconds).
Float
4
TSOUSER
TSO user ID in workload.
Char
7
VOLSER
Volume serial number for I/O wait reasons.
Char
6
WAITNAME
Full name of wait reason (see “Wait reasons (WAITNAME)” on
page 292).
Char
12
WAITPCT
Percent of average session elapsed time spent on this wait reason
for the workload.
Float
4
WAITTIME
Average elapsed time spent on this wait reason for the workload (in Float
seconds).
4
WKLNAME
Workload name for the workload (see notes following the table).
12
Char
Filters
Notes
The allowed values for PWKLNAME and WKLNAME are as follows:
Workload names (PWKLNAME, WKLNAME)
TSO USER
ACCOUNT NUM
PROGRAM NAME
Chapter 13. Report element tables
297
298
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Part 3. Appendixes
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
299
300
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Appendix A. Data dictionary
This section describes the EPILOG data elements (and corresponding SAS variables), and is intended to
be used as a reference when you write programs that use data generated by the OBTAIN command (as
described in Chapter 11, “Exporting EPILOG data,” on page 193) or the SAS interface (as described in
Chapter 12, “Reporting with SAS graphics,” on page 201). The data elements are listed in alphabetical
order.
For each data element, the dictionary lists the following information:
Tables
The report tables that include this data element or SAS variable. For an
explanation of report tables, see “Report table organization” on page 233.
The report table is identical to the SAS data set that is used by the SAS
interface commands.
Value range
Range of values for the data element.
Data type
The data type and length of the data element. When this information
varies depending on usage, the field is marked INT (for the INT format on
the OBTAIN command), COL (for the COL and PC formats in the OBTAIN
command), and SAS (for the formats used by the EXTRACT and
COMPEXT commands in the SAS interface).
The following symbols are used in the Data Type columns:
Symbol
C
D
F
H
I
N
P
T
U
X
Definition
Character (all formats)
Floating point number (INT format)
Floating point number (SAS format)
Hexadecimal (in EBCDIC for COL and PC formats)
Numeric signed integer (INT format)
General numeric format. Can be integer or decimal. If
positive, will be unsigned; if negative, will be signed. (COL
and PC formats)
Packed decimal (INT format)
Time in hundredths of a second (for INT format)
Unsigned binary integer (INT format)
Hexadecimal (INT format)
290
The EPILOG display functions that generate the information in the data
element.
Commands
The EPILOG commands that can be used to generate the data element.
Note that EXTRACT and COMPEXT commands indicate the SAS
interface.
Filters
The exception threshold filters that can be used with this data element.
(Valid keywords are RIF and SIF. For further information, see “Workload
exception filters” on page 143 and “Resource exception filters” on page
155.)
Environments
The MVS environments that support the data element. Valid environments
are 370, XA, and ESA. In addition, environments that restrict or require
the 3090 processor, the logical partitioning machine, or expanded storage
are noted.
ACCTCDE
Account code for batch jobs and TSO users. The source of the data is the accounting field
on the job card or the LOGON procedure. This data element will only be valid for batch
jobs and TSO session workloads if you have chosen to collect it in your installation’s
collection filter process.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
301
For COMPEXT data, this field refers to the batch job, started task, or TSO user session in
the workload, not the profile.
Short name
None
Tables
BTCHDETL, BTCHCOMP, STCDETL, STCCOMP,
TSODETL, TSOCOMP
Value range
User-defined
Data type
C, 12
290
Batch job, started task, and TSO session degradation
displays, and corresponding profile/workload comparison
displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
AVGCTRX
Average concurrent transactions. The average number of concurrent transactions in the
performance group for the interval.
Short name
AVTX
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
AVGNSWAP
Average non-swappable users. The average number of non-swappable users who were
concurrently active during the interval.
Short name
ANSP
Tables
RDOM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
I, 10
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
AVGRESP
Average response time for a transaction (in seconds).
Short name
ARSP
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN
Filters
RIF
Environments
370, XA, ESA
AVGTRXIN
Average transactions in. The average number of concurrent transactions in storage from
the performance group.
Short name
TXIN
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
302
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
N, 11
F, 4
Performance group resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
AVGUGOUT
Average users going out. The average number of address spaces in the domain that were
marked for swap-out during the interval.
Short name
AGOT
Tables
RDOM
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
AVGUIN
Average user in. The average number of swappable users from the domain that were
in-storage during the interval.
Short name
AUIN
Tables
RDOM
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
AVGUOUT
Average users out. The average number of users in the domain that were swapped-out
during the interval.
Short name
AUOT
Tables
RDOM
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
AVGURDY
Average user ready. The average number of users in the domain that were ready to use
the CPU.
Short name
ARDY
Tables
RDOM
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
Appendix A. Data dictionary
303
AVGWKST
Average working set. The average working set size for an address space in the
performance group (in bytes).
Short name
WKST
Tables
RPGN
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
BATAVGAS
Batch average address spaces. The average number of concurrent batch address spaces
running during the interval.
Short name
BAVG
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
BATMAXAS
Batch maximum address spaces. The highest number of concurrent batch address spaces
running during the interval.
Short name
BMAX
Tables
RCPU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
BATMINAS
Batch minimum address spaces. The lowest number of concurrent batch address spaces
running during the interval.
Short name
BMIN
Tables
RCPU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
BPSSCCH1
The bypass cache percentage for the cache device, expressed in tenths of a percent. This
statistic comes from the 3880 first storage director or the 3990 subsystem.
Short name
BPC1
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
304
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
N, 11
F, 4
RCCH resource displays
DISPLAY, OBTAIN, EXTRACT
RIF
XA, ESA
BPSSCCH2
The bypass cache percentage for the cache device, expressed in tenths of a percent. This
statistic comes from the 3880 second storage director.
Short name
BPC2
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
CACHSTT1
The caching status of the 3990 or the 3880 first storage director.
Short name
CST1
Tables
RCCH
Value range
On 3880 displays: “activ»»” or “Inact»»”; on 3990 displays:
“PenAc»»,” “PenDa»»,” “TimeO»»,” “activ*»»,” “inact*»»,”
“PenAc*»»,” “PenDa*»»,” or “TimeO*»»”
Data type
C, 8
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
CACHSTT2
The caching status of the 3880 second storage director.
Short name
CST2
Tables
RCCH
Value range
“activ»»,” “inactiv»»,” “activ*»»,” or “inact*»»”
Data type
C, 8
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
CCHDASDN
The normal track transfers from cache to DASD for cache device, expressed in tenths of a
track per second. This statistic comes from the 3990 subsystem.
Short name
TCDN
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
CHNNUM
Channel number.
Short name
Tables
CHNM
RCHN
Appendix A. Data dictionary
305
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
Positive hexadecimal digit
U, 1
N, 3
F, 4
Channel resource displays
OBTAIN, EXTRACT, COMPEXT
RIF
370 only
CHNSETID
Channel set identifier. The channel set identification number for the channel.
Short name
CHNS
Tables
RCHN
Value range
Positive hexadecimal digit
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Channel resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
370 only
CHNTYPE
Channel type, including byte multiplexor, block multiplexor, ESCON channel, ESCON
converter, and ESCON director.
EPILOG field
Typ
Short name
CTYP
Tables
RCHN
Value range
Valid channel types are:
Byte
Block
ESchn
EScnv
ESdir
CHPID
306
Data type (INT)
Data type (COL, PC)
Data type (SAS)
EPILOG source
Commands
Filters
Environments
The following values indicate the type of the last period in
a COMBINE. The asterisk (*) indicates that the type has
changed during the COMBINE interval.
Byte*
Block*
ESchn*
EScnv*
ESdir*
C,6
C,6
C,8
Channel resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
Channel path ID.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
CPID
RCHN, RLCU
Two hexadecimal digits
C, 2
Channel resource and I/O queuing displays
OBTAIN, EXTRACT
RIF
XA, ESA
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
CHPINSEC
The rate at which I/O requests
this channel path.
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
for devices attached to this control unit are completed by
CINS
RLCU
Positive integers
I, 4
N, 11
F, 4
I/O queuing displays
OBTAIN, EXTRACT
RIF
370, XA, ESA (3090 only)
CIOPRSEC
Channel I/O per second. The average number of successful START and RESUME I/O
instructions issued to the channel per second.
Short name
CIOS
Tables
RCHN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Channel resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
370 only
CNTRLMOD
The controller model.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
CMOD
RCCH
3880-13, 3880-23, or 3990-3
C, 7
RCCH resource displays
DISPLAY, OBTAIN, EXTRACT
RIF
XA, ESA
CONNTIME
Connect time. The average number of milliseconds the device is connected to the channel
path for search and data transfer.
Short name
CTIM
Tables
RDAS
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
CONTIDX
Contention index. The contention index of a domain is used to select which domain will
have its target MPL adjusted when the SRM determines the system’s MPL must be
changed. The contention index indicates how busy the system looks to the domain as a
whole. Its method of calculation is determined by the installation’s IPS.
Short name
CIDX
Tables
RDOM
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Appendix A. Data dictionary
307
Data type (SAS)
290
Commands
Filters
Environments
F, 4
SRM domain resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
CPCONFIG
The dynamic I/O update status of the channel.
Short name
CPCF
Tables
RCHN
Value range
Blank, del, add, mod, add*, del*, mod*
Data type
C, 4
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA only
CPCTBUSY
Channel percent busy. The percentage of the interval that the channel was busy. The total
or complex percent utilization of the channel.
Short name
BUSY
Tables
RCHN
Value range
Positive real number between 0.0% and 100.0%
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Channel resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
CPENVIRN
The active environment for the CPU complex.
Short name
CPEN
Tables
RCHN
Value range
Basic, Logical partitioning
Data type
C, 8
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA only
CPLPSTAT
Channel path status with respect to the monitored partition.
Short name
CPST
Tables
RCHN
Value range
On, Off, Inv, Par, On*, Off*, Inv*, Par*
Data type
C, 4
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA only
CPLSHARE
Channel path logical partitioning sharable status.
Short name
CPSH
Tables
RCHN
Value range
Y, N, Y*, N*
Data type
C, 2
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA only
CPLTBUSY
The percent of the RMF interval time the channel was busy to the logical partition.
Short name
CPLU
Tables
RCHN
Value range
Blank or positive real number between 0.0% to 100.0%
308
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
I, 4
N, 11
F, 4
Channel resource displays
OBTAIN, EXTRACT
RIF
ESA only
CPMFSTAT
The Channel Path Measurement Facility status.
Short name
CPMF
Tables
RCHN
Value range
Available, unavailable, not installed, restarted
Data type
C, 12
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA only
CPPCTBSY
The percentage of the time that all channel paths for this LCU were busy at the same
time.
Short name
CBSY
Tables
RLCU
Value range
Positive real number between 0.0% and 100.0%
Data type
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
CPUFLAG
Indicates any change to the status of the CPU configuration that may have affected RCPU
values during the interval. The CPU flags are represented as one decimal value. The
decimal value is equivalent to a 32 bit unsigned binary field. Each bit represents a status
change. Bit 0 represents the most significant bit.
Bit
0-7
8
9
10
11
12
13 - 14
15
16
17
18
19
20
21
22
Description
Unused.
The number of physical processors assigned for use by
logical partitioning changed during the combined reporting
interval.
The number of logical processors assigned to this logical
partition changed during the combined reporting interval.
The wait state assist status of the logical partition changed
during the combined reporting interval.
The combined record includes interval records with
different logical partition names or numbers.
The logical partition management data indicator.
Unused.
Indicates the availability of LPAR management data during
the user-requested combined reporting interval.
The number of CPUs in the configuration changed during
the combined reporting interval.
Reserved.
The record contains queuing I/O information (XA and ESA
only).
Some combined interval records lack queuing information.
SRB service definition coefficient changed during the
combined interval.
Indicates that the system was running as logical partition.
Logical partition information is missing.
Appendix A. Data dictionary
309
23
24
25
26
27
28
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
Logical partition information is missing (combined interval
record).
CPU speed factor is missing from the record. The CPU
speed factor is used to calculate the SRB, TCB, and MVS
percent utilization figure. If it is missing, EPILOG derives a
CPU speed factor from the current processor configuration
and uses it in calculating the percent utilization figures.
The number of CPUs in the configuration changed during
the combined reporting interval.
Reserved.
An error was encountered in calculating the MVS
overhead.
Indicates dedicated processors in the logical partition.
None
RCPU
1 or 0 in the first two bits
X, 4
H, 8
F, 4
CPU resource displays
OBTAIN, EXTRACT, COMPEXT
None
Logical partitioning machine
CPUID
CPU identifier. It identifies a specific CPU in the processor complex.
Short name
None
Tables
RCPU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
CPUMODEL
The CPU model type, for example, 3081K.
Short name
CPM
Tables
RINF
Value range
Valid CPU model numbers
Data type
C, 6
290
General information resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
CPUSERL
CPU serial. The CPU serial number is a unique number assigned by the manufacturer for
this specific processor.
Short name
CPS
Tables
RINF
Value range
Positive integers. In a logical partitioning environment, the
second position from the left holds a hexadecimal digit
(0-F).
Data type
C, 6
290
General information resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
310
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Environments
370, XA, ESA
CPUTRXTM
Total CPU time for the job, task, session, or step (in CPU seconds).
Short name
None
Tables
BTCHDETL, STCDETL, TSODETL
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Batch job, started task, and TSO user session displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
CSRVTIME
Channel service time. The average time in milliseconds that the channel is active for an
I/O.
Short name
SERV
Tables
RCHN
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Channel resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
370 only
CUDEFSEC
The percent of I/O requests deferred for that control unit.
Short name
UDEF
Tables
RLCU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA (3090 only)
CUID
Identifies a control unit in the LCU.
Short name
None
Tables
RLCU
Value range
Positive integers
Data type
C, 3
290
I/O queuing resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
CUPCTBSY
The percentage of I/O requests to the LCU deferred due to a CONTROL UNIT BUSY
condition.
Short name
UBSY
Tables
RLCU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing displays
Appendix A. Data dictionary
311
Commands
Filters
Environments
OBTAIN, EXTRACT
RIF
Non-3090
CURMPL
Current multiprogramming level. The actual multiprogramming level assigned to the
domain for the interval.
Short name
CUM
Tables
RDOM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
DASDCCHN
The normal track transfers from DASD to cache for cache device, expressed in tenths of a
track per second. This statistic comes from the 3990 subsystem.
Short name
TDCN
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
DASDCCHS
The sequential track transfers from DASD to cache for cache device, expressed in tenths
of a track per second. This statistic comes from the 3990 subsystem.
Short name
TDCS
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
DASDFWST
DASD Fast Write status. (3990 only)
Short name
DFWS
Tables
RCCH
Value range
“activ»»,” “inact»»,” “PenAc»»,” “PenDa»»,” “activ*»»,”
“inact*»»,” “PenAc*»»,” “PenDa*»»,” or “n/a”
Data type
C, 8
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
DCUBSY
Average delay for an I/O request caused by the control unit busy condition (in
milliseconds).
Short name
CUBY
Tables
RDAS
312
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
Positive integers
I, 4
N, 11
F, 4
DASD resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA (3090 only)
DDEVBSY
Average delay for an I/O request caused by the device busy condition (in milliseconds).
Short name
DVBY
Tables
RDAS
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA (3090 only)
DEVADD
Device address. The channel, control unit, and unit numbers used to identify a device (in
MVS/370) or the device number (in MVS/XA or MVS/ESA).
For COMPEXT data, a blank address for ACTIVEIO, QUEUEDIO, and RESRVEIO
signifies that the observation is a composite of all the extracted data for the wait reason
and is not device specific.
Short name
DEVA
Tables
RCCH, RDAS, RPDS, RSDS, PGNDETL, PGNCOMP,
BTCHDETL, BTCHCOMP, STCDETL, STCCOMP,
TSODETL, TSOCOMP
Value range
Unsigned hexadecimal digit, 000 to FFF (in MVS/370),
0000 to FFFF (in MVS/XA and MVS/ESA).
Data type (INT)
Char, 4 (for workload tables)
Char, 3 (for resource tables)
Data type (COL, PC)
Char, 4 (for workload tables)
Char, 3 (for resource tables)
Data type (SAS)
Char, 4
290
DASD resource, I/O queuing, page data set resource,
swap data set resource, workload degradation and
profile/workload comparison displays.
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF (resources)
None (workloads)
Environments
370, XA, ESA
DEVQTIME
Device queue time. The average time an I/O spends queued for the device (in
milliseconds).
Short name
DQTM
Tables
RDAS
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
370 only
Appendix A. Data dictionary
313
DEVTYPE
The device model type, such as 3380.
Short name
DTYP
Tables
RPDS, RSDS
Value range
Valid device types
Data type
C, 4
290
Page data set resource and swap data set resource
displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA (3090 only)
DFWBYPSS
The DASD fast write bypass percentage for the cache device, expressed in tenths of a
percent. This statistic comes from the 3990.
Short name
DFWR
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N,11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
DIOPRSEC
Device I/Os per second. The average number of I/Os per second issued to the device.
Short name
DIOP
Tables
RDAS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
DISCTIME
Disconnect time (in milliseconds). For XA channels only. This represents the time the
device is busy but not transferring data. This time is composed of such items as SEEK
and SET SECTOR.
Short name
DTIM
Tables
RDAS
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
DLCUID
Logical control unit ID.
Short name
Tables
Value range
Data type
290
Commands
314
DLCU (in RDAS)
RLCU (in RLCU)
RDAS, DLCU
Positive integers
C, 3
DASD resource and I/O queuing displays
OBTAIN, EXTRACT
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Filters
Environments
RIF
XA, Logical partitioning, ESA
DOMNUM
The domain identification number.
Short name
DMN
Tables
RDOM
Value range
Positive integers
Data type (INT)
U, 1
Data type (COL, PC)
N, 3
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
DPBBUSY
Average delay time to an I/O request due to a busy ESCON director port.
EPILOG field
-Short Name
DBSY
Tables
RDAS
Value Range
Positive integers
Data Type (INT)
I,4
Data Type (COL, PC)
N,11
Data Type (SAS)
F,4
EPILOG Source
DASD resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA (3090 and ES9000 only)
DPCTBUSY
Device percent busy. The percentage of the interval time that the device was busy.
Short name
PBSY
Tables
RDAS
Value range
Positive percentage, 0.0% to 100.0%
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
DPCTBSY
The percentage of I/O requests to the LCU deferred due to a DEVICE BUSY condition.
Short name
DBSY
Tables
RLCU
Value range
Positive percentage, 0.0% to 100.0%
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
Non-3090
DPRODIDX
Delta productivity index. Difference in productivity index between the workload and the
profile. See also the data element PRODIDX below.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
0 - 100
Data type
F, 4
Appendix A. Data dictionary
315
Commands
Filters
Environments
COMPEXT
None
370, XA, ESA
DSWPPHSC
Number of swaps per second from central storage to auxiliary storage that were not
marked as logically swapped before the physical swap was performed. This field is used
differently, depending on report table. For detailed information, see the discussion of this
field in tables RSWA and RSWR in Chapter 13, “Report element tables,” on page 233.
Short name
DSWP
Tables
RSWA, RSWR
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity and swap reasons resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
See description
DSXAFLAG
Architecture flag. XA hardware behaves differently than S/370 hardware with respect to I/O
activity. There are several data elements unique to MVS/XA and MVS/ESA. This flag
indicates, for the RDAS data set, that unique XA and ESA data elements apply.
Short name
None
Tables
RCHN, RCPU, RDAS, RSWA
Value range
Valid values are:
“XA»” for XA and ESA
“370” for S/370
Data type (INT)
C, 3
Data type (COL, PC)
C, 3
Data type (SAS)
C, 3
290
Channel resource, CPU resource, DASD resource, I/O
queuing, and Swap Activity resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
DTRXELAP
Delta average elapsed time (in seconds). The difference in average elapsed times
between the workload and the profile.
Short name
None
Tables
BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive or negative real numbers
Data type (INT)
None
Data type (COL, PC)
None
Data type (SAS)
F, 4
290
Profile/workload comparison displays
Commands
EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
DTRXRESP
Delta average transaction response time (in seconds). The difference in average
transaction response time between the workload and the profile.
Short name
None
Tables
PGNCOMP
Value range
Positive or negative real numbers
Data type
F, 4
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
316
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Environments
370, XA, ESA
DWAITTME
Delta average transaction time for this wait reason (in seconds). For performance group
data, this data element contains the difference in average transaction time spent on this
wait reason between the workload and the profile. For batch job, started task, and TSO
session data, it contains the difference in average elapsed time spent on this wait reason
between the workload and the profile.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive or negative real numbers
Data type
F, 4
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
EDATE
End date. The date value for the end of the interval that covers this OBTAIN record. This
data element contains a time value. If multiple RMF-based intervals are being combined
into larger intervals, this figure reflects the end date of the larger combined interval.
Short name
None
Tables
All (except PGNCOMP, BTCHCOMP, STCCOMP, and
TSOCOMP)
Value range
Positive integers
Data type (INT)
P, 4
Data type (COL, PC)
N, 8
Data type (SAS)
None
290
All
Commands
OBTAIN
Filters
None
Environments
370, XA, ESA
EDATTIME
End date time. The date and time value for the end of the interval that covers this SAS
observation. For COMPEXT data, this field applies to the workload, not the profile. This
data element contains a standard SAS date/time value. If multiple RMF-based intervals
are being combined into larger intervals, this figure reflects the end date and time of the
larger combined interval.
Short name
None
Tables (SAS)
All
Value range
Positive integers
Data type (SAS)
F, 8
290
All
Commands
EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
ENQNAME
Enqueue name. The major name component of an enqueue request.
Short name
None
Tables
PGNDETL, PGNCOMP, BTCHDETL, BTCHCOMP,
STCDETL, STCCOMP, TSODETL, TSOCOMP
Value range
Enqueue major names are defined during product
installation. (For details, see the EPILOG for MVS
Customization Guide.) The default names are described in
Table 17 on page 150.
Data type
C, 8
290
All workload degradation and profile/workload comparison
displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Appendix A. Data dictionary
317
Environments
370, XA, ESA
ESASWPSC
Number of swaps per second that were first swapped from central storage to expanded
storage, and then migrated from expanded storage to auxiliary storage. This field is used
differently, depending on report table. For detailed information, see the discussion of this
field in tables RSWA and RSWR in Chapter 13, “Report element tables,” on page 233.
Short name
ESMG
Tables
RSWA, RSWR
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity and swap reasons resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
See description
ESAVG
Average number of expanded storage frames available during the interval.
Short name
ESAV
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA (expanded storage only)
ESDSWPSC
Number of swaps per second from central storage to expanded storage that were not
marked as logical swaps before the physical swap was performed. This field is used
differently, depending on report table. For detailed information, see the discussion of this
field in tables RSWA and RSWR in Chapter 13, “Report element tables,” on page 233.
Short name
DESW
Tables
RSWA, RSWR
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity and swap reasons resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
See description
ESFRREAL
Average number of pages per second sent to expanded storage from central storage for
paging and swapping requests.
Short name
ESRL
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA (expanded storage only)
ESINST
The total amount of expanded storage installed (in frames).
318
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
ESIN
RPAG
Positive real numbers
I, 4
N, 10
F, 4
Paging and storage resource display
OBTAIN, EXTRACT
RIF
XA, ESA (expanded storage only)
ESMAX
Maximum number of expanded
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
storage frames available during the interval.
ESMX
RPAG
Positive real numbers
I, 4
N, 10
F, 4
Paging and storage resource display
OBTAIN, EXTRACT
RIF
XA, ESA (expanded storage only)
ESMIGRAT
Expanded storage migration age; that is, the average length of time that a page remains
unreferenced in expanded storage before it is migrated to auxiliary storage.
Short name
EMIG
Tables
RSRM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
I, 10
Data type (SAS)
None
290
SRM MPL adjustment values resource display
Commands
OBTAIN
Filters
None
Environments
370, XA, ESA (3090 only)
ESMIN
Minimum number of expanded
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
ESONLINE
The amount of expanded storage online during the interval (in frames).
Short name
ESON
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
storage frames available during the interval.
ESMN
RPAG
Positive real numbers
I, 4
N, 10
F, 4
Paging and storage resource display
OBTAIN, EXTRACT
RIF
XA, ESA (expanded storage only)
Appendix A. Data dictionary
319
Environments
XA, ESA (expanded storage only)
ESTOAUX
Average number of pages per second sent to auxiliary storage from expanded storage
during the interval.
Short name
ESAX
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA (expanded storage only)
ESTSWPSC
Number of swaps per second from central storage to expanded storage that were marked
as logical swaps before the physical swap was performed. This field is used differently,
depending on report table. For detailed information, see the discussion of this field in
tables RSWA and RSWR in Chapter 13, “Report element tables,” on page 233.
Short name
TESW
Tables
RSWA, RSWR
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity and swap reasons resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
See description
ETIME
End time. The time value for the end of the interval that covers this OBTAIN record. This
data element contains a time value. If multiple RMF-based intervals are being combined
into larger intervals, this figure reflects the end time of the larger combined interval.
Short name
None
Tables
All (except PGNCOMP, BTCHCOMP, STCCOMP, and
TSOCOMP)
Value range
Positive integers
Data type (INT)
T, 4
Data type (COL, PC)
C, 8
290
All
Commands
OBTAIN
Filters
None
Environments
370, XA, ESA
EXTDTIME
Extract date time. The date and time the EXTRACT command which produced this
observation was executed. This field is common to all observations in a data set which
were produced by a particular EXTRACT command and to the title observation in the title
data set which defines the parameters of the extract. This data element contains a
standard SAS date/time value.
Short name
None
Tables
All SAS data sets
Value range
Positive integers
Data type (SAS)
F, 8
290
All
Commands
EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
320
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
FIXED
Average fixed frame counts (in 4K frames).
Short name
None
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
HIGHSTEP
Highest step number of the job or started task.
Short name
None
Tables
BTCHDETL, STCDETL
Value range
Positive integers
Data type (INT)
U, 2
Data type (COL, PC)
N, 5
Data type (SAS)
F, 4
290
Batch job and started task degradation displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
ICS
The current ICS suffix.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
INFFLAG
None
RINF
Valid ICS suffixes
C, 2
General information resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
Bits 8-15 indicate changes made during a combined interval (not used unless the interval
is combined.) The flags are represented as one decimal value. The decimal value is
equivalent to a 32 bit unsigned binary field. Each bit represents a flag. Bit 0 represents the
most significant bit.
Bit
0-7
8
9
10
11
12
13
14 - 15
16
17
18
19
20
21
22 - 23
24
25
Description
Unused.
The partition number changed.
The partition name changed.
The number of physical processors changed.
The relative share changed.
The global time slice changed.
The CPU capping changed.
Unused.
VM is running in native mode.
Logical partition.
Dedicated processors.
Wait completion = yes.
CPU capping is in effect for the partition in which EPILOG
is running.
The logical partitioning management data is present.
Unused.
MVS level.
IPS.
Appendix A. Data dictionary
321
26
27
28
29
30
31
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
ICS.
OPT.
CPU model.
Serial.
Mode.
RMF version.
None
RINF
1 or 0 in each of the bits
I, 4
C, 10
F, 4
General information resource displays
OBTAIN, EXTRACT, COMPEXT
None
370, XA, ESA
INHBCCH1
The inhibit cache percentage for the cache device, expressed in tenths of a percent. This
statistic comes from the 3880 first storage director or the 3990 subsystem.
Short name
IHC1
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
INHBCCH2
The inhibit cache percentage for the cache device, expressed in tenths of a percent. This
statistic comes from the 3880 second storage director.
Short name
IHC2
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
INTVLSRV
Average number of SRM service units used by all members of the domain during the
interval.
Short name
SRVU
Tables
RDOM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
IODEFSEC
Total percent of I/O requests on the LCU that were deferred.
Short name
IDFS
322
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
RLCU
Positive real numbers
I, 4
N, 11
F, 4
I/O queuing resource display
OBTAIN, EXTRACT
RIF
XA, ESA (non-3090 only)
IOINTRPT
Average number of I/O interrupts per second handled by the processor during the interval.
Short name
IONT
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
IOPERSEC
Average number of I/O requests for the data set per second.
Short name
IOSC
Tables
RPDS, RSDS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Page data set and swap data set displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
IOPIOQLN
The average number of I/O requests queued for this I/O processor.
Short name
IOQL
Tables
RLCU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA (3090 only)
IOPIOSEC
The rate per second at which I/O requests are placed on the IOP initiation queue.
Short name
IOPS
Tables
RLCU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA (3090 only)
Appendix A. Data dictionary
323
IOPROCID
I/O processor ID.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
None
RLCU
Positive real numbers
C, 2
I/O queuing displays
OBTAIN, EXTRACT
None
XA, ESA (3090 only)
IOPTLDEF
The rate at which the IOP placed delayed I/O requests on the control unit header queue
during the interval.
Short name
IDEF
Tables
RLCU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA (3090 only)
IOSQTIME
The average time (in milliseconds) that an I/O spent on the IOS queue for the device.
Short name
QTIM
Tables
RDAS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
IPS
The current IPS suffix.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
ITRGMPL
324
None
RINF
Valid IPS suffixes
C, 2
General information resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
Average In-Target MPL over the interval, as calculated by SRM.
Short name
ITDM
Tables
RDOM
Value range
Positive Integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA/SP 4.2 and above
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
JESCLASS
Batch job execution class. For COMPEXT data, this field refers to the batch job in the
workload, not the profile.
Short name
None
Tables
BTCHDETL, BTCHCOMP
Value range
Valid execution classes
Data type
C, 3
290
Batch job and profile/workload comparison displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
JESNUM
JES job number for the batch job, started task, or TSO user session. For COMPEXT data,
this field refers to the batch job, started task, or TSO session in the workload, not the
profile.
Short name
None
Tables
BTCHDETL, BTCHCOMP, STCDETL, STCCOMP,
TSODETL, TSOCOMP
Value range
Positive integers
Data type (INT)
C, 5
Data type (COL, PC)
C, 5
Data type (SAS)
C, 4
290
Batch job, started task, and TSO session displays, and all
corresponding profile/workload comparison displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
JOBNAME
Batch job name. For COMPEXT data, this field refers to the batch job in the workload, not
the profile.
Short name
None
Tables
BTCHDETL, BTCHCOMP
Value range
User-defined
Data type
C, 8
290
Batch job and profile/workload comparison displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
LCUCPST
Indicates the status of the channel path during the reporting interval. The status flags are
represented as one decimal value. The decimal value is equivalent to a 32-bit unsigned
binary field. Each bit represents a status flag. Bit 0 represents the most significant bit.
0
= Channel path installed
1
= Channel path online
2
= Channel path varied
3
= Channel path offline to all devices
4
= Channel path to all devices altered
5
= Path data invalid
Short name
None
Tables
RLCU
Value range
Described above
Data type (INT)
X, 1
Data type (COL, PC)
H, 2
Data type (SAS)
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
XA, ESA
Appendix A. Data dictionary
325
LCUERROR
If this data element contains a value other than hexadecimal zero (x’0’), this indicates that
the collector encountered an error while collecting this data, and that the data in this
observation may not be reliable.
Short name
None
Tables
RLCU
Value range
Hexadecimal numbers
Data type (INT)
X, 1
Data type (COL, PC)
H, 2
Data type (SAS)
C, 1
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
XA, ESA
LCUIOQLN
The average number of I/O requests queued for this LCU.
Short name
LOQL
Tables
RLCU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
LCUIOSEC
The rate per second at which I/O requests were successfully selected for initiation by the
associated LCU.
Short name
LIOS
Tables
RLCU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
Non-3090
LCUTYPE
LCU record subtype.
Short name
Tables
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
LSWPRSEC
326
LTYP
RLCU
U, 1
N, 3
C, 1
I/O queuing displays
OBTAIN, EXTRACT
None
370, XA, ESA
Logical swap rate (in pages per second). This rate consists of the total number of address
spaces that were marked as logically swapped during the interval, divided by the number
of seconds in the interval. (Some of these address spaces may have been subsequently
transitioned to auxiliary or expanded storage during the interval.) This field is used
differently, depending on report table. For detailed information, see the discussion of this
field in tables RSWA and RSWR in Chapter 13, “Report element tables,” on page 233.
Short name
LSW
Tables
RSWA, RSWR
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
Positive integers
I, 4
N, 11
F, 4
Swap activity and swap reasons resource displays
OBTAIN, EXTRACT
RIF
See description
LTPPRSEC
Percent of logical swaps that became physical swaps.
Short name
LSPS
Tables
RSWA
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity and swap reasons resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA (Non-3090)
MACHFLAG
Machine flag. 3090 processors differ from other processors with respect to I/O activity and
expanded storage. Because of this, there are several data elements unique to 3090
processors. This flag indicates that unique 3090 data elements apply.
Short name
None
Tables
RDAS, RLCU, RPAG, RSRM, RSWA, RSWR
Value range
Valid values are:
“309X” for 3090 records or SAS observations
“»»»»” for non-3090 records or SAS observations
Data type
290
Commands
Filters
Environments
For the RPAG table, the valid values are:
“EXT” for expanded storage systems
“»»»” for non-expanded storage systems
C, 4
DASD resource, I/O queuing, Paging and Swapping
resource, SRM resource, Swapping Activity resource, and
Swap Reason resource displays
OBTAIN, EXTRACT, COMPEXT
None
370, XA, ESA (3090 only)
MAXMPL
Maximum MPL specified for the domain in the IPS.
Short name
MXM
Tables
RDOM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
MINMPL
Minimum MPL specified for the domain in the IPS.
Short name
MNM
Tables
RDOM
Value range
Positive integers
Data type (INT)
I, 4
Appendix A. Data dictionary
327
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
MODE
N, 10
F, 4
SRM domain resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
Processor mode during the time period.
Short name
None
Tables
RINF
Value range
NATIVE
VM
VM/PMA
PARTITIONED
Data type
290
Commands
Filters
Environments
= Native MVS
= MVS under VM
= MVS under VM/PMA
= MVS in logical
partitioning mode
C, 20
General information resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
NONFIXED
Average non-fixed frame counts (in 4K frames).
Short name
None
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
NUMICMB
Number of EDS records extracted for the record (or SAS observation). This value is of
interest in situations where data has been combined in some fashion.
Short name
None
Tables
All
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
All
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
NUMPHYS
The number of physical processors available to the logically partitioned processor
complex.
Short name
NPRO
Tables
RINF
Value range
Positive real numbers
Data type
C, 1
290
General information resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
Logical partitioning machine
328
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
NUMTRX
Number of transactions in the record.
Short name
None
Tables
PGNDETL
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
OPENDSN
Average number of DCBs and ACBs concurrently open on a volume. On MVS/ESA
systems, this field contains the average number of allocations against a device.
Short name
ODSN
Tables
RDAS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
OTRGMPL
Average Out Target MPL over the interval, as calculated by SRM.
Short name
OTDM
Tables
RDOM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
ESA SP 4.2 and above
OSLEVEL
The level of MVS/ESA under which you are running.
Short name
OSLV
Tables
RCPU, RCHN, RDAS, RDOM, RLCU
Value range
1 to 4. A value of 1 indicates that you are running under
ESA 3.1.0; a value of 2 indicates that you are running
under ESA 3.1.3; a value of 3 indicates that you are
running under SP 4.1.0; a value of 4 indicates that you are
running under SP 4.2.0.
Data type (INT)
U, 1
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Channel resource, CPU resource, DASD resource, SRM
domain resource, and I/O queuing displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
MVS/ESA
OPT
The current OPT suffix.
Short name
Tables
Value range
None
RINF
Valid OPT suffixes
Appendix A. Data dictionary
329
Data type
290
Commands
Filters
Environments
C, 2
General information resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
PACCTCDE
Account code for the profile. (For details, see ACCTCDE.)
Short name
None
Tables
BTCHCOMP, STCCOMP, TSOCOMP
Value range
User-defined
Data type
C, 12
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PAGEIN
Average page-ins per second.
Short name
None
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
PAGEOUT
Average page-outs per second.
Short name
None
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
PAGES
Average number of page operations per second (page-ins plus page-outs).
Short name
PGS
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
290
Paging and storage resource display
Commands
OBTAIN
Filters
None
Environments
370, XA, ESA
PAGPERIO
Average number of pages per I/O request.
Short name
PGIO
Tables
RPDS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
330
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
290
Commands
Filters
Environments
Page data set and swap data set displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
PARTNAME
Partition name. The name assigned to the partition in which EPILOG is running.
Short name
PTN
Tables
RINF
Value range
Positive real numbers
Data type
C, 8
290
General information resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
Logical partitioning machine
PARTNUM
The partition number in which EPILOG is running.
Short name
PNUM
Tables
RINF
Value range
Positive integers
Data type (INT)
U, 1
Data type (COL, PC)
N, 3
Data type (SAS)
F, 4
290
General information resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
Logical partitioning machine
PARTWGT
Processor weight. Relative importance of the logical partition with respect to the other
partitions that share the same CPU.
Short name
PWT
Tables
RINF
Value range
0 - 1000. A value of MIX indicates a logical partition with
both shared and dedicated processors.
Data type
C, 3
290
General information resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
Logical partitioning machine
PATHQTME
The average time that an I/O spent queued on the path (in milliseconds).
Short name
PQTM
Tables
RDAS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
370 only
PAVGELAP
The average elapsed time for the profile (in seconds).
Short name
None
Tables
BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
C, 10
Data type (SAS)
F, 4
Appendix A. Data dictionary
331
290
Commands
Filters
Environments
Profile/workload comparison displays
COMPEXT
RIF, SIF
370, XA, ESA
PAVGTRX
The average transaction response time for the profile (in seconds).
Short name
None
Tables
PGNCOMP
Value range
Positive real numbers
Data type
F, 4
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PCTCMPLX
Processor complex CPU utilization. Percentage of total CPU cycles used by all the
partitions in the processor complex.
Short name
PCTX
Tables
RINF
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
General information resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
Logical partitioning machine
PCTCPMVS
Percent CPU used by MVS for all processors during the interval. If RMF-based intervals
were combined into larger intervals, this figure represents the percent CPU used over the
larger combined interval. In processors that support logical partitioning, this value pertains
to the logical CPU (that is, the partition under which EPILOG is running).
Short name
PMVS
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PCTCPSRB
Percent CPU used by SRBs for all processors during the interval. If RMF-based intervals
were combined into larger intervals, this figure represents the percent CPU used over the
larger combined interval. In processors that support logical partitioning, this value pertains
to the logical CPU (that is, the partition under which EPILOG is running).
Short name
PSRB
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
332
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
PCTCPTCB
Percent CPU used by TCBs for all processors during the interval. If RMF-based intervals
were combined into larger intervals, this figure represents the percent CPU used over the
larger combined interval. In processors that support logical partitioning, this value pertains
to the logical CPU (that is, the partition under which EPILOG is running).
Short name
PTCB
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PCTCPU
Average percent of total CPU capacity used over the interval. If RMF-based intervals were
combined into larger intervals, this figure represents the percent CPU used over the larger
combined interval.
Short name
PCPU
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PCTFULL
Average percent of slots used on the data set.
Short name
PFUL
Tables
RPDS, RSDS
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Page data set and swap data set displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PCTINUSE
Percent of the time that the data set was considered busy by ASM.
Short name
PUSE
Tables
RPDS, RSDS
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Page data set and swap data set displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PCTPARTA
Average logical processor utilization. Average percent of total CPU cycles assigned to this
partition that this partition used during the time period.
Short name
PRTA
Tables
RCPU
Value range
Positive real numbers
Appendix A. Data dictionary
333
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
I, 4
N, 11
F, 4
CPU resource displays
OBTAIN, EXTRACT, COMPEXT
RIF
Logical partitioning machine
PCTPRMCP
The logical partitioning management percentage for this partition.
Short name
PRMC
Tables
RCPU
Value range
1 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
Partitioned, RMF 4.2.1 and above
PCTMGPLX
The logical partitioning management percentage for the complex.
Short name
PCMX
Tables
RCPU
Value range
1 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
General information resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
Partitioned, RMF 4.2.1 and above
PCTPRTCP
Physical CPU utilization. Average percent of complex-wide CPU cycles that this partition
used during the time period.
Short name
PRTC
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
Logical partitioning machine
PCTRES
Percent of the time that at least one transaction was in storage.
Short name
PRES
Tables
RPGN
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
334
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
PCTRSVDL
Percent of the time during the interval that a shared device was reserved by another
processor or a control unit was busy with another path.
Short name
PRSV
Tables
RDAS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PCTTPI
Percent of the total I/O interrupts handled with the TPI instruction during the interval.
Short name
PTPI
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
PDATTIME
The date and time of the COMPEXT command.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive real numbers
Data type
F, 8
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PDSFLAG
If the page data set is found to be unusable, bit 0 of this field is set to 1.
Short name
None
Tables
RPDS
Value range
0 or 1 in bit 0
Data type (INT)
X, 4
Data type (COL, PC)
H, 8
Data type (SAS)
F, 4
290
Page data set resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PEDATTIM
The date and time of the interval end for the profile.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive real numbers
Data type
F, 8
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PENDTIME
The average time (in milliseconds) that an I/O spent in the pending condition.
Appendix A. Data dictionary
335
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
PTIM
RDAS
Positive real numbers
I, 4
N, 10
F, 4
DASD resource displays
OBTAIN, EXTRACT
RIF
XA, ESA
PERF
Performance group number. See also the data element PGN.
Short name
None
Tables
RPGN
Value range
Positive integers
Data type (INT)
I, 2
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN
Filters
RIF (for resources)
None (for workloads)
Environments
370, XA, ESA
PERIOD
Performance group period. See also the data element PGP.
Short name
PRD
Tables
RPGN
Value range
Positive numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group resource panels
Commands
OBTAIN
Filters
RIF (for resources)
None (for workloads)
Environments
370, XA, ESA
PGMNAME
Batch job or started task program name. For COMPEXT data, this field refers to the batch
job or started task in the workload, not the profile.
Short name
None
Tables
BTCHDETL, BTCHCOMP, STCDETL, STCCOMP
Value range
User-defined
Data type
C, 8
290
Batch job and started task degradation and
profile/workload comparison displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
PGN
Performance group number. For COMPEXT data, this field refers to the performance
group in the workload, not the profile.
Short name
None
Tables
PGNDETL, PGNCOMP, RPGN
Value range
Positive integers
Data type (INT)
I, 2
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
336
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
290
Commands
Filters
Environments
Performance group and profile/workload comparison
displays
EXTRACT, COMPEXT
None
370, XA, ESA
PGNFLAG
Indicates whether the CPU speed was recorded for the interval (0 = recorded, 64 = not
recorded). Since OBTAIN values are formatted as hexadecimal, the “not recorded”
indicator will appear as 40, the hexadecimal equivalent of 64.
Short name
None
Tables
RPGN
Value range
Bit string
Data type (INT)
X, 4
Data type (COL, PC)
H, 8
Data type (SAS)
F, 4
290
Performance group resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
PGNTYPE
Performance group type. For COMPEXT data, this field refers to the performance group in
the workload, not the profile.
Short name
None
Tables
PGNDETL, PGNCOMP
Value range
For OBTAIN, the values are:
C = Control
R = Report
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
For EXTRACT and COMPEXT, the values are:
CONTROL PGN
REPORT PGN
C, 1
C, 1
C, 11
Performance group and profile/workload comparison
displays
OBTAIN, EXTRACT, COMPEXT
None
370, XA, ESA
PGP
Performance group period. For profile/workload comparisons, this refers to the
performance period of the workload, not the profile.
Short name
None
Tables
PGNDETL, PGNCOMP, RPGN
Value range
Positive numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group degradation panels and
profile/workload comparison displays
Commands
EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
PGPRSWPI
Average number of pages per address space swap-in.
Short name
PPSI
Tables
RSWA
Value range
Positive real numbers
Appendix A. Data dictionary
337
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
I, 4
N, 11
F, 4
Swap activity resource display
OBTAIN, EXTRACT
RIF
370, XA, ESA (non-expanded storage only)
PGPRSWPO
Average number of pages per address space swap-out.
Short name
PPSO
Tables
RSWA
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA (non-expanded storage only)
PIOPRSEC
Average EXCPs per second for the performance group (as counted by SMF).
Short name
PIOS
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
PJESCLAS
Batch job execution class for the profile.
Short name
None
Tables
BTCHCOMP
Value range
Valid execution classes
Data type
C, 4
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PJOBNAME
Batch job name for the profile.
Short name
None
Tables
BTCHCOMP
Value range
User-defined
Data type
C, 8
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PPGMNAME
Batch job or started task program name for the profile.
Short name
None
Tables
BTCHCOMP, STCCOMP
Value range
User-defined
Data type
C, 8
290
Profile/workload comparison displays
338
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Commands
Filters
Environments
COMPEXT
None
370, XA, ESA
PPGN
Performance group number for the profile.
Short name
None
Tables
PGNCOMP
Value range
Positive real numbers
Data type
F, 4
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PPGNTYPE
Performance group type for the profile.
Short name
None
Tables
PGNCOMP
Value range
CONTROL PGN or REPORT PGN
Data type
C, 11
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PPGP
Performance group period for the profile.
Short name
None
Tables
PGNCOMP
Value range
Positive real numbers
Data type
F, 4
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PPRODIDX
Productivity index of the profile. For an explanation of the productivity index, see the
description of the PRODIDX data element below.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, and TSOCOMP
Value range
1 - 100
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
All profile/workload comparisons
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PROCNAME
Batch job or started task procedure name.
Short name
None
Tables
BTCHDETL, STCDETL
Value range
User-defined
Data type
C, 8
290
Batch job and started task degradation displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
PRODIDX
Productivity index. This is a general indicator of how severely the workload is being
degraded. It is derived by dividing the amount of time spent doing productive work (such
Appendix A. Data dictionary
339
as active I/O or using CPU) by the total amount of time spent in productive and
unproductive activities. Idle time (such as STIMER waits) is disregarded in the calculation,
since it is not relevant to performance degradation.
For more detailed information about how execution states are categorized as productive,
unproductive, and idle, and instructions about how you can modify these definitions, see
“The PRODUCTS command” on page 114.
For COMPEXT data, this data element contains the productivity index of the workload, not
the profile.
Short name
None
Tables
PGNDETL, BTCHDETL, STCDETL, TSODETL,
PGNCOMP, BTCHCOMP, STCCOMP, and TSOCOMP
Value range
1 - 100
Data type
F, 4
290
Workload degradation displays and profile/workload
comparisons
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
PROFNAME
Profile name.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
None
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
User-supplied
C, 16
Profile/workload comparison displays
COMPEXT
None
370, XA, ESA
PSDATTIM
Date and time of interval start for the profile.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive real numbers
Data type
F, 8
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PSMFID
SMF ID for the profile. Up to five IDs may be included.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Valid SMD IDs
Data type
C, 20
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PSTCNAME
Started task name for the profile.
Short name
None
Tables
STCCOMP
Value range
User-defined
Data type
C, 8
290
Started task profile/workload comparison displays
Commands
COMPEXT
Filters
None
340
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Environments
370, XA, ESA
PSWPRSEC
Physical swap rate (in pages per second). This field is used differently, depending on
report table. For detailed information, see the discussion of this field in tables RSWA and
RSWR in Chapter 13, “Report element tables,” on page 233.
Short name
PSW
Tables
RSWA, RSWR
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity and swap reasons resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
See description
PSYMBNAM
Symbolic name of the performance group, if one was defined during installation, for the
profile.
Short name
None
Tables
PGNCOMP
Value range
8-character descriptive name
Data type
C, 8
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PTSOUSER
TSO user ID in the profile record.
Short name
None
Tables
TSOCOMP
Value range
User-defined
Data type
C, 7
290
TSO session profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PWAITPCT
For performance group data, this data element contains the percent of average transaction
response time spent on this wait reason for the profile. For batch jobs, started tasks, and
TSO sessions, it contains the percent of average elapsed time spent on this wait reason
for the profile.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive real numbers
Data type
F, 4
290
Profile/workload comparison displays
Commands
COMPEXT
Filters
None
Environments
370, XA, ESA
PWAITTME
For performance groups, this data element contains the average transaction response time
spent on this wait reason for the profile (in seconds). For batch jobs, started tasks, and
TSO sessions, it contains the average elapsed time spent on this wait reason for the
profile (in seconds).
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive real numbers
Data type
F, 4
Appendix A. Data dictionary
341
290
Commands
Filters
Environments
PWKLNAME
Profile/workload comparison displays
COMPEXT
None
370, XA, ESA
Workload name for the profile.
Short name
None
Tables
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
For performance groups, valid workload names are:
ALL SYSTEM
PERIOD
SYMBOLIC PGN
PGN
Data type
290
Commands
Filters
Environments
For batch jobs, started tasks, and TSO sessions, valid
workload names are:
ACCOUNT NUM
PROGRAM NAME
BATCH CLASS
BATCH JOB/STARTED TASK/TSO USER ID
Char, 12
Profile/workload comparison displays
COMPEXT
RIF, SIF
370, XA, ESA
RDHITCFW
The total cache fast write read
a percent. This statistic comes
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
RDHITNM1
The normal read hit percentage for the cache device, expressed in tenths of a percent.
This statistic comes from the 3880 first storage director or the 3990 subsystem.
Short name
RHN1
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
RDHITNM2
The normal read hit percentage for the cache device, expressed in tenths of a percent.
This statistic comes from the 3880 second storage director.
Short name
RHN2
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
342
hit percentage for the cache device, expressed in tenths of
from the 3990 subsystem.
RHCF
RCCH
Positive Integer
I, 4
N, 11
F, 4
RCCH resource displays
OBTAIN, EXTRACT
RIF
XA, ESA
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
N, 11
F, 4
RCCH resource displays
OBTAIN, EXTRACT
RIF
XA, ESA
RDHITSQ1
The sequential read hit percentage for the cache device, expressed in tenths of a percent.
This statistic comes from the 3880 first storage director or the 3990 subsystem.
Short name
RHS1
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
RDHITSQ2
The sequential read hit percentage for the cache device, expressed in tenths of a percent.
This statistic comes from the 3880 second storage director.
Short name
RHS2
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
RDHITTOT
The total read hit percentage for the cache device, expressed in tenths of a percent.
Short name
RHTT
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
RDRTIME
Time (in seconds) that the job waited on the input queue either for the job to start, or, if
the EXTRACT was done at the step level, for the step to start.
Short name
None
Tables
BTCHDETL
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Batch job displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
Appendix A. Data dictionary
343
RDWRTRT
The percentage of total read and write operations that involve read processing.
Short name
RDWR
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
READYIN
Average number of ready users that were observed as “swapped-in” during the interval. If
RMF-based intervals were combined into larger intervals, this figure reflects the larger
combined interval.
Short name
RIN
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
READYOUT
Average number of ready users that were observed as “swapped-out” during the interval.
If RMF-based intervals were combined into larger intervals, this figure reflects the larger
combined interval.
Short name
ROUT
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
REAL
Average real frame counts (in 4K frames).
Short name
None
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
REPTCNTL
Performance group type flag.
Short name
Tables
Value range
Data type (INT)
344
None
RPGN
R = report performance group
C = control performance group
C, 1
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
C, 1
C, 1
Performance group resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
RESP
Average response time for a transaction (in seconds).
Short name
None
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
EXTRACT
Filters
RIF
Environments
370, XA, ESA
RMF
RMF version number.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
None
RINF
Valid RMF version numbers
C, 20
General information resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
SDATE
Start date. The date value for the start of the interval that covers this OBTAIN record. This
data element contains a time value. If multiple RMF-based intervals are being combined
into larger intervals, this figure reflects the start date of the larger combined interval.
Short name
None
Tables
All (except PGNCOMP, BTCHCOMP, STCCOMP, and
TSOCOMP)
Value range
Positive integers
Data type (INT)
P, 4
Data type (COL, PC)
N, 8
290
All
Commands
OBTAIN
Filters
None
Environments
370, XA, ESA
SDATTIME
Date and time of interval start (for SAS data only). For COMPEXT data, this field applies
to the workload, not the profile. If multiple RMF-based intervals are being combined into
larger intervals, this figure reflects the start date and time of the larger combined interval.
This data element contains a standard SAS date/time value.
Short name
None
Tables
All SAS data sets
Value range
Positive integers
Data type (SAS)
F, 8
290
All
Commands
EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
SDSFLAG
Indicates the swap data set condition. If the X’80’ bit is on, ASM has found the swap data
set to be unusable.
Appendix A. Data dictionary
345
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
None
RSDS
bit indicator
X, 4
H, 8
F, 4
Swap data set resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
SMFID
The SMF ID of the system on which the data was collected. For COMPEXT data, this field
applies to the workload, not the profile.
Short name
None
Tables
All
Value range
Valid SMF IDs
Data type
C, 4
290
All
Filters
None
Environments
370, XA, ESA
SPCTYPE
Page space type.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
STYP
RPDS
Valid space types are: PLPA, COMMON, DUPLEX,
LOCAL, or SWAP
C, 6
Page data set resource display
OBTAIN, EXTRACT
RIF
370, XA, ESA
SRASMQ
Average ASM queue length, as calculated by SRM.
Short name
ASMQ
Tables
RSRM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
I, 10
Data type (SAS)
F, 4
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SRBPCT
Percent of CPU used by SRBs in the performance group.
Short name
SRBP
Tables
RPGN
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SRCPU
Average percent CPU utilization, as calculated by SRM.
Short name
ACPU
346
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
RSRM
Positive integers
I, 4
I, 10
F, 4
SRM MPL adjustment values resource display
OBTAIN, EXTRACT
RIF
370, XA, ESA
SRDMDPAG
Demand paging rate (non-swap, non-VIO page-ins plus reclaims) (in pages per second).
Short name
DPGR
Tables
RSRM
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SRFIXB16
Percent of the first 16M of real storage that was fixed (that is, non-pageable).
Short name
FXB
Tables
RSRM
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SRFIXTOT
Percent of total real storage that was fixed (that is, non-pageable).
Short name
FXT
Tables
RSRM
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
SRMVALS
SRM MPL adjustment values and observed values.
Short name
SRVS
Tables
RSRM
Value range
The five allowed values are: HIGH THRESHOLD, LOW
THRESHOLD, OBSERVED MAX, OBSERVED MIN, and
OBSERVED AVG
Data type (INT)
C, 14
Data type (COL, PC)
C, 14
Data type (SAS)
C, 16
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Appendix A. Data dictionary
347
Environments
370, XA, ESA
SRPAGDLY
Page delay time (excluding swap paging) in milliseconds, as calculated by SRM.
Short name
PGDL
Tables
RSRM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SRPAGFLT
Page fault rate (non-swap, non-VIO page-ins plus page-outs) (in pages per second).
Short name
PFLT
Tables
RSRM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SRUIC
Average unreferenced interval count, as calculated by SRM.
Short name
AUIC
Tables
RSRM
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
SRM MPL adjustment values resource display
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SRVTIME
The average time that an I/O spent transferring data to the device (in milliseconds).
Short name
SVTM
Tables
RDAS
Value range
Positive real integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
RIF
Environments
370 only
STCAVGAS
The average number of started tasks running concurrently during the interval. If multiple
RMF-based intervals were combined into larger intervals, this figure reflects the number of
started tasks running during the larger combined interval.
Short name
SAVG
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
348
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Data type (SAS)
290
Commands
Filters
Environments
F, 4
CPU resource displays
OBTAIN, EXTRACT
RIF
370, XA, ESA
STCMAXAS
The maximum number of started tasks running concurrently during the interval. If multiple
RMF-based intervals were combined into larger intervals, this figure reflects the number of
started tasks running during the larger combined interval.
Short name
SMAX
Tables
RCPU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
STCMINAS
The minimum number of started tasks running concurrently during the interval. If multiple
RMF-based intervals were combined into larger intervals, this figure reflects the number of
started tasks running during the larger combined interval.
Short name
SMIN
Tables
RCPU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
STCNAME
The name of the started task. For COMPEXT data, this refers to the started task name in
the workload, not the profile.
Short name
None
Tables
STCDETL, STCCOMP
Value range
Positive integers
Data type
C, 8
290
Started task degradation and profile/workload comparison
displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
STEPNAME
Batch job or started task step name.
Short name
None
Tables
BTCHDETL, STCDETL
Value range
User-defined
Data type
C, 8
290
Batch job and started task degradation displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
STGDIR1
The first storage director. (3880 only)
Short name
STD1
Appendix A. Data dictionary
349
Tables
Value range
Data type
Commands
Filters
Environments
RCCH
Hexadecimal digit, 00 to FF
C, 2
DISPLAY, OBTAIN, EXTRACT
RIF
XA, ESA
STGDIR2
The second storage director. (3880 only)
Short name
STD2
Tables
RCCH resource displays
Value range
Hexadecimal digit, 00 to FF
Data type
C, 2
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
STIME
Start time. The time value for the start of the interval that covers this OBTAIN record. This
data element contains a time value. If multiple RMF-based intervals are being combined
into larger intervals, this figure reflects the start time of the larger combined interval.
Short name
None
Tables
All (except PGNCOMP, BTCHCOMP, STCCOMP, and
TSOCOMP)
Value range
Positive integers
Data type (INT)
T, 4
Data type (COL, PC)
C, 8
290
All
Commands
OBTAIN
Filters
None
Environments
370, XA, ESA
STORNAME
Name of storage area that is mapped.
Short name
None
Tables
RPAG
Value range
One of the following areas of storage:
SQA
CSA
LPA
LSQA
PVT
FREE
NUC
TOTAL
EPVT (MVS/XA and MVS/ESA only)
ECSA (MVS/XA and MVS/ESA only)
ELPA (MVS/XA and MVS/ESA only)
ESQA (MVS/XA and MVS/ESA only)
EXP. STOR. (MVS/XA and MVS/ESA only)
Data type (INT)
C, 12
Data type (COL, PC)
C, 12
Data type (SAS)
C, 6
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
SUBSYSID
The cache device subsystem id. (3990 only)
Short name
SSYS
Tables
RCCH
350
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Value range
Data type
290
Commands
Filters
Environments
Hexadecimal digit, 0000 to FFFF
C, 4
RCCH resource display
DISPLAY, OBTAIN, EXTRACT
RIF
XA, ESA
SUPERSEC
Average SRM service units used per second.
Short name
SUPS
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SWAPERTX
Average number of swaps per transaction.
Short name
SWTX
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
SWAPIN
Average number of pages swapped in per second.
Short name
None
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA (non-expanded storage)
SWAPOUT
Average number of pages swapped out per second.
Short name
None
Tables
RPAG
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA (non-expanded storage)
SWAPS
Average number of pages swapped per second (swap-ins plus swap-outs).
Short name
SPS
Tables
RPAG
Appendix A. Data dictionary
351
Value range
Data type (INT)
Data type (COL, PC)
290
Commands
Filters
Environments
SWPCATG
Swap category.
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
Positive real numbers
I, 4
N, 11
Paging and storage resource display
OBTAIN
None
370, XA, ESA
CAT
RSWR
There are sixteen swap categories:
TERMINAL WAIT
DETECTED LONG THINK
LONG WAIT
DETECTED WAIT
EXCHANGE
UNILATERAL
REQUEST
ENQ EXCHANGE
AUX STORAGE SHORTAGE
REAL STORAGE SHORTAGE
TRANSITION TO NONSWAP
CENTRAL STORAGE (MVS/SP 4.2 and above)
SYSTEM PAGING (MVS/SP 4.2 and above)
OUT TOO LONG (MVS/SP 4.2 and above)
APPC WAIT (MVS/SP 4.2 and above)
TOTAL PHY SWAPS / SEC
Char, 21
Char, 21
Char, 21
Swap reasons resource display
OBTAIN, EXTRACT
RIF
370, XA, ESA
SYMBNAME
Symbolic name of the performance group, if one was defined during installation. For
COMPEXT data, this field refers to the performance group in the workload, not the profile.
Short name
None
Tables
PGNDETL, PGNCOMP
Value range
8-character descriptive name
Data type
C, 8
290
Performance group and profile/workload comparison
displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
SYSLEVEL
Operating system level, that is, MVS SP3.1.0.
Short name
SYS
Tables
RINF
Value range
Valid operating system levels
Data type
C, 8
290
General information resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
352
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
TARGMPL
Target MPL specified for the domain in the IPS.
Short name
TDM
Tables
RDOM
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
SRM domain resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
TCBPCT
Percent of CPU used by TCBs in the performance group.
Short name
TCBP
Tables
RPGN
Value range
0 - 100
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Performance group resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
TOTCPUTM
Total CPU time used by all transactions in the interval (in seconds). If multiple RMF-based
intervals are being combined into larger intervals, this figure reflects the number of
transactions in the larger combined interval.
Short name
None
Tables
PGNDETL
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
TOTIOTME
The average end-to-end service time per I/O to the device (in milliseconds).
Short name
TTME
Tables
RDAS
Value range
Positive real integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
DASD resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
TOTSTGK
Average amount of storage used by the performance group (in bytes).
Short name
TSTG
Tables
RPGN
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group resource displays
Appendix A. Data dictionary
353
Commands
Filters
Environments
OBTAIN, EXTRACT
RIF
370, XA, ESA
TRNSTIME
Average time required to complete a page transfer (in milliseconds).
Short name
TRTM
Tables
RPDS, RSDS
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Page data set and swap data set displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
TRXELAP
Total transaction time for performance group data, or total elapsed time for batch job,
started task, or TSO session data (in seconds). For COMPEXT data, this field refers to the
batch job, started task, or TSO session elapsed time in the workload, not the profile.
Short name
None
Tables
PGNDETL, BTCHDETL, STCDETL, TSODETL,
BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive integers
Data type (INT)
D, 8
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
All workload degradation and all batch job, started task,
and TSO session profile/workload displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
TRXRESP
Average transaction response time (in seconds). For COMPEXT data, this field refers to
the performance group in the workload, not the profile.
Short name
None
Tables
PGNDETL, PGNCOMP
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group and profile/workload comparison
displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
TRXTHINK
Average transaction think time (in seconds).
Short name
None
Tables
PGNDETL
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Performance group displays
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
354
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
TSOAVGAS
The average number of TSO users during the interval. If multiple RMF-based intervals are
combined into larger intervals, this figure reflects the number of TSO users during the
larger combined interval.
Short name
TAVG
Tables
RCPU
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
TSOMAXAS
The maximum number of TSO users during the interval. If multiple RMF-based intervals
are combined into larger intervals, this figure reflects the number of TSO users during the
larger combined interval.
Short name
TMAX
Tables
RCPU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
TSOMINAS
The minimum number of TSO users during the interval. If multiple RMF-based intervals
are combined into larger intervals, this figure reflects the number of TSO users during the
larger combined interval.
Short name
TMIN
Tables
RCPU
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
CPU resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
370, XA, ESA
TSOUSER
The TSO user ID. For COMPEXT data, this refers to the TSO user ID in the workload, not
the profile.
Short name
None
Tables
TSODETL, TSOCOMP
Value range
Positive integers
Data type
C, 7
290
TSO user session degradation displays and
profile/workload comparisons
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
TSWPSEC
Number of transition swaps per second. Transition swaps are swap-outs of address
spaces that were already marked as logically swapped when the physical swap-out was
Appendix A. Data dictionary
355
made. This field is used differently, depending on report table. For detailed information,
see the discussion of this field in tables RSWA and RSWR in Chapter 13, “Report element
tables,” on page 233.
Short name
TSWP
Tables
RSWA, RSWR
Value range
Positive real numbers
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
Swap activity and swap reasons resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
See description
VLFADDS
The number of objects in this class that were added to virtual storage during the interval.
Short name
VLFA
Tables
RVLF
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RVLF resource display
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
ESA 3.1 and above
VLFCLASS
The name of this class.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
VLFC
RVLF
User-defined
C, 8
RVLF resource display
DISPLAY, OBTAIN, EXTRACT
RIF
ESA 3.1 and above
VLFDELES
The number of objects in this class that were deleted from virtual storage during the
interval.
Short name
VLFD
Tables
RVLF
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RVLF resource display
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
ESA 3.1 and above
VLFHITS
The number of times an object in this class was found in cache, divided by the number of
times cache was searched for objects in this class, expressed as a percentage in tenths of
a percent.
Short name
VLFH
Tables
RVLF
Value range
1-100
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
356
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
290
Commands
Filters
Environments
RVLF resource display
DISPLAY, OBTAIN, EXTRACT
RIF
ESA 3.1 and above
VLFFULL
The storage used by objects in this class, divided by the maximum storage available for
objects in this class, expressed as a percentage in tenths of a percent.
Short name
VLFF
Tables
RVLF
Value range
1-100
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RVLF resource display
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
ESA 3.1 and above
VLFLRGST
The size of the largest object which VLF attempted to load into virtual storage.
Short name
VLFL
Tables
RVLF
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RVLF resource display
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
ESA 3.1 and above
VLFMXSTG
The maximum number of virtual storage pages that this class can use.
Short name
VLFM
Tables
RVLF
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RVLF resource display
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
ESA 3.1 and above
VLFRATE
The number times cache was searched for an object in this class during an interval,
divided by the number of seconds in the interval.
Short name
VLFR
Tables
RVLF
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RVLF resource display
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
ESA 3.1 and above
VLFTRIMS
The number of objects in this class that were trimmed from virtual storage during the
interval.
Short name
VLFT
Appendix A. Data dictionary
357
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
RVLF
Positive integers
I, 4
N, 10
F, 4
RVLF resource display
DISPLAY, OBTAIN, EXTRACT
RIF
ESA 3.1 and above
VLFVSUSD
The number of pages of virtual storage used by this class.
Short name
VLFU
Tables
RVLF
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
RVLF resource display
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
ESA 3.1 and above
VIRT
Virtual storage (in kilobytes). Used in conjunction with STORNAME.
Short name
None
Tables
RPAG
Value range
Positive integers
Data type (INT)
I, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
Paging and storage resource display
Commands
OBTAIN, EXTRACT
Filters
None
Environments
370, XA, ESA
VOLSER
Volume serial number.
Short name
Tables
Value range
Data type
290
Commands
Filters
Environments
WAITNAME
Wait reason.
Short name
Tables
Value range
358
VOLS
RCCH, RDAS, RPDS, RSDS, PGNDETL, BTCHDETL,
STCDETL, TSODETL, PGNCOMP, BTCHCOMP,
STCCOMP, TSOCOMP
Valid volume serial numbers
C, 6
DASD resource, page data set resource, swap data set
resource, and workload degradation and profile/workload
comparison displays
OBTAIN, EXTRACT, COMPEXT
RIF (resources)
None (workloads)
370, XA, ESA
None
PGNDETL, BTCHDETL, STCDETL, TSODETL,
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Wait reasons are defined in Table 14 on page 146.
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
For COMPEXT data, a blank DEVADD for ACTIVEIO,
QUEUEDIO, or RESRVEIO signifies that the observation
is a composite of all the extracted data for the wait reason
and is not device specific.
C, 20
C, 20
C, 10
All workload degradation and profile/workload comparison
displays
OBTAIN, EXTRACT, COMPEXT
None
370, XA, ESA
WAITPCT
Percent of the average transaction time spent on this wait reason. For COMPEXT data,
this field applies to the workload, not the profile.
Short name
None
Tables
PGNDETL, BTCHDETL, STCDETL, TSODETL,
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
All workload degradation and profile/workload comparison
displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
WAITTIME
Average transaction time spent on this wait reason (in seconds). For COMPEXT data, this
field applies to the workload, not the profile.
Short name
None
Tables
PGNDETL, BTCHDETL, STCDETL, TSODETL,
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
Value range
Positive integers
Data type (INT)
D, 4
Data type (COL, PC)
N, 10
Data type (SAS)
F, 4
290
All workload degradation and profile/workload comparison
displays
Commands
OBTAIN, EXTRACT, COMPEXT
Filters
None
Environments
370, XA, ESA
WKLNAME
Workload name.
Short name
Tables
Value range
None
PGNDETL, BTCHDETL, STCDETL, TSODETL,
PGNCOMP, BTCHCOMP, STCCOMP, TSOCOMP
For performance groups, valid workload names are:
ALL SYSTEM
PERIOD
SYMBOLIC PGN
PGN
For batch jobs, started tasks, and TSO sessions, valid
workload names are:
ACCOUNT NUM
PROGRAM NAME
Appendix A. Data dictionary
359
Data type
290
Commands
Filters
Environments
BATCH CLASS
BATCH JOB/STARTED TASK/TSO USER ID
C, 12
All workload degradation and profile/workload comparison
displays
OBTAIN, EXTRACT, COMPEXT
None
370, XA, ESA
WRHITCFW
The cache fast write write hit percentage for the cache device, expressed in tenths of a
percent. (3990 subsystem)
Short name
WHCF
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRHITDFN
The DASD fast write normal write hit percentage for the cache device expressed in tenths
of a percent. (3990 subsystem)
Short name
WHDN
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRHITDFS
The DASD fast write sequential write hit percentage for the cache device, expressed in
tenths of a percent. (3990 subsystem)
Short name
WHDS
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRHITDFT
The total DASD fast write write
a percent. (3990 subsystem)
Short name
Tables
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
360
hit percentage for the cache device, expressed in tenths of
WHDT
RCCH
Positive Integer
I, 4
N, 11
F,4
RCCH resource displays
DISPLAY, OBTAIN, EXTRACT
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Filters
Environments
RIF
XA, ESA
WRHITNM1
The normal write hit percentage for the cache device, expressed in tenths of a percent.
This statistic comes from the 3880 first storage director or the 3990 subsystem.
Short name
WHN1
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRHITNM2
The normal write hit percentage for the cache device, expressed in tenths of a percent.
This statistic comes from the 3880 second storage director.
Short name
WHN2
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRHITSEQ
The sequential write hit percentage for the cache device, expressed in tenths of a percent.
This statistic comes from the 3990 subsystem.
Short name
WHSQ
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRHITTOT
The total write hit percentage for the cache device, expressed in tenths of a percent.
Short name
WHTT
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
DISPLAY, OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRITENM1
The normal write requests for cache device, expressed in tenths of a write per second.
This statistic comes from the 3880 first storage director.
Short name
WRN1
Tables
RCCH
Appendix A. Data dictionary
361
Value range
Data type (INT)
Data type (COL, PC)
Data type (SAS)
290
Commands
Filters
Environments
Positive Integer
I, 4
N, 11
F, 4
RCCH resource displays
OBTAIN, EXTRACT
RIF
XA, ESA
WRITENM2
The normal write requests for cache device, expressed in tenths of a write per second.
This statistic comes from the 3880 second storage director.
Short name
WRN2
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRITESQ1
The sequential write requests for cache device, expressed in tenths of a write per second.
This statistic comes from the 3880 first storage director.
Short name
WRS1
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
WRITESQ2
The sequential write requests for cache device, expressed in tenths of a write per second.
This statistic comes from the 3880 second storage director.
Short name
WRS2
Tables
RCCH
Value range
Positive Integer
Data type (INT)
I, 4
Data type (COL, PC)
N, 11
Data type (SAS)
F, 4
290
RCCH resource displays
Commands
OBTAIN, EXTRACT
Filters
RIF
Environments
XA, ESA
362
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Appendix B. Documentation library
This appendix contains information about the publications in the OMEGAMON XE on z/OS library and
about other publications related to OMEGAMON XE on z/OS.
See IBM Tivoli Monitoring and OMEGAMON XE Products: Documentation Guide, SC23-8816, for
information about accessing and using the publications. You can find the Documentation Guide in the IBM
Tivoli Monitoring and OMEGAMON XE Information Center at http://publib.boulder.ibm.com/infocenter/
tivihelp/v15r1/.
To find a list of new and changed publications, click What's new on the Welcome page of the IBM Tivoli
Monitoring and OMEGAMON XE Information Center. To find publications for the previous version of a
product, click Previous information centers on the Welcome page for the product.
OMEGAMON XE on z/OS library
The following documents are available for OMEGAMON XE on z/OS:
v Program Directory GI13-2207
Contains information about the material and procedures associated with the installation of IBM Tivoli
OMEGAMON XE on z/OS. The Program Directory is intended for the system programmer responsible
for program installation and maintenance.
v Planning and Configuration Guide SC27-4032
Provides information that helps plan the deployment and configuration of OMEGAMON XE on z/OS and
the required common services component. It also provides detailed instructions for configurating product
components. This document is intended for system administrators and others who are responsible for
configuring OMEGAMON XE on z/OS.
v User's Guide SC27-4028
Introduces the features, workspaces, attributes, and predefined situations for the OMEGAMON XE on
z/OS product and supplements the user assistance provided with this product. This document is written
for data center operators and analysts responsible for monitoring and troubleshooting system
performance and availability or performing trend analysis for resource planning.
v Troubleshooting Guide GC27-4029
Provides explanations for the messages issued by the OMEGAMON XE on z/OS product, its
OMEGAMON II for MVS component, and common agent components. This booksalso provides
troubleshooting advice for installation and configuration, security, and usage problems, and instructions
for setting up tracing on z/OS.
v Parameter Reference SC27-4033
Provides names and descriptions for all OMEGAMON XE on z/OS configuration parameters.
v OMEGAMON for MVS User's Guide SC27-4030
Describes the features and commands used in OMEGAMON for MVS. Reference information for
OMEGAMON major and minor commands is included by functional area, along with a description of the
following features: User Profile Facility, Exception Analysis, CSA Analyzer, End-to-End Response Time
Feature, Bottleneck Analysis, DEXAN, Impact Analysis, Workload Profile Facility.
v OMEGAMON for MVS User's Guide SC27-4031
Contains complete descriptions of OMEGAMON for MVS commands, organized alphabetically by
command name. Includes a chapter on “Command Groupings” that is an introduction organized by topic
(exception analysis, hiperspace, paging, and so on) where you can refresh your memory as to the
proper spelling of a command or keyword.
v EPILOG User's Guide SC27-4034
Describes the basic reporting features of EPILOG for MVS. The introduction provides a product
overview and a discussion of the EPILOG approach to performance management. The rest of the
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
363
manual explains how to use the reporter, including the various types of reports and the use of the
DISPLAY command. Topics, such as advanced reporting options, the Workload Profile Facility,
exception filtering, exporting historical data, and reporting with SAS graphics are also documented.
v EPILOG Command Reference SC27-4035
Contains complete descriptions of EPILOG for MVS commands, organized alphabetically by command
name.
OMEGAMON XE and Tivoli Management Services on z/OS common
library
The books in this library are common to some or all of the OMEGAMON XE products or Tivoli
Management Services on z/OS:
v Quick Start Guide, GI11-8918
Provides an overview of the installation and setup process for a monitoring agent on z/OS.
v Common Planning and Configuration Guide, SC23-9734
Covers planning and configuration information common to the OMEGAMON XE monitoring agents and
to the components of Tivoli Management Services on z/OS.
v Common Parameter Reference SC14-7280
Provides reference information on parameters used for setting up runtime environments and configuring
hub and remote Tivoli Enterprise Monitoring Servers on z/OS.
v OMEGAMON Enhanced 3270 User Interface Guide SC22-5426
Describes the features of the interface and provides operating instructions and reference material.
v Upgrade Guide, SC23-9745
Provides an overview and instructions for performing the upgrades from prior versions of OMEGAMON
XE monitoring agents and Tivoli Management Services components.
v End-to-End Response Time Feature Reference, SC27-2303
Documents the End to End Response Time feature, a common component used by four OMEGAMON
XE monitoring agents on z/OS: CICS, z/OS, IMS, and Mainframe Networks.
v Reports for Tivoli Common Reporting, SC27-2304
Provides information about the Tivoli Common Reporting tool that is specific to products that run under
the Tivoli Enterprise Portal and use the Tivoli Data Warehouse database.
OMEGAMON II for MVS V520 library
The OMEGAMON II® documentation has not been updated since V520. Any changes relevant to
configuration of the version currently incorporated into OMEGAMON XE on z/OS V4.2.0 are documented
in the IBM Tivoli OMEGAMON XE on z/OS: Planning and Configuration Guide. The existing documents
refer to the Candle Support structure rather than to IBM software support. References to Candle Support
processes and procedures are invalid. Direct questions to IBM Software Support. For details about the
IBM support structure, see “Support information” on page 367.
The following information sources are available in the OMEGAMON II for MVS V520 library.
v Configuration and Customization Guide, GC32-9277
Describes how to configure and customize the OMEGAMON II for MVS product. It provides background
on the product components, addresses maintenance and migration considerations, gives an overview of
the configuration and customization process, and documents step-by-step procedures.
v User's Guide, GC32-9280
Contains an overview of OMEGAMON II features, the types of panels displayed, and how to navigate
from one panel to another; instructions for adjusting an OMEGAMON II environment; usage scenarios
describing how to use OMEGAMON II to monitor realtime and historical performance; instructions for
364
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
using some of the commands for creating OMEGAMON screen spaces described in the OMEGAMON
for MVS Command Language Reference Manual and commands for generating EPILOG resports
described in the EPILOG for MVS Command Language Reference Manual.
v OMEGAMON for MVS Command Language Reference Manual, GC32-9276
Provides the syntax and available keywords for OMEGAMON II for MVS commands
v EPILOG for MVS Command Language Reference Manual, GC32-9275
Documents the syntax and available keywords for EPILOG for MVS commands.
IBM Tivoli Monitoring library
The following publications provide information about IBM Tivoli Monitoring V6.2 and about the commonly
shared components of Tivoli Management Services:
v Quick Start Guide, GI11-8058
Introduces the components of IBM Tivoli Monitoring.
v Installation and Setup Guide, GC32-9407
Provides instructions for installing and configuring IBM Tivoli Monitoring components on Windows, Linux,
and UNIX systems.
v Program Directory for IBM Tivoli Management Services on z/OS, GI11-4105
Gives instructions for the SMP/E installation of the Tivoli Management Services components on z/OS.
v Configuring the Tivoli Enterprise Monitoring Server on z/OS, SC27-2313
Gives detailed instructions for using the Configuration Tool to configure Tivoli Enterprise Monitoring
Server on z/OS systems. Includes scenarios for using batch mode to replicate monitoring environments
across the z/OS enterprise. Also provides instructions for setting up security and for adding application
support to a Tivoli Enterprise Monitoring Server on z/OS.
v Administrator's Guide, SC32-9408
Describes the support tasks and functions required for the Tivoli Enterprise Portal Server and clients,
including Tivoli Enterprise Portal user administration.
v Tivoli Enterprise Portal online help
Provides context-sensitive reference information about all features and customization options of the
Tivoli Enterprise Portal. Also gives instructions for using and administering the Tivoli Enterprise Portal.
v User's Guide, SC32-9409
Complements the Tivoli Enterprise Portal online help. The guide provides hands-on lessons and detailed
instructions for all Tivoli Enterprise Portal features.
v Command Reference, SC32-6045
Provides detailed syntax and parameter information, as well as examples, for the commands you can
use in IBM Tivoli Monitoring.
v Troubleshooting Guide, GC32-9458
Provides information to help you troubleshoot problems with the software.
v Messages, SC23-7969
Lists and explains messages generated by all IBM Tivoli Monitoring components and by z/OS-based
Tivoli Management Services components (such as Tivoli Enterprise Monitoring Server on z/OS and
TMS:Engine).
Other sources of documentation
You can also obtain technical documentation about Tivoli Monitoring and OMEGAMON XE products from
the following sources:
v IBM Tivoli Integrated Service Management Library
http://www.ibm.com/software/tivoli/opal
Appendix B. Documentation library
365
The Integrated Service Management Library is an online catalog that contains integration documentation
as well as other downloadable product extensions. This library is updated daily.
v Redbooks
http://www.redbooks.ibm.com/
IBM Redbooks®, Redpapers, and Redbooks Technotes provide information about products from platform
and solution perspectives.
v Technotes
You can find Technotes through the IBM Software Support Web site at http://www.ibm.com/software/
support/probsub.html, or more directly through your product Web site, which contains a link to
Technotes (under Solve a problem).
Technotes provide the latest information about known product limitations and workarounds.
366
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Support information
If you have a problem with your IBM software, you want to resolve it quickly. IBM provides the following
ways for you to obtain the support you need:
Online
Go to the IBM Software Support site at http://www.ibm.com/software/support/probsub.html and
follow the instructions.
IBM Support Assistant
The IBM Support Assistant (ISA) is a free local software serviceability workbench that helps you
resolve questions and problems with IBM software products. The ISA provides quick access to
support-related information and serviceability tools for problem determination. To install the ISA
software, go to http://www.ibm.com/software/support/isa.
Troubleshooting Guide
For more information about resolving problems, see the product's Troubleshooting Guide.
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
367
368
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Notices
This information was developed for products and services offered in the U.S.A. IBM may not offer the
products, services, or features discussed in this document in other countries. Consult your local IBM
representative for information on the products and services currently available in your area. Any reference
to an IBM product, program, or service is not intended to state or imply that only that IBM product,
program, or service may be used. Any functionally equivalent product, program, or service that does not
infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document.
The furnishing of this document does not give you any license to these patents. You can send license
inquiries, in writing, to:
IBM Director of Licensing
IBM Corporation
North Castle Drive
Armonk, NY 10504-1785 U.S.A.
For license inquiries regarding double-byte (DBCS) information, contact the IBM Intellectual Property
Department in your country or send inquiries, in writing, to:
IBM World Trade Asia Corporation
Licensing
2-31 Roppongi 3-chome, Minato-ku
Tokyo 106, Japan
The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law:
INTERNATIONAL BUSINESS MACHINES CORPORATION PROVIDES THIS PUBLICATION "AS IS"
WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT, MERCHANTABILITY OR FITNESS
FOR A PARTICULAR PURPOSE.
Some states do not allow disclaimer of express or implied warranties in certain transactions, therefore, this
statement might not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically
made to the information herein; these changes will be incorporated in new editions of the publication. IBM
may make improvements and/or changes in the product(s) and/or the program(s) described in this
publication at any time without notice.
Any references in this information to non-IBM Web sites are provided for convenience only and do not in
any manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of
the materials for this IBM product and use of those Web sites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without
incurring any obligation to you.
Licensees of this program who wish to have information about it for the purpose of enabling: (i) the
exchange of information between independently created programs and other programs (including this one)
and (ii) the mutual use of the information which has been exchanged, should contact:
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
369
IBM Corporation
2Z4A/101
11400 Burnet Road
Austin, TX 78758 U.S.A.
Such information may be available, subject to appropriate terms and conditions, including in some cases
payment of a fee.
The licensed program described in this document and all licensed material available for it are provided by
IBM under terms of the IBM Customer Agreement, IBM International Program License Agreement or any
equivalent agreement between us.
If you are viewing this information in softcopy form, the photographs and color illustrations might not be
displayed.
Trademarks
IBM, the IBM logo, and ibm.com® are trademarks or registered trademarks of International Business
Machines Corporation in the United States, other countries, or both. If these and other IBM trademarked
terms are marked on their first occurrence in this information with a trademark symbol (® or ™), these
symbols indicate U.S. registered or common law trademarks owned by IBM at the time this information
was published. Such trademarks may also be registered or common law trademarks in other countries. A
current list of IBM trademarks is available on the Web at "Copyright and trademark information" at
http://www.ibm.com/legal/copytrade.shtml.
Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel
SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its
subsidiaries in the United States and other countries.
Linux is a trademark of Linus Torvalds in the United States, other countries, or both.
Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the
United States, other countries, or both.
UNIX is a registered trademark of The Open Group in the United States and other countries.
Other company, product, and service names may be trademarks or service marks of others.
370
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Index
A
A-matrix 161
A-matrix commands 24
abbreviations 25
ACCOUNT keyword 36, 171, 280, 290
ACCT keyword 36, 171
ACCTCDE data element 284, 293, 294, 296, 301
active datastore list 106, 111
ACTIVEIO keyword 58, 146
ADD operand 107
AIO keyword 58, 146
ALL operand 92, 172
ALLOC CLIST 111
allocating a SAS file 210
allocating datastores 106
ALLWAITS operand 150
AMATRIX keyword 96, 130, 165, 166
ANY operand 150
APPC 49
APPCWAIT keyword 58, 146
APPEND in OBTAIN command 195
APS keyword 58, 146
asterisks 24, 25
AUTO
See AUTOMATIC keyword
AUTO keyword 96
automatic analysis 161
AUTOMATIC keyword 96, 163
AUX keyword 58, 146
AUXSTOR keyword 58, 146
AVERAGE keyword 46, 54
AVG keyword 46
AVGCTRX data element 265, 302
AVGNSWAP data element 250, 302
AVGRESP data element 302
AVGTRXIN data element 265, 302
AVGUGOUT data element 250, 303
AVGUIN data element 250, 303
AVGUOUT data element 250, 303
AVGURDY data element 250, 303
AVGWKST data element 265, 304
B
BACKUP keyword 58, 146
BAND keyword 91, 92, 171
BATAVGAS data element 242, 304
batch workloads 35, 39, 41
job displays 39
selecting for display 35
step displays 41
BATMAXAS data element 242, 304
BATMINAS data element 242, 304
BATNAME override 215
BKP keyword 58, 146
blank spaces 24
blinking displays 105
© Copyright IBM Corp. 1990, 1992, 1997, 2009, 2012
bottleneck analysis 6
BPSSCCH1 data element 238, 304
BPSSCCH2 data element 238, 305
BTCHCOMP data set 293
BTCHDETL data set 284
BUFND keyword 20
BUFNI keyword 20
C
cache subsystem data 64
CACHSTT1 data element 236, 305
CACHSTT2 data element 236, 305
CAN keyword 58, 146
CANCEL keyword 58, 146
capacity planning 11, 201
CCHDASDN data element 238, 305
CENSTOR keyword 58
CENTSTOR keyword 146
CHANNEL override 215
channel utilization data 65
CHNNUM data element 239, 305
CHNSETID data element 239, 306
CHNTYPE data element 239, 306
CHPID data element 239, 257, 306
CHPINSEC data element 257, 307
CIL keyword 58, 146
CILOCATE keyword 58, 146
CIOPRSEC data element 240, 307
CLASS keyword 36, 171, 280, 290
CLEAR keyword 95, 190, 231
CLISTS 19
KEDSPF 19
KEDTSO 19
KEPTSO 19
CLS keyword 36, 171
CMAT command 162
CMB
See COMBINE keyword
CNTRLMOD data element 236, 307
collector 4
color terminals 104
COM keyword 58, 146
COMBINE keyword 97, 99, 100, 110
with batch jobs, started tasks, and TSO
sessions 100
with performance groups 97
with resources 97, 99
COMBINE keyword with exception analysis
combining time intervals 97
command abbreviation 25
command reshow 103
command syntax 24
commas 24
comment lines 24
COMMON keyword 58, 146
COMPARE command 24, 183, 186
COMPARE Command 188
151
371
COMPARE Command (continued)
examples 188
COMPARE keywords 187, 188
IODEVICE 188
LINECNT 188
PROFNAME 187
REPORTIF 187
SUMMARY 188
SUMWAIT 188
SYSID 187
time/date 187
workload 187
COMPEXT command 229, 230
examples 230
COMPEXT keywords 229, 230
date and time 229
PROFNAME 229
REPORTIF 230
SYSID 230
COMPLETE keyword 103
CONNTIME data element 248, 307
consolidated workloads 12
CONTIDX data element 250, 307
continuation character 24
CONTROL keywords 102, 103, 104
COLOR 104
FOLDOFF 104
FOLDON 104
LOG 104
MODE 102
RECALL 103
TITLE 103
control units 76
CPCONFIG data element 240, 308
CPCTBUSY data element 240, 308
CPENVIRN data element 240, 308
CPLPSTAT data element 240, 308
CPLSHARE data element 240, 308
CPLTBUSY data element 240, 308
CPMFSTAT data element 240, 309
CPPCTBSY data element 256, 309
CPU capping 75
CPU keyword 58, 146
CPU utilization data 66
CPUFLAG data element 243, 244, 245, 309
CPUID data element 242, 310
CPUMODEL data element 252, 310
CPUSERL data element 252, 310
CPUTIL keyword 58, 146
CPUTRXTM data element 284, 311
CPUWAIT keyword 58, 146
CPW keyword 58, 146
CREATEM command 162
cross-system contention 108
cross-system reporting 44
CSRVTIME data element 240, 311
CSS keyword 58, 146
CU operand 150
CUDEFSEC data element 257, 311
CUID data element 257, 311
CUPCTBSY data element 256, 311
372
CURMPL data element
250, 312
D
D navigational command 129
D operand 150
DASD 44, 68, 109, 134
device activity 109
utilization data 68
viewing workload utilization 134
workload utilization 44
DASD operand 150
DASDCCHN data element 238, 312
DASDCCHS data element 238, 312
DASDFWST data element 237, 312
data buffers 20
data sample 174
DATASTOR command 107
datastore 4, 20, 27, 111
EPILOG datastore 4
initialization 111
inquiry 111
opening 20
sample 27
status 111
datastore list 106
datastores, profile 167
date 92, 171, 172
defaults 172
format 172
Gregorian 92, 172
Julian 92, 172
keywords 171
date and time 91, 92, 93
defaults 92
keywords 91
relative 93
date format 92
date keywords format 92
DAY keyword 92, 172
DAYOFWK keyword 92, 172
DCUBSY data element 248, 312
DDEVBSY data element 248, 313
DDNAME override 215
degradation data 5
DEL keyword 58, 146
DELETEDS keyword 58, 146
DELETEJOB keyword 58, 146
DELETEM command 163
delimiters 24
DET keyword 46, 58, 146
DETAIL keyword 46
Detail Wait Reason 49, 50, 51
Active I/O 49
APPC Wait 49
Auxiliary Storage 49
Central Storage 49
Common Page-In Wait 49
Detected Wait 49
Disk Mount Pending 49
ECB Wait 49
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Detail Wait Reason (continued)
with Stimer 49
Enqueue Exchange Swap 50
Enqueue Wait 50
Exchange Swap 50
HSM Delay 50
Backup 50
JES C/I 50
Migrate 50
Read CTL 50
Recall 50
Recover 50
TSO HLIST 50
HSM Delay (Delete) 50
JES Cancel 50
JES Delay 50
Requeue 50
Status 50
Sysout 50
JES Delay (Delete) 50
Long Wait Swap 50
Out Too Long 50
Private Page-In Wait 50
Queued I/O 50
Real Storage Swap 50
Request Swap 50
Reserve I/O 50
SRM Delay 50
MPL 50
Requeue 50
Stimer Wait 50
Swap Page-In Wait 50
Swapped with WTOR 50
System Paging 50
Tape Mount Pending 51
Terminal Input Wait 51
Terminal Output Wait 51
Transition Swap 51
Unilateral Swap 51
Using CPU 51
Waiting for CPU 51
Waiting for MVS Lock 51
Waiting For Staging 51
DETECTED keyword 58, 146
DEVADD data element 236, 247, 263, 267, 281, 284,
291, 293, 294, 296, 313
DEVICE override 215
device types and addresses 149
DEVQTIME data element 247, 313
DEVTYPE data element 263, 267, 314
DFWBYPSS data element 238, 314
DIOPRSEC data element 247, 314
DISCTIME data element 248, 314
DISK operand 150
DISKMOUNT keyword 58, 146
DISPLAY command 26, 29, 91, 95, 97, 143, 179
combining data 97
exception analysis 95, 143
overview 26
selecting the time period 91
setting defaults 95
DISPLAY command (continued)
workloads 29
DISPLAY Command 180, 182
examples 182
DISPLAY keywords 27, 30, 32, 33, 36, 37, 44, 46, 54,
55, 57, 60, 64, 65, 66, 68, 73, 74, 76, 78, 81, 82, 84,
85, 87, 89, 91, 95, 97, 103, 104, 143, 145, 151, 163,
165, 181, 183, 228
ACCOUNT 36
AMATRIX 165
AUTOMATIC 163
AVERAGE 54
BAND 91
CLASS 36
COMBINE 97
COMPLETE 103
DETAIL 46
display options 181
ELAPSED 143, 145
ENDDATE 91
ENDTIME 91
EXDATE 91
EXWAIT 57
INTERVAL 36
IODEVICE 181
JDAS 44
JOB 36
LASTMONTH 91
LASTWEEK 91
LASTYEAR 91
LIMIT 103, 143
LOGOFF 104
LOGON 104
MAXSCALE 55
MERGE 27
PDAS 44
PERFGROUP 32
PGN 32
PGPERIOD 33
PLOTMIN 54
PROFNAME 181
PROGRAM 37
RANGE 91
RCCH 64
RCHN 65
RCPU 66
RDAS 68
RDOM 73
REPORTIF 95, 143, 151, 181
RESPONSE 143, 145
RINF 74
RLCU 76
RPAG 78
RPDS 81
RPGN 82
RSDS 84
RSRM 84
RSWA 85
RSWR 87
RVLF 89
SELECTIF 95, 143, 151
Index
373
DISPLAY keywords (continued)
SINGLE 97
STARTDATE 91
STARTTIME 91
STARTTSK 37
STC 37
STEP 37
SUMMARY 46
SYMBOLIC 33
SYSID 27, 181, 183, 228
SYSTEM 30
THISMONTH 91
THISWEEK 91
THISYEAR 91
time/date 181
TODAY 91
TOTAL 54
TSOUSER 37
TWAITOFF 60
TWAITON 60
workload 181
XPG 33
YESTERDAY 91
display options 181
display screen 21
splitting 21
swapping 21
display scrolling 102, 103
COMPLETE keyword 103
scrolling mode 102
DLCUID data element 247, 256, 314
DLD keyword 58, 146
DMAT command 163
DMP keyword 58, 146
documentation
See publications
domains 73
DOMNUM data element 250, 315
DPBUSY data element 315
DPCTBSY data element 256, 315
DPCTBUSY data element 247, 315
DPRODIDX data element 291, 293, 294, 296, 315
DROP operand 107
DSNAME override 213
DSWPPHSC data element 271, 276, 316
DSXAFLAG data element 240, 245, 248, 272, 316
DTRXELAP data element 293, 294, 296, 316
DTRXRESP data element 291, 316
DWAITTIME data element 291
DWAITTME data element 293, 294, 296, 317
dynamic i/o reconfiguration 5
E
ECB keyword 59, 146
ECBSTIMER keyword 58, 146
ECBWAIT keyword 59, 146
ECS keyword 58, 146
EDATE
See ENDDATE keyword
EDATE data element 317
374
EDATE keyword 92, 172
EDATE variable 240, 246, 248, 251, 254, 258, 262,
263, 266, 268, 269, 272, 277, 281, 284, 286, 288
EDATTIME data element 240, 246, 249, 251, 255,
258, 262, 263, 266, 268, 269, 272, 277, 281, 284,
291, 293, 294, 296, 317
EDSLIST 111
EEX keyword 59, 146
ELAPSED keyword 143, 145
ELEMENTS in OBTAIN command 194
EMS
See Exception Management System
END 19
ENDDATE keyword 91, 92, 172
ENDTIME keyword 91, 92, 172
ENQ keyword 59, 146
ENQEXC keyword 59, 146
ENQNAME data element 281, 284, 291, 293, 295,
296, 317
enqueue exception operands 150
ENQUEUE keyword 59, 146
enqueue names 150
enqueue types 149, 150
environments 109
EPPARM ddname 111
equal signs 24
ESASWPSC data element 272, 277, 318
ESAVG data element 261, 318
ESDSWPSC data element 272, 277, 318
ESFRREAL data element 261, 318
ESINST data element 261, 318
ESMAX data element 261, 319
ESMIGRAT data element 269, 319
ESMIN data element 261, 319
ESONLINE data element 261, 319
ESTOAUX data element 261, 320
ESTSWPSC data element 272, 277, 320
ETIME
See ENDTIME keyword
ETIME data element 320
ETIME keyword 92, 172
ETIME variable 240, 246, 249, 251, 255, 258, 262,
263, 266, 268, 270, 272, 277, 281, 284, 286, 288
EVERY operand 150
exception analysis 95, 143, 151, 170, 187
keyword format 143
on combined data 151
on the COMPARE command 187
on the PROFILE command 170
exception filters 143, 155
resource 155
workload 143
Exception Management System 5
EXCHANGE keyword 59, 146
EXD keyword 92, 172
EXDATE keyword 91, 92, 172
execution states 144, 146
NO prefix 144
EXG keyword 59, 146
exporting EPILOG data 193
EXT command 206
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
EXTDTIME data element 241, 246, 249, 251, 255,
258, 262, 264, 266, 268, 270, 272, 277, 281, 284, 320
EXTRACT command overview 227
extracted data 280
EXW keyword 46
EXWAIT keyword 46, 57
F
fast-path navigation 125
filters 143, 155
resource 155
workload 143
FIXED data element 260, 321
FORECAST override 215
FORMAT in OBTAIN command 194
FRIDAY operand 92, 172
G
graph replay procedure
212
H
HIGHSTEP data element
HLIST keyword 59, 147
HLS keyword 59, 147
HOLD mode 102
hyphen 24
284, 321
I
I/O exception operands 150
I/O queuing 76
IBM Support Assistant 367
ICS data element 252, 321
IKT00405I 105
index buffers 20
INFFLAG data element 253, 254, 321
INHBCCH1 data element 238, 322
INHBCCH2 data element 238, 322
initial datastore list 106
INQUIRE command 111, 112
INSUF SCRN SIZE 21
interval boundary 110
INTERVAL keyword 36
INTVLSRV data element 250, 322
IODEFSEC data element 256, 322
IODEVICE keyword 180, 181, 185, 188
IOINTRPT data element 323
IOINTRPT variable 242
IOPERSEC data element 263, 267, 323
IOPIOQLN data element 256, 323
IOPIOSEC data element 256, 323
IOPROCID data element 256, 324
IOPTLDEF data element 256, 324
IOSQTIME data element 248, 324
IPS data element 252, 324
ISA 367
ISPF SPLIT function 137
ISPF split-screen operation 20
ISPF SWAP function 137
ITRGMPL data element 324
J
J navigational command 131
JCL KEPPROC 19
JCL KEPPROFJ 169
JDAS keyword 44, 134
JESCLASS data element 284, 293, 325
JESNUM data element 284, 293, 295, 296, 325
job 40
JOB keyword 36, 280, 290
JOBNAME data element 284, 293, 325
JOBNAME keyword 36, 171
JOBSTATUS keyword 59, 147
JST keyword 59, 147
K
KEBRSET utility 107
KEDPMJCL 214
KEDPMTSO 214
KEPBJOBC report program
KEPBJOBT report program
KEPBLJBC report program
KEPBLJBT report program
KEPBPGML report program
KEPBTCHM macro 214
KEPCOLLC macro 214
KEPCOLLS macro 214
KEPCOLOR macro 214
KEPCOLOR override 215
KEPDELGR macro 214
KEPGNMA macro 214
KEPILGPM procedure 211
KEPMREPL macro 214
KEPPLPGP report program
KEPPLPGT report program
KEPPMJCL 214
KEPPMTSO 214
KEPPPGNP report program
KEPPPGNT report program
KEPPROFJ 169
KEPRCHBH report program
KEPRCHBP report program
KEPRCHIH report program
KEPRCHIP report program
KEPRCHNS report program
KEPRDASD report program
KEPRLCBH report program
KEPRLCBP report program
KEPRLCIH report program
KEPRLCIP report program
KEPRLPAP report program
KEPRLPAS report program
KEPRPAGP report program
KEPRPAGS report program
KEPRPGNC report program
KEPRPGNS report program
221
222
221
222
226
220
221
220
221
216
217
216
217
223
225
216
217
216
217
218
218
218
218
219
219
Index
375
KEPSRCMP 115, 117
KEPSRDIR 115
KEPSROPM 115
KEPSRSPG 115, 116
KEPSUMMG macro 214
KEPVERSI macro 214
keyword abbreviation 25
KPMJCL6 214
KPMTSO6 214
L
L operand 150
LASTMONTH keyword 91, 92, 172
LASTWEEK keyword 91, 92, 172
LASTYEAR keyword 91, 92, 172
LCH operand 150
LCK keyword 59, 147
LCUCPST data element 258, 325
LCUERROR data element 258, 326
LCUIOPLN data element 256
LCUIOQLN data element 326
LCUIOSEC data element 256, 326
LCUTYPE data element 257, 326
LEAD override 215
libraries
IBM Tivoli Monitoring 365
library, OMEGAMON XE on z/OS 363
LIM keyword 96
LIMIT keyword 96, 103, 143
LINECNT keyword 188
LISTM command 163
LMAT command 163
LMN keyword 92, 172
Logical channel operand 150
logical control units 76
logical partitiong 68
logical partitioning 76, 309, 321
logical tuning 11
logical tuning approach 4, 9
LOGOFF keyword 96, 104
LOGON keyword 96, 104
LON keyword 59, 147
long form 25
LONGWAIT keyword 59, 147
LSWPRSEC data element 271, 277, 326
LTPPRSEC data element 271, 327
LWK keyword 92, 172
LYR keyword 92, 172
M
M operand 150
MACHFLAG data element 249, 258, 262, 270, 272,
278, 327
major enqueue names 150
MAJOR operand 150
management reporting 201
masks 25
MAX keyword 46
MAXJOB override 215
376
MAXMPL data element 250, 327
MAXSCALE keyword 46, 55
MERGE keyword 27, 108
MIG keyword 59, 147
MIGRATE keyword 59, 147
MINMPL data element 250, 327
minus sign 91, 93
MISC SYS operand 150
MISC USR operand 151
MODE data element 252, 328
Mode field 102
MONDAY operand 92, 172
MPL 99
MSS keyword 59, 147
multi-system configuration 108
multi-system environment 17
MVS environments 109
MVSLOCK keyword 59, 147
N
navigation 125
navigational commands 129
NO prefix 144
NONFIXED data element 260, 328
NOOUTLIER 170
NOPLOT keyword 170
NOPRODIDX keyword 96
NUMICMB data element 241, 246, 249, 251, 255, 259,
262, 264, 266, 268, 270, 273, 278, 281, 285, 328
NUMPHYS data element 328
NUMTRX data element 281, 329
O
OBTAIN command 193, 195, 196, 197
examples 197
keywords 193
output files 195, 196, 197
defined 195
examples 197
for mainframe 195, 196
for PC 197
syntax 193
OLS keyword 59, 147
OMEGAMON II for MVS
library 364
ONE mode 102
OPAL documentation 365
OPENDSN data element 248, 329
OPT data element 252, 329
OSLEVEL data element 241, 242, 246, 249, 251, 259,
329
OTRGMPL data element 329
OUTLIER keyword 170
outlier, definition 174
OUTLONG keyword 59, 147
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
P
P navigational command 131
PACCTCDE data element 293, 295, 296, 330
PAG keyword 59, 147
page data set resource data 81
PAGE mode 102
PAGEIN data element 261, 330
PAGEIN keyword 59, 147
PAGEOUT data element 261, 330
PAGES data element 261, 330
PAGESEP command 24, 142
paging 78
PAGPERIO data element 263, 330
parentheses 24
PARTNAME data element 331
PARTNUM data element 331
PARTWGT data element 331
PATHQTME data element 247, 331
PAVGELAP data element 293, 295, 296, 331
PAVGTRX data element 291, 332
PCTCMPLX data element 332
PCTCPMVS data element 242, 332
PCTCPSRB data element 242, 332
PCTCPTCB data element 242, 333
PCTCPU data element 242, 333
PCTFULL data element 263, 267, 333
PCTINUSE data element 263, 267, 333
PCTMGPLX data element 334
PCTPARTA data element 333
PCTPRMCP data element 334
PCTPRTCP data element 334
PCTRES data element 265, 334
PCTRSVDL data element 248, 335
PCTTPI data element 242, 335
PDAS keyword 44
PDATTIME data element 291, 293, 295, 296, 335
PDSFLAG data element 263, 335
PEDATTIM data element 291, 293, 295, 296, 335
PENDTIME data element 248, 335
percent sign 145
PERF data element 336
PERFGROUP keyword 32, 171
performance groups 31, 33, 82
cross-performance group analysis 33
degradation analysis 31
performance period analysis 33
resource data 82
symbolic 33
performance problem solving 201
PERIOD data element 336
PF key assignments 21, 137
ISPF SPLIT function 137
ISPF SWAP function 137
PFK command 24
PGM keyword 37, 171
PGMNAME data element 285, 293, 295, 336
PGN data element 264, 281, 291, 336
PGN keyword 32, 171, 280, 290
PGN override 216
PGNCOMP data set 290
PGNDETL data set 280
PGNFLAG data element 266, 337
PGNTYPE data element 281, 291, 337
PGP data element 265, 281, 291, 337
PGP keyword 290
PGPERIOD keyword 33, 83, 171
PGPRSWPI data element 271, 337
PGPRSWPO data element 271, 338
PIOPRSEC data element 265, 338
PJESCLAS data element 293, 338
PJOBNAME data element 293, 338
PLOT keyword 170, 175
PLOTMIN keyword 46, 54, 96
plus sign 91, 93
PMIN keyword 46, 96
PNAME keyword 229
PPGMNAME data element 293, 295, 338
PPGN data element 291, 339
PPGNTYPE data element 291, 339
PPGP data element 291, 339
PPRODIDX data element 291, 293, 295, 296, 339
PR/SM 243, 253
PRDXDEF keyword 96, 141
problem solving 11
PROCNAME data element 285, 339
PRODIDX data element 282, 285, 291, 293, 295, 296,
339
PRODIDX keyword 96
producing reports 209
product display functions 111
productivity index definitions 141
PRODUCTS command 114
PROFILE command 169, 176
examples 176
Profile datastore 167
PROFILE keywords 169, 170
NOOUTLIER 170
NOPLOT 170
OUTLIER 170
PLOT 170
PROFNAME 170
SELECTIF 170
SYSID 170
TEST 170
time/date 170
workloads 169
profile reports 179, 180
standard 179
with device detail 180
profile utility 167
profile workload comparisons 7
Profile/Workload Comparison report 184
Profile/Workload Comparison Summary Report 186
PROFNAME data element 291, 293, 295, 296, 340
PROFNAME keyword 170, 181, 187, 229
PROG471 105
PROGRAM keyword 37, 171, 280, 290
PSDATTIM data element 291, 293, 295, 296, 340
PSMFID data element 291, 293, 295, 296, 340
PSO keyword 60, 147
PSTCNAME data element 295, 340
PSWPRSEC data element 271, 276, 341
Index
377
PSYMBNAM data element 291, 341
PTSOUSER data element 296, 341
publications
OMEGAMON XE on z/OS 363
OPAL 365
Redbooks 366
Technotes 366
types 363
PWAITPCT data element 291, 294, 295, 296, 341
PWAITTIME data element 294
PWAITTME data element 291, 295, 296, 341
PWKLNAME data element 291, 294, 295, 296, 342
Q
QIO keyword 59, 147
QSTEPS 170
quartile values 174
QUEUEDIO keyword 59, 147
QUIT 19
quotes 24, 25
R
R navigational command 127, 129
RALL keyword 63, 233
RANGE keyword 91, 92, 172
RCCH data set 235
RCCH keyword 63, 64, 233
RCHN data set 239
RCHN keyword 65, 233
RCL keyword 59, 147
RCPU data set 241
RCPU keyword 66, 233
RCR keyword 59, 147
RCV keyword 59, 147
RDAS data set 247
RDAS display examples 109
RDAS keyword 68, 134, 233
RDHITCFW data element 237, 342
RDHITNM1 data element 236, 342
RDHITNM2 data element 236, 342
RDHITSQ1 data element 236, 343
RDHITSQ2 data element 236, 343
RDHITTOT data element 236, 343
RDOM data set 250
RDOM keyword 73, 233
RDRTIME data element 285, 343
RDWRTRT data element 237, 344
RDY keyword 59, 147
READ keyword 139
READCTL keyword 59, 147
READY keyword 59, 147
READYIN data element 242, 344
READYOUT data element 242, 344
REAL data element 260, 344
REALSTOR keyword 59, 147
RECALL keyword 59, 147
RECOVER keyword 59, 147
Redbooks 366
relative date and time 93
378
relative dates and times 173
REOPEN command 20
REPLACE in OBTAIN command 195
REPLACEM command 163
replaying graphs 212
report element tables 233
reporter 19, 20, 23, 24
batch operation 19
command syntax rules 24
full-screen operation 19
list of commands 23
opening the datastores 20
overview 19
physical I/Os 20
split-screen operation 19, 20
VSAM buffers 20
reporter batch operation 142
BLOCK keyword 142
DUPLICATE keyword 142
TITLE keywords 142
reporter commands 19, 20, 23, 25, 26, 95, 111, 114,
130, 142, 162, 163, 165
CREATEM 162
DELETEM 163
DISPLAY 26
END 19
INQUIRE 111
LISTM 163
PAGESEP 142
PRODUCTS 114
QUIT 19
REOPEN 20
REPLACEM 163
RESOURCE 130, 165
SET 95
STOP 19
TSO 25
reporter keywords date and time 91
REPORTIF keyword 95, 143, 151, 181, 187, 230
REPORTIF Keyword 155
reporting with SAS graphics 201
representative range 174
REPTCNTL data element 265, 344
REQ keyword 59, 147
REQUEST keyword 59, 147
REQUEUE keyword 59, 147
RESOURCE command 130, 165
resource data 63
resource exception filters 155
resource keywords 63
RESP data element 265, 345
RESPONSE keyword 143, 145
RESRVEIO keyword 59, 147
reverse displays 105
RIF
See REPORTIF keyword
RIF keyword 143
RINF data set 252
RINF keyword 74, 233
RIO keyword 59, 147
RLCU data set 255
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
RLCU keyword 76, 234
RMF data element 252, 345
ROLL mode 102
RPAG data set 259
RPAG keyword 78, 234
RPDS data set 263
RPDS keyword 81, 234
RPGN data set 264
RPGN keyword 82, 234
RQU keyword 59, 147
RSDS data set 267
RSDS keyword 84, 234
RSRM data set 269
RSRM keyword 84, 234
RST keyword 59, 147
RSWA data set 270
RSWA keyword 85, 234
RSWR data set 273
RSWR keyword 87, 234
RTO keyword 59, 147
RVLF data set 278
RVLF keyword 89, 234
S
S navigational command 126, 129
sample EPILOG datastore 27
Sample Reports workload profiling 184
SAS data sets 235, 239, 241, 247, 250, 252, 255,
259, 263, 264, 267, 269, 270, 273, 278, 280, 284,
286, 288, 290, 293, 294, 296
BTCHCOMP 293
BTCHDETL 284
PGNCOMP 290
PGNDETL 280
RCCH 235
RCHN 239
RCPU 241
RDAS 247
RDOM 250
RINF 252
RLCU 255
RPAG 259
RPDS 263
RPGN 264
RSDS 267
RSRM 269
RSWA 270
RSWR 273
RVLF 278
STCCOMP 294
STCDETL 286
TSOCOMP 296
TSODETL 288
SAS file allocation 210
SAS interface 201
SATURDAY operand 92, 172
SAVE keyword 139
screen display functions 101
scrolling 102, 103
COMPLETE keyword 103
scrolling (continued)
mode 102
SDATE
See STARTDATE keyword
SDATE data element 345
SDATE keyword 92, 172
SDATE variable 241, 246, 249, 251, 255, 259, 262,
264, 266, 268, 270, 273, 278, 282, 285, 287, 289
SDATTIME data element 241, 246, 249, 251, 255,
259, 262, 264, 266, 268, 270, 273, 278, 282, 285,
291, 294, 295, 296, 345
SDSFLAG data element 267, 345
SEC override 216
SELECTIF keyword 95, 143, 151, 170
selecting data 91
service level management 115
SET command 95
SET command with OBTAIN command 193
SETP command 24, 189, 231
SETP Command 190
examples 190
SETP keywords 190
CLEAR 190
SYSID 190
time/date 190
workloads 190
shared-DASD 17, 44, 108, 134
contention 17
data 108
display 134
reporting 44
short form 25
SHORT keyword 96
SIF
See SELECTIF keyword
SINGLE keyword 97
single quotes 24
SMF 5
SMF ID 27, 170, 181, 228
SMF System ID
See SYSID
SMFID data element 241, 247, 250, 252, 255, 259,
262, 264, 267, 268, 270, 273, 278, 282, 285, 291,
294, 295, 297, 346
Software Support 367
SPCTYPE data element 263, 346
split-screen operation 20
SPS keyword 60, 148
SRASMQ data element 269, 346
SRBPCT data element 265, 346
SRCPU data element 269, 346
SRDMDPAG data element 269, 347
SRFIXB16 data element 269, 347
SRFIXTOT data element 269, 347
SRM 73, 82, 84
domain resource data 73
MPL adjustment values 84
parameters 84
performance group resource data 82
SRMRTO keyword 59, 147
SRMVALS data element 269, 347
Index
379
SRPAGDLY data element 269, 348
SRPAGFLT data element 269, 348
SRUIC data element 269, 348
SRVTIME data element 247, 348
STAGING keyword 59, 147
STARTDATE keyword 91, 92, 172
started tasks 35, 41
selecting for display 35
step displays 41
starter kit 209, 214, 215, 216, 217, 218, 219, 220,
221, 222, 223, 225, 226
description 215
macros 214
reports 216, 217, 218, 219, 220, 221, 222, 223,
225, 226
Batch Job Comparison 221
Batch Job Trending 222
Batch Program Resource 226
Channel I/O Rate 216
Channel Peak I/O Rate 217
Channel Peak Utilization 217
Channel Service Rate 223
Channel Utilization 216
CPU Utilization 219
DASD Utilization 225
Demand Paging Activity 218
Performance Group Degradation 220
Performance Group Trending 221
Real Storage Utilization 219
Swap Paging Activity 218
user overrides 215
STARTTIME keyword 91, 92, 172
STARTTSK keyword 37, 171
STC keyword 37, 171, 280, 290
STCAVGAS data element 242, 348
STCCOMP data set 294
STCDETL data set 286
STCMAXAS data element 242, 349
STCMINAS data element 242, 349
STCNAME data element 295, 349
STEP keyword 37
STEPNAME data element 285, 349
STGDIR1 data element 236, 349
STGDIR2 data element 236, 350
STI keyword 59, 147
STIME
See STARTTIME keyword
STIME data element 350
STIME keyword 92, 172
STIME variable 241, 247, 250, 252, 255, 259, 262,
264, 267, 268, 270, 273, 278, 282, 285, 287, 289
STIMER keyword 59, 147
STOP 19
storage resource data 78
STORNAME data element 260, 350
SUBSYSID data element 236, 350
SUM keyword 46
SUMDEF keyword 54, 96, 140
SUMMARY keyword 46, 188
summary wait categories 51, 140
summary wait definitions 51, 140
380
SUMWAIT keyword 35, 46, 96, 188
on XPG displays 35
SUNDAY operand 92, 172
SUPERSEC data element 265, 351
support assistant 367
swap activity resource data 85
swap data set resource data 84
swap reason resource data 87
SWAPCTG data element 276
SWAPERTX data element 265, 351
SWAPIN data element 261, 351
SWAPIN keyword 60, 147
SWAPOUT data element 261, 351
swapping 78
SWAPS data element 261, 351
SWI keyword 60, 147
SWPCATG data element 274, 275, 352
SYMBNAME data element 282, 292, 352
SYMBOLIC keyword 33, 171, 280, 290
symbolic names 33
SYSDSN operand 151
SYSID 230
SYSID keyword 27, 96, 107, 170, 181, 187, 190, 228,
230, 231
SYSIEA01 operand 151
SYSIEFSD operand 151
SYSIEWLP operand 151
SYSIGGV1 operand 151
SYSIGGV2 operand 151
SYSIKJBC operand 151
SYSLEVEL data element 252, 352
SYSOUT keyword 60, 147
SYSPAGE keyword 60, 148
SYSSMF01 operand 151
SYSTEM keyword 30, 169, 171, 181, 280, 290
system resource data 74
SYSVSAM operand 151
SYSVTOC operand 151
SYSZJES2 operand 151
SYSZVARY operand 151
T
T operand 150
TAPE operand 150
TAPEMOUNT keyword 60, 147
TARGMPL data element 250, 353
TCBPCT data element 265, 353
TDAY keyword 92, 172
Technotes 366
TERMIN keyword 60, 148
terminal input waits 60
TERMINAL override 216
TERMOUT keyword 60, 148
TEST keyword 170, 176
THISMONTH keyword 91, 92, 172
THISWEEK keyword 91, 92, 172
THISYEAR keyword 91, 92, 172
THRESHLD override 216
THURSDAY operand 92, 172
time keywords 173
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
time keywords (continued)
format 173
time keywords format 93
time period reporter 91
TIN keyword 60, 148
title block 103
TITLE keyword 96, 142
TITLE2 keyword 96, 142
TITLE3 keyword 96, 142
TMN keyword 92, 172
TMP keyword 60, 147
TODAY keyword 91, 92, 172
tolerance limits 110
TOT keyword 46
TOTAL keyword 46, 54
TOTCPUTM data element 282, 353
TOTIOTME data element 247, 353
TOTSTGK data element 265, 353
TOU keyword 60, 148
TRANSWAP keyword 60, 148
TRAP override 216
TRNSTIME data element 263, 267, 354
TRXELAP data element 282, 285, 294, 295, 297, 354
TRXRESP data element 282, 292, 354
TRXTHINK data element 282, 354
TSO keyword 37, 280, 290
TSO performance 14
TSO users selecting for display 35
TSOAVGAS data element 242, 355
TSOCOMP data set 296
TSODETL data set 288
TSOMAXAS data element 242, 355
TSOMINAS data element 242, 355
TSOUSER data element 297, 355
TSOUSER keyword 37, 171
TSW keyword 60, 148
TSWPSEC data element 271, 277, 355
TUESDAY operand 92, 172
tuning methodology 4
TWAITOF keyword 46
TWAITOFF keyword 60, 96
TWAITON keyword 46, 60, 96
TWK keyword 92, 172
TYR keyword 92, 172
U
U operand 150
underscored displays 105
UNI keyword 60, 148
UNILATERAL keyword 60, 148
UNIT operand 150
uppercase output 104
USE operand 107
user overrides 211
using the reporter 91
V
V navigational command
V operand 150
131
variable characters 25
version number 114
VIRT data element 260, 358
VLF resource data 89
VLFADDS data element 279, 356
VLFCLASS data element 356
VLFDELES data element 279, 356
VLFFULL data element 279, 357
VLFHITS data element 279, 356
VLFLRGST data element 279, 357
VLFMXSTG data element 278, 357
VLFRATE data element 279, 357
VLFTRIMS data element 279, 357
VLFVSUSD data element 278, 358
VOLSER data element 236, 247, 263, 267, 282, 285,
292, 294, 295, 297, 358
VOLUME operand 150
VSAM buffers 20
W
W navigational command 127, 129
wait reason 47, 49, 51
detail 49
graphic display 47
summary 51
wait reasons 144, 146, 148
detail 146
NO prefix 144
summary 148
WAITNAME data element 282, 285, 292, 294, 295,
297, 358
WAITPCT data element 282, 285, 292, 294, 295, 297,
359
WAITTIME data element 282, 285, 292, 294, 295,
297, 359
WEDNESDAY operand 92, 172
WEEKDAY operand 92, 172
WEEKEND operand 92, 172
wildcard characters 25
WKDAY operand 92, 172
WKEND operand 92, 172
WKLNAME data element 283, 285, 292, 294, 295,
297, 359
workload degradation 29, 30, 31, 35, 46
batch jobs 35
display options 46
display screen 29
performance groups 31
started tasks 35
systemwide 30
TSO user sessions 35
workload displays 39, 41
job level 39
step level 41
workload exception filters 143
workload keywords 169, 171, 181, 187, 190, 229, 231
workload profiling 167, 168, 169, 179, 183, 189, 231
comparing workloads to profiles 183
creating profiles 168
displaying profiles 179
Index
381
workload profiling (continued)
overview 167
SETP 189, 231
Workload Profiling Facility 4, 5, 7
overview 4
workload utilization 44
workload-profile comparison data 290
workload-profile comparisons 229
WPF
See also workload profiling
See Workload Profiling Facility
WRHITCFW data element 237, 360
WRHITDFN data element 237, 360
WRHITDFS data element 237, 360
WRHITDFT data element 239, 360
WRHITNM1 data element 237, 361
WRHITNM2 data element 237, 361
WRHITSEQ data element 237, 361
WRHITTOT data element 237, 361
WRITENM1 data element 239, 361
WRITENM2 data element 239, 362
WRITESQ1 data element 239, 362
WRITESQ2 data element 239, 362
WTO keyword 60, 148
WTORWAIT keyword 60, 148
X
X navigational command 129
X navigational command XPG display
navigational facility 130
XPG keyword 33
130
Y
YDAY keyword 92, 172
YESTERDAY keyword 91, 92, 172
382
IBM Tivoli OMEGAMON XE on z/OS: EPILOG for MVS User's Guide
Printed in USA
SC27-4034-00
Download PDF

advertising