Chapter 5. Specifying IMS execution parameters. IBM Version 11 System Definition

PDF

文档

Chapter 5. Specifying IMS execution parameters

You can specify control region execution parameters in the appropriate control region JCL, or in the IMS, DBC, and DCC members of the IMS PROCLIB data set.

Parameters specified in these members override any system definition. The parameters are not positional.

Use the PARM1= and PARM2= parameters on the EXEC statements for the control region to specify the JCL for IMS online execution.

Recommendation:

For optimal performance, use the PARM1= and PARM2= parameters rather than creating your own JCL. These parameters are specified symbolically for the control region.

System control and performance EXEC parameters for the IMS control region

Using IMS EXEC parameters, you can control many IMS system resources such as number of active regions, performance options, z/OS options, DLISAS options, dump formatting options, IRLM options, Fast Path settings, subsystem options, and storage pool definitions.

This topic describes the EXEC parameters that can be used to control IMS system resources.

Identifying the nucleus

The SUF parameter is generated as a null value for the IMS procedure, indicating that the default nucleus name has a one-character suffix of 0. If you are using an alternative control program, you must specify this parameter.

Overriding the number of active regions

Use the PST parameter to override the expected number of regions that are to be in operation during the online execution. Additional regions can be dynamically allocated, up to the maximum allowable number permitted by your operating system. If you do not specify the PST parameter, the default number of regions is the number specified for the MAXREGN keyword in the IMSCTRL macro.

Specifying performance options

Three parameters contribute to general performance strategy. The FIX= parameter specifies a two-character suffix for the member DFSFIXxx in the IMS PROCLIB data set. This suffix indicates which member should describe all modules and control blocks that are to be page fixed. The EXVR= parameter allows you to page-fix buffers used for the management of message queues. The parameter is generated with a null value; you must specify EXVR=1 to indicate the page-fixing action. Similarly, the PRLD= parameter specifies a two-character suffix for the member DFSMPLxx, where all preloaded modules are listed.

145

146

System Definition

Specifying z/OS options

Several parameters apply only to the z/OS operating system. The SRCH parameter allows you to take advantage of any special library structure to optimize the search for loaded modules. You can override the default value of 0 with a value of 1 if you want the JPA and LPA to be searched before IMS program libraries.

To reduce the amount of z/OS CSA (common storage area) that IMS uses you can use one of the following techniques:

One technique is to specify LSO=Y on the IMS procedure to cause some control blocks and some IMS modules that are used for DL/I processing to be loaded into the control program's private storage. To allow for this transfer, use the following sum to increase the size of the control region:

220K + OSAM buffer pool + VSAM buffers

+ enqueue/dequeue tables + system log buffers

Another way to reduce IMS use of CSA is to take advantage of log buffer CSA

VSCR (virtual storage constraint relief). Using CSA VSCR, several buffers can be moved out of CSA and placed above the 16-MB line into ECSA (extended CSA).

The buffers that are relocated are the online log buffers, the batch log buffers, and the IMS Monitor buffers. This method not only frees CSA space, but it also allows these buffers more space in ECSA. To use this buffer relocation technique, specify

LSO=S on the IMS procedure.

Specifying DL/I separate address space options

An additional variation of the local storage option is to use the DL/I separate address space. You do this by specifying LSO=S. This address space contains most of the DL/I code, control blocks, and database buffers for full-function databases.

Again, z/OS cross memory services are used.

Specifying dump formatting options

In a DC environment, you can request the following types of dump outputs for errors that terminate IMS: SDUMP, SYSMDUMP, SYSABEND, or SYSUDUMP. To do this, specify the FMTO startup parameter in combination with z/OS dump DD statements.

For SYSMDUMP, you should provide operational procedures for saving and formatting dumps; otherwise, you can overlay a SYSMDUMP if you must restart

IMS before the previous SYSMDUMP is transferred.

You can also request dump outputs for some errors that do not terminate IMS.

Your choice of dump depends on several factors: the type of failure, the FMTO parameter option, and the IMS spin-off and z/OS dump DD statements that have been selected. For more information about the FMTO parameter and about using dumps in these situations, see

IMS Version 11 System Utilities

Specifying IRLM options

If an execution of the IMS online system is to use IRLM as lock manager or participate in block-level sharing, specify IRLM=YES. Specify the z/OS subsystem name for the IRLM associated with this control region on the IRLMNM parameter.

You cannot use the UHASH parameter with the IRLM to specify the name of an alternative Fast Path hashing module. The UHASH parameter is ignored, and

DBFLHSH0 is always used as the hashing module.

Enabling Fast Path

Use FP= to enable or disable Fast Path.

The FPCTRL macro, although allowed for compatibility, is ignored by system definition. Fast Path is enabled by setting FP=Y in the DFSPBxxx member of the

IMS PROCLIB data set or by specifying FP=Y as a startup parameter. The default is FP=N, so you must explicitly specify FP=Y to enable Fast Path.

If you specify FP=N and you attempt to use a Fast Path resource or command, results are unpredictable.

To override the initial value coded on DFSPBxxx, use the FP= parameter at execution time.

Specifying subsystem identification parameters

Use the SSM parameter to reference a member in the IMS PROCLIB data set which identifies the DB2

® for z/OS subsystems that can be accessed from application programs executing in dependent regions. The one- to four-character suffix that you specify, together with the currently assigned name for IMSID, forms the member name. The member contains entries, each identifying a DB2 for z/OS subsystem (by its z/OS subsystem name). All subsystems to be accessed from programs executing in dependent regions must be identified by an entry.

Depending on its entry, a region can access all, some, or none of these subsystems.

To allow a dependent region to access all subsystems identified to the IMS control region, do not specify the SSM parameter for the dependent region or have the

SSM entry specify the same member as the IMS procedure. To allow a dependent region to access only selected DB2 for z/OS subsystems, specify an SSM parameter on the dependent region procedure that points to a member containing only those specific subsystems. To prevent a dependent region from having access to any subsystem identified to the IMS control region, specify a member containing no entries.

See

IMS Version 11 System Administration

for details about how to specify the individual entries and construct the member of the IMS PROCLIB data set.

Modifying storage pool definitions for the storage manager

You can set an upper expansion limit for the AOIP, HIOP, CIOP, CMDP, SPAP,

DYNP, LUMP, LUMC, FPWP, and EMHB storage manager pools by using the appropriate parameters at execution time. IMS establishes these storage pool definitions without an upper expansion limit, because they are dynamic storage pools that expand and contract as needed during execution.

Use caution in specifying upper expansion limits. If an upper limit is too low, IMS might abend. Under normal circumstances, a pool should never reach its upper limit. The intent of the upper limit is to keep pools from consuming so much storage that an out-of-storage condition occurs.

Chapter 5. Specifying IMS execution parameters

147

Use the SPM=nn parameter to specify the suffix for DFSPMnn. DFSPMnn identifies the member of the IMS PROCLIB data set that overrides the storage pool definitions established by IMS.

Dynamic resource definition EXEC parameters

You can use the DFSDF= EXEC parameter to override dynamic resource definition

(DRD) values that were specified on the DFSDFxxx member of the IMS PROCLIB data set.

Use DFSDF= to identify the DRD PROCLIB member.

By using DFSDF= as an EXEC parameter in the control region, you can override the suffix for DFSDFxxx that was provided during system definition.

Database and PSB exec parameters for the control region

The IMS procedure includes parameters that enable you to control database buffer requirements, override predefined buffer sizes, and manage performance.

Three kinds of parameters in the IMS procedure that are related to database and

PSB processing are: v Pointers to database buffer requirements v

Overrides of predefined buffer sizes v

Performance options

Database buffer requirements

You can use the VSPEC parameter of the IMS procedure to predefine buffer pool requirements for databases that use the OSAM or VSAM access method.

Use the VSPEC parameter to specify a two-character suffix for member DFSVSMxx in the IMS PROCLIB data set. This member predefines the buffer pool requirements for databases that use OSAM or VSAM as the access method. The

IMS procedure generates a default value of 00 that points to a general- purpose buffer definition in member, but you can change this suffix.

See “IMS buffer pools” on page 177 to specify subpool sizes.

Buffer sizes for databases that use OSAM or VSAM

Performance studies often indicate a need to increase DMB or PSB buffer space.

You can override the system definition values to implement this change.

Several parameters enable you to redefine the size of buffer pools that hold DMBs and PSBs. Specify a value that represents the size as a number of 1024-byte blocks.

For example, a value of 18 represents the size as 18432 bytes (18 x 1024 = 18432).

IMS then rounds up this number based on the nearest whole number of pages on which these bytes fit. For example, if one page in z/OS equals 4096 bytes, the value of 18 blocks would result in the allocation of 20480 bytes. This value overrides the value defined in system definition. The calculation is as follows:

Formula

18 x 24 = 18432/4096 = 4.5 pages (round up to 5)

5 x 4096 = 20480 bytes

148

System Definition

If you use the DL/I address space option, two PSB pools exist: one in the z/OS common area and one in the DL/I address space. Use the SASPSB parameter on the BUFPOOLS macro to specify sizes of the PSB pools during IMS system definition; use the CSAPSB and DLIPSB parameters on the IMS procedure to override the size of the PSSB pools. If the DL/I address space is not used, specify the size of the single PSB pool using the PSB= parameter on the BUFPOOLS macro; override this value with the PSB= parameter on the IMS procedure.

If Fast Path resources are used, an EPCB pool must contain extensions to Fast Path

PCBs. For each PCB referencing a DEDB or MSDB database, an EPCB is required.

This pool is used only by MPP regions, not by IFP or BMP regions.

Use the parameters of the IMS procedure in Table 25 to override the buffer pool

sizes that were specified at system definition time with the BUFPOOLS macro.

Table 25. Overriding buffer pool sizes with database and PSB parameters

IMS Procedure

Parameter

DMB

PSB

PSBW

CSAPSB

DLIPSB

EPCB

System Definition Specification

Macro

BUFPOOLS

Parameter

DMB

PSB

PSBW

SASPSB(size1)

SASPSB(size2)

EPCB

The IMS procedure is generated with null value parameters for these storage areas.

The sizes default to their system definition values.

Performance options for data management and program specification blocks

You can override the system definition specifications for individual database DMBs and PSBs using the RES parameter on the IMS procedure.

System definition lets you specify individual database DMBs and PSBs to be resident with the DATABASE and APPLCTN macros, respectively. You can override this specification with the RES parameter. The IMS procedure generates this as a null parameter; the default is RES=Y.

Request that all the control blocks are not to be made resident by specifying

RES=N. You might do this when executing a test system or if storage is temporarily constrained.

Data communication EXEC parameters for the control region

Using EXEC parameters, you can override definitions for data communication that were initially set during IMS system definition.

Three kinds of parameters in the IMS procedure are related to data communication: v Overrides of predefined buffer sizes v

Overrides and options particular to Message Format Service (MFS)

Chapter 5. Specifying IMS execution parameters

149

150

System Definition v

Performance options

Buffer sizes for data communications

By monitoring communication traffic you might decide to increase buffer space.

Use the parameters of the IMS procedure shown below to allocate virtual storage and override the buffer pool sizes predefined in system definition.

Table 26. Overriding buffer pool sizes with data communication parameters

IMS procedure parameter

FBP

QBUF

System definition specification

Macro

BUFPOOLS

MSGQUEUE

Parameter

FORMAT(size1)

BUFFERS

RECA COMM RECANY

Specify a value representing the number of 1024-byte blocks.

By increasing the number of buffers (used for the message queue data sets), you can reduce the frequency of I/Os.

Overrides of predefined parameters

Monitoring communication traffic often results in a decision to increase buffer space or the number of buffers. The following execution parameters are provided to override the size and number of some of the buffer pools predefined at system definition:

FBP

Overrides the size specified on the FORMAT(size1) keyword of the

BUFPOOLS macro.

QBUF

Message queue buffer. Adjusts the size of the message queue buffer pool. QBUF enables you to override the number of blocks to hold messages in the control program's storage. By increasing the number of buffers used for the message queue data sets, you can reduce the frequency of I/Os. The default number of buffers is specified in the

MSGQUEUE macro.

EMHL

Specifies the EMH buffer length for Fast Path transactions.

RECASZ

Specifies the size of the Receive Any buffer.

RECA

Overrides the number of Receive Any buffers predefined in system definition by the RECANY keyword of the COMM macro.

SAV

Specifies the allowed maximum number of concurrently active device

I/Os.

ETO parameters

Use the following parameters to define ETO options:

ALOT

Specifies the amount of time allotted before a terminal in session is automatically logged off if no user successfully signs on.

ASOT

Specifies the amount of time allotted before a signed-on user is automatically signed off if no input or output activity occurs.

LHTS

Specifies the number of slots for the CNT/LNB/RCNT hash table.

NHTS

Specifies the number of slots for the VTCB hash table.

UHTS

Specifies the number of slots for the SPQB hash table.

DLQT

Specifies the number of days allotted before a queue containing data that has not been allocated is classified as a potential dead-letter queue.

ETO

Specifies whether terminals and queues that are not defined to IMS are to be supported.

DSCT

Specifies the suffix of the descriptor member DFSDSCTy.

Performance options

Use the following parameters to define performance options:

VAUT

Specifies the use of VTAM-Authorized Path.

NLXB

Specifies that parallel sessions are added during system startup.

FESTIM

Overrides the timeout value for Front End Switching; this value is predefined during system definition.

MFS options

When you use MFS, you can adjust the capacity of the system by adjusting the availability of format blocks. Use the FRE parameter to specify the maximum number of active blocks in the message format buffer pool. You need to coordinate this number with the value specified for the FBP pool size parameter.

Performance options for VTAM systems

The VAUT parameter offers a performance option to systems using VTAM. You can override the null value generated in the IMS procedure and specify the use of

VTAM Authorized Path.

The NLXB parameter allows you to add parallel sessions during system startup.

The number you specify is added to the number defined in the SESSION keyword of the MSPLINK macro.

The FESTIM parameter overrides the timeout value for front-end switching that is predefined in system definition. You specify the parameter on the COMM macro with a value from 1 second to 300 seconds.

Chapter 5. Specifying IMS execution parameters

151

Fast Path EXEC parameters in DCCTL or DB/DC

Use Fast Path EXEC parameters to specify database buffer sizes and specify DEDB options in a DCCTL or IMS DB/DC environment.

Note:

Although the DCCTL environment does not support Fast Path databases, it does support Fast Path processing and transactions.

The PARM1= and PARM2= positional parameters for the control region's EXEC statement that can be used to specify the Fast Path parameters are shown below.

Table 27. Categories and purpose of Fast Path control region parameters for DCCTL and

DB/DC

Category Parameter Purpose

MSDB load MSDB

Database buffer sizes BSIZ

DEDB options

DBBF

DBFX

OTHR

EPCB

EMHL

LGNR

EPCB

EMHL

Specify the DBFMSDBx suffix

Specify the common size of a buffer

Specify the maximum number of buffers

Specify system buffer allocation

Specify the number of DEDB updates that can be concurrently waiting for I/O after sync point

Specify the maximum DEDB buffer alterations before CI logging

Specify the size of the EPCB pool

Specify the size of the EMHL buffer

When an emergency restart is performed, the values specified for the EXEC parameters MSDB, BSIZ, DBBF, OTHR, and LGNR, and the contents of the member DBFMSDBx, must remain unchanged from the last normal start. These values are used when reestablishing buffer contents from checkpoint records.

MSDB loading

The MSDB parameter enables you to control which MSDBs must be loaded for the current online session and how many segments each MSDB requires. The parameter is generated with a null value, implying that the MSDBs are loaded directly from a Fast Path system data set. You specify a one-character suffix that points to an IMS.PROCLIB member, DBFMSDBx, containing the detailed requirements.

If the MSDB requirements for an online session are a subset of the MSDBs, or if the number of segments for any MSDB is to be increased to reserve space for dynamic terminal-related MSDBs, you must coordinate the content of the

DBFMSDBx member and the MSDB parameter value. The control statements in the

DBFMSDBx member must explicitly name all MSDBs required to be loaded. The control statements also enable you to individually select MSDBs for page fixing.

LU 6.2 devices allow read-only access to dynamic MSDBs. LU 6.2 devices cannot access terminal-related MSDBs.

Terminals defined dynamically with ETO are not available to terminal-related

MSDBs.

152

System Definition

Database buffer sizes

Using the DBBF and BSIZ parameters, you manually specify the total buffers available for the IMS online system's processing of DEDB activity. Dependent regions then claim portions of this allocation. Use DBFX to specify additional buffers (not separate pools) that are set aside and page-fixed at when the control region is initialized. These buffers allow for asynchronous processing.

If you prefer to have IMS automatically manage Fast Path buffer pools, enable the

Fast Path 64-bit buffer manager using the DFSDFxxx PROCLIB member.

DEDB options

The OTHR parameter specifies the number of asynchronous output threads. You can revise the estimate of DEDB updates that occupies buffer space while they are waiting to be committed to the area data sets after sync point logging.

The OTHR= parameter can have any value from 1 to 32,767. If you do not specify a value for this parameter, the value defaults to 255. If you specify 0 or a number greater than 32,767, the value defaults to 2.

The LGNR parameter determines a maximum number of individually logged

DEDB alterations made in a buffer. Above that number, the whole control interval

(CI) is logged. The default value is 7, and you can specify a number from 7 to 99.

If the application program is using large CIs, the LGNR value can be increased to economize on the system log activity.

Related reference

“FASTPATH section of the DFSDFxxx member” on page 687

“Parameter descriptions for IMS procedures” on page 448

Fast Path EXEC parameters in DBCTL

Use Fast Path EXEC parameters to specify database buffer sizes and specify DEDB options in a DBCTL environment.

The PARM1= and PARM2= positional parameters for the control region's EXEC statement that can be used to specify the Fast Path parameters in a DBCTL region are shown below.

Table 28. Categories and purpose of Fast Path control region parameters for DBCTL

Category Parameter

Database buffer sizes BSIZ

DBBF

DEDB options

DBFX

OTHR

EPCB

LGNR

EPCB

Purpose

Specify the common size of a buffer

Specify the maximum number of buffers

Specify system buffer allocation

Specify the number of DEDB updates that can be concurrently waiting for I/O after sync point

Specify the maximum DEDB buffer alterations before CI logging

Specify the size of the EPCB pool

Chapter 5. Specifying IMS execution parameters

153

When an emergency restart is performed, the values specified for the EXEC parameters BSIZ, DBBF, OTHR, and LGNR must remain unchanged from the last normal start. These values are used when reestablishing buffer contents from checkpoint records.

Database buffer sizes

If you prefer to have IMS automatically manage Fast Path buffer pools, enable the

Fast Path 64-bit buffer manager using the DFSDFxxx PROCLIB member.

DEDB options

The OTHR parameter specifies the number of asynchronous output threads. It allows you to revise the estimate of DEDB updates occupying buffer space while waiting to be committed to the area data sets after sync point logging.

The LGNR parameter determines a maximum for the number of DEDB alterations made in a buffer that are individually logged. Above that number, the whole control interval (CI) is logged. The default value is 7, and you can specify a number from 7 to 99. If the application program is using large CIs, the LGNR value can be increased to economize on the system log activity.

Related reference

“Parameter descriptions for IMS procedures” on page 448

“FASTPATH section of the DFSDFxxx member” on page 687

Fast Path dependent region parameters in DCCTL or DB/DC

The positional parameters (PARM=) for a Fast Path dependent region in DCCTL or

DB/DC environments are described in this topic.

Note:

The DCCTL environment does not support Fast Path databases. It does support Fast Path processing and transactions.

The PARM= positional parameters for Fast Path dependent region's EXEC

statements are shown in Table 29.

Table 29. Fast Path dependent region parameters for DCCTL and DB/DC

Category

Database and PSB

Parameter

Data communications

NBA

OBA

MBR

PSB none

154

System Definition

Table 29. Fast Path dependent region parameters for DCCTL and DB/DC (continued)

Category Parameter

Region control and performance

IFP

OPT

PRLD

PREINIT

STIMER

DIRCA

CPUTIME

DBLDL

SSM

ALTID

PARDLI

Recovery and restart

TLIM

SOD

Security options

IMSID

The EXEC statement's first positional parameter, IFP, causes this region to be active for Fast Path processing only. The MBR parameter specifies the name of the message-driven application program.

Two database-related parameters, NBA and OBA, control how many of the total

Fast Path buffers this region can appropriate for its use. If you are using the Fast

Path 64-bit buffer manager, the combined values of NBA and OBA make up the maximum number of buffers reserved for storage. If you are not using the Fast

Path 64-bit buffer manager, DBBF controls the maximum number of buffers reserved for storage.

The NBA parameter specifies how many buffers are reserved for DEDB and MSDB processing by this region. This normal allotment is obtained at region startup and must be available from the total number specified for the control region. Specify a number from 1 to 9 999. If you are not using the Fast Path 64-bit buffer manager, coordinate the value across all regions that are to be concurrently active and specify the total on the DBBF parameter. If too few buffers are available from the

DBBF total, the dependent region abends. If you are using the Fast Path 64-bit buffer manager, IMS controls the expansion of the buffer pools, and the DBBF parameter is not valid.

The OBA parameter specifies how many buffers are requested from the control region as overflow when the normal allotment is used up. The control region page fixes the largest number of overflow buffers specified for all active BMPs and

CCTL threads. If you are not using the Fast Path 64-bit buffer manager, the buffers are made available to only one program at a time. If you are using the Fast Path

64-bit buffer manager, multiple dependent regions can extend into the overflow buffers concurrently.

Both the NBA and OBA parameters are generated with a 0 (zero) value so that you must respecify the individual values.

Chapter 5. Specifying IMS execution parameters

155

|||

Fast Path parameters in BMP and CCTL regions in DBCTL

The Fast Path positional parameters (PARM1= and PARM2=) for batch message processing and CCTL regions in a DBCTL environment are described in this topic.

The PARM1= and PARM2= positional parameters for a BMP region's EXEC statements are shown below. The Fast Path parameters in the DRA startup table that the CCTL must use when it connects to a DBCTL environment are also shown.

Table 30. Fast Path dependent region parameters for DBCTL

Category Parameter

Database and PSB NBA

OBA

MBR

CCTL

PSB

CNBA

FPB

FPOB

Purpose

Specifies the number of database buffers

Specifies the number of overflow buffers

Specifies name of message-driven program

Specifies PSB name

Total number of buffers for this CCTL

Number of database buffers each thread uses (from the

CNBA total)

Number of overflow buffers each thread might need

Two database-related parameters, NBA and OBA, control how many of the total

Fast Path buffers this region can appropriate for its use.

The NBA or CNBA parameter specifies how many buffers are reserved for DEDB processing. This normal allotment is obtained at region startup and must be available from the total number specified for the control region. Specify a number from 1 to 9999. If you are not using the Fast Path 64-bit buffer manager, coordinate the value across all regions that are to be concurrently active and specify the total on the DBBF parameter. If too few buffers are available from the DBBF total, the dependent region abends. If you are using the Fast Path 64-bit buffer manager, IMS controls the expansion of the buffer pools, and the DBBF parameter is not valid.

After a CCTL is connected, each CCTL thread request for Fast Path PSBs receives an allotment of buffers.

The OBA or FPOB parameter specifies how many buffers are requested from the control region as overflow when the normal allotment (NBA, FPB) is depleted. The control region page fixes the largest number of overflow buffers specified for all active BMPs and CCTL threads. If you are not using the Fast Path 64-bit buffer manager, the buffers are made available to only one program at a time. If you are using the Fast Path 64-bit buffer manager, multiple dependent regions can extend into the overflow buffers concurrently. Choose a value from 1 to 9999 that is a suitable common value for all regions.

Both the NBA and OBA parameters are generated with a 0 (zero) value so that you must respecify the individual values.

Related reading

: For more information about DEDB buffer considerations in a

DBCTL environment, see

IMS Version 11 Database Administration

156

System Definition

Online DEDB utility region parameters in DCCTL, DBCTL, or DB/DC

There are several positional parameters used by the Fast Path Online DEDB utility in DCCTL, DBCTL, and DB/DC environments.

The procedure FPUTIL uses the first few positional parameters defined for the

IMSFP procedure, but interprets several of the parameters as control for the online

DEDB utilities. Table 31 shows these parameters.

Table 31. Generated values and purposes for FPUTIL procedure parameters

FPUTIL Parameter

IFP

DBD

N/A

REST

N/A

DIRCA

PRLD

N/A

Generated Value

IFP

N/A

DBF#FPU0

Null

Purpose

Specifies Fast Path region

Specifies the DEDB name

Utility program name

Restart indicator

No overflow buffers

Default startup, ask operator

Allows one abend

2 KB block for PCB

No preload

No time limit

The function performed by the utility depends on the input control statements. A series of DEDB areas can be scanned, have dependent segments deleted, or have the units of work reorganized. If the utility is to be restarted, the REST parameter is coded nonzero.

Recovery-related EXEC parameters for the control region

The EXEC parameters that you can use to control the kind of recovery that is performed for the current system execution include ARC=, DBRC=, DBRCNM=,

WADS=, QTU=, AND QTL=.

ARC=

Use the ARC parameter to specify whether automatic archiving of the online log data sets (OLDS) is to be performed. Automatic archiving is recommended; however, you can arrange for the MTO to monitor the availability of the OLDS and perform archiving when necessary. Specify ARC=0 for no automatic archiving; otherwise, specify a one- or two-digit number representing the number of full

(closed) OLDSs that invoke automatic archiving for those data sets.

DBRC=

Batch and utility regions use the DFSIDEF0 module during initialization. The

DFSIDEF0 module that is shipped in the ADFSSMPL library contains a batch and utility region default of DBRC=YES. This value is coded on the DFSIDEF macro.

You are not required to use the DFSIDEF0 module. If you do not use the module, or if the module cannot be loaded at initialization time, IMS defaults to

Chapter 5. Specifying IMS execution parameters

157

158

System Definition

DBRC=YES. Therefore, if DBRC is to be used for your batch and utility regions, then creating the DFSIDEF0 module is not required.

For online environments (DBCTL, DCCTL, or DB/DC), the only valid value is

DBRC=YES.

For batch and utilities, you might want to change the default or current value for

DBRC.

For batch and utilities. the default is DBRC=Y. To change the default value for

DBRC, you can do either of the following: v

Assemble and bind DFSIDEF0 into the IMS execution library; you can specify

DBRC=Y, DBRC=N, or DBRC=FORCE. If you specify RMODE, you must specify

RMODE=24 and AMODE=24 v

Override the value in a DFSPBxxx PROCLIB member. You can specify only

DBRC=N or DBRC=Y.

To do this, you must define the RGSUF= parameter in your DLIBATCH or

DBBBATCH procedure.

To change the current value for DBRC, override the value in JCL. You can specify only DBRC=N or DBRC=Y.

Sample JCL to assemble and bind the DFSIDEF0 module is shown below:

//ASSEMBLE EXEC PGM=ASMA90,PARM=’NOOBJ,DECK’

//SYSLIB DD DSN=IMS.SDFSMAC,DISP=SHR

//SYSPUNCH DD DISP=OLD,DSN=IMS.OBJDSET(DFSIDEF0)

//SYSPRINT DD SYSOUT=*

//SYSUT1 DD UNIT=SYSDA,DISP=(,DELETE),SPACE=(CYL,(15,15))

//SYSIN

IDEF0

DD *

TITLE ’DFSIDEF0 - IMS INSTALLATION DEFAULTS BLOCK’

DFSIDEF0 CSECT

SPACE 1

DFSIDEF TYPE=BEGIN

DFSIDEF TYPE=PARM,DBRC=YES

*****

DFSIDEF TYPE=PARM,DBRC=NO

DFSIDEF TYPE=PARM,DBRC=FORCE

DFSIDEF TYPE=END

END DFSIDEF0

//STEP1 EXEC PGM=IEWL,

// PARM=’SIZE=(880K,64K),NCAL,LET,REUS,XREF,LIST’

//SYSPRINT DD SYSOUT=*

//SYSPUNCH DD DSN=IMS.OBJDSET,DISP=SHR

//SYSLMOD DD DSN=IMS.SDFSRESL,DISP=SHR

//SYSUT1 DD UNIT=(SYSDA,SEP=(SYSLMOD,SYSPUNCH)),SPACE=(CYL,(10,1))

//SYSLIN DD *

INCLUDE SYSPUNCH(DFSIDEF0)

NAME DFSIDEF0(R)

DBRCNM=

Specify a name (do not accept the default) for the start procedure for the DBRC region on the DBRCNM parameter. The default is a null parameter indicating a procedure name of DBRC. Also as part of your installation's recovery strategy, the

AUTO parameter specifies whether to use automatic restart of the IMS online system. You need to override the generated default value (N) by specifying

AUTO=Y if you plan to use automatic restart.

QTU= and QTL=

For a non-shared queues environment, the QTU and QTL parameters are associated with protection of the message queues. Use them to adjust the values in the message queue space notification exit routine (DFSQSPC0). Given the number of records reserved on each message queue data set for automatic shutdown

(specified in system definition with the SHUTDWN parameter on the MSGQUEUE macro), a finite number of residual records are available for message use. The notification exit routine keeps track of the current level of utilization. If the usage exceeds the upper limit (expressed as a percentage of records used), the MTO is notified so that steps can be taken to reduce the backlog. The MTO is also notified if the usage falls below the lower limit.

For a shared queues environment, the QTU and QTL parameters are used to monitor the device relative record number (DRRN) in-use count. If the high threshold value is reached, a DFS2281I message is issued, and all messages retrieved from shared queues are put in an IWAIT state. When the DRRN in-use count becomes less than the low threshold value, a DFS2282I message is issued, and the messages in an IWAIT state are then posted.

The QTU and QTL values are one or two-digit percentages. QTU overrides the upper threshold value and QTL overrides the lower one (100 is allowed for QTU).

Both upper and lower threshold values are generated with null values; the defaults are those of the IBM-supplied exit routine—75% and 60%, respectively. Unless your installation establishes its own exit routine to control the MTO notification, you should probably use these defaults. In the long term, you can tune them to fixed values based on the feedback from the statistics given in the

/DISPLAY POOL command; enter this command during peak periods of activity.

WADS=

The WADS parameter indicates whether the IMS online system is to use a single write-ahead data set (WADS) or dual data sets. The WADS helps protect the integrity of the online logging. In case of write errors, additional data sets can be defined up to a total of 10. Specify WADS=S for single recording or WADS=D for dual recording.

Security-related EXEC parameters for the control region

Several parameters control the kind of security checking that is done during the current execution by acting as switches for different types of security.

Moreover, these parameters determine how much flexibility the MTO has to override the choice of security checking. You must coordinate the setting of these switches with both overall security design and operational procedures. The parameters are RCLASS, SECCNT, TRN, SGN, RCF, ISIS, ASOT, ALOT, AOI1,

AOIS, and TCORACF.

All the values generated for the IMS procedure parameters specify no security. You need to reset them to match the security design. You must also coordinate the level of the security tables with the suffix identifier for the nucleus.

Related reading:

Operational restrictions for the MTO, and the parameter values and their meaning are discussed in

IMS Version 11 System Administration

Chapter 5. Specifying IMS execution parameters

159

The IMSID parameter is partially related to security and to operations. The one- to four-character value that you specify uniquely identifies the control region.

Dependent regions executing under control of this nucleus must specify the same identifier. The definition default is IMSA.

Any message sent to the console from subsystem execution is identified by this name; therefore, the name chosen should be unique to other subsystems executing in the z/OS system, including any batch executions. The IMSID name should not be the same as the procedure name in order to avoid confusion as to the source of the message.

EXEC parameters for IMS message processing regions

Using IMS EXEC parameters, you can control many IMS message processing region characteristics including data communications, PSBs, performance, recovery and restart, and security.

These EXEC parameters do not apply to DBCTL.

The PARM= positional parameters for a message processing region's EXEC

statement are shown in Table 32. See the DFSMPR procedure for more information.

Table 32. Categories and purpose of message processing region parameters

Category

Database and PSB

Data Communication

Region Control and

Performance

Parameter

PCB

VALCK

NBA

OBA

CL1

CL2-4

LOCKMAX

Purpose

Specify size of pool for PSB copy

Check validity of addresses in DL/I calls

Specify Fast Path database buffers

Specify Fast Path overflow buffers

Specify priority of message class

Specify up to three lower-priority message classes

Turn off locking limitations

MSG

OPT

OVLA

PRLD

PREINIT

STIMER

DBLDL

SOD

SSM

ALTID

Specify MSG for message region

Coordinate region start with control region

Retain overlay supervisor

Specify suffix of DFSMPLxx for program preload

Specify suffix of DFSINTxx for modules to be preinitialized in readiness for takeover

Specify timer options

Specify maximum number of

IMS.PGMLIB directory entries to be saved

Specify output class for spin-off dump

Specify access to DB2 for z/OS subsystems

Specify alternate IMS system

160

System Definition

Table 32. Categories and purpose of message processing region parameters (continued)

Category Parameter Purpose

Recovery and restart

Security options

PWFI

APARM

SPIE

TLIM

IMSID

Activate pseudo wait-for-input

Specify application parameter

Allow user program SPIE during DL/I call

Specify number of abnormal terminations allowed

Override IMS subsystem identifier

PSB-related EXEC parameters for a message processing region

The PCB parameter is an optional parameter that specifies the size of the interregion communication area. This area holds the PCBs included in the PSBs that are used by programs executing in this region. If you do not specify this parameter, the default value is 000. This allocates an area large enough to hold all the PCBs for the largest PSB defined to this IMS system and present in the active

IMS.ACBLIBA/B. To specify a value, enter the number of 1024-byte blocks as a three-digit number (for example, 001).

The default value for the VALCK parameter signifies that address validity checking is not to be performed for DL/I calls issued by the application programs in this region. (An address is invalid if it is either lower than the lowest address not in the z/OS nucleus or higher than the highest address in virtual storage.) With adequate testing of and controls over the DL/I call parameter coding, validity checking should not be necessary.

Data communication EXEC parameters for a message region

A message region requires the set of four parameters, CL1=, CL2=, CL3=, CL4=, to specify transaction classes. Express each class as a three-digit number. The first message class for the region causes all messages assigned to that class to be selected first as eligible for scheduling. Only when all possibility of scheduling a transaction in that class has been exhausted does scheduling begin for the second message class. Priorities determine the order that programs are to be selected for scheduling into the region. The message classes you specify need to be coordinated with your transaction scheduling rules and the numbers entered with the PRTY keyword on the TRANSACT macro.

Region control EXEC parameters for a message region

MSG is the first positional parameter for a message processing region. When transactions are encountered on the queue, programs are automatically scheduled in these regions, if the message class priority is suitable.

The parameter OPT= helps you control the region startup. If, for some reason, the control region is not active or is terminating when the MPP region is invoked, you can have the MTO decide whether to start the MPP region again (the default), let it wait until the control region is ready, or cancel it.

Use STIMER= to invoke timer facilities. If you use the non-default LSO option,

LSO=S, both application program and DL/I processing times are recorded. Most

DL/I processing time is not included when you use the default, LSO=Y. A value of

Chapter 5. Specifying IMS execution parameters

161

162

System Definition

1 causes a timer sequence to be issued once for each DL/I call that can adversely affect the performance of the region. This parameter is used with the TIME= parameter on the //JOB statement.

Recommendation:

Because a timeout of a message region causes an abnormal termination of the message region, have processing limits coded within the application program.

Use the SSM= parameter, which is required when using DB2 for z/OS group names to point to a member in IMS.PROCLIB that identifies the DB2 for z/OS subsystems that can be accessed from this MPP region. To prevent any access to subsystems from an MPP scheduled into this region, use the name of a null member (no entries). The SSM= default, a null value, allows the MPP region to attach to any of the subsystems declared to the IMS control region. If necessary, coordinate this parameter with the corresponding SSM parameter in the IMS

procedure. Refer to the description in “System control and performance EXEC parameters for the IMS control region” on page 145.

Use the APARM= parameter to specify execution time parameters unique to this dependent region. This parameter specifies a character string for the application program or Data Capture exit routine. The APARM= can be used to specify the frequency for checkpoint calls. The APARM= can also be used to pass parameters to the Data Capture exit routine to indicate whether the data should be captured.

The maximum length of the parameter is 32 characters; the parameter must be enclosed in single quotation marks (') if special characters are used; embedded commas are not allowed. The INQY call with the ENVIRON subfunction is used to receive the APARM character string.

Related reading:

For more information, see

IMS Version 11 Application Programming

Use the SOD parameter to specify the output class for a spin-off dump. In this way, you can obtain a printed copy of the storage immediately at abnormal termination of the message region immediately rather than being placed on general

SYSOUT queues.

Performance-related EXEC parameters for a message region

Four performance-related options are available. The parameter OVLA= enables you to retain a copy of the overlay supervisor in the message region. The default is to have the overlay control loaded each time you schedule an overlay application program in the region. If you have arranged the message priorities so that an overlay program is likely to be scheduled in the region, you should use this option.

Use the DBLDL= parameter to minimize the cost of program loading by increasing the number of program directory entries maintained in the message region. The effect is to reduce the I/O to the program library (and the search) to obtain the direct address of the program. The default is 20.

Use the PRLD= parameter to specify the two-character suffix of the member

DFSMPLxx in IMS.PROCLIB. This member contains the list of programs that are to be preloaded. The effect is to reduce the repetition of I/O for program load. The region address space retains the modules.

PREINIT allows you to specify the two-character suffix of the member DFSINTxx in IMS.PROCLIB. This member contains the preinitialization modules to are to receive control.

The type of program load selected depends on the order of the following:

Preload list

Standard z/OS fetch

The fourth performance-related parameter is the PWFI= (Pseudo WFI) parameter.

Specifying PWFI=Y can potentially reduce CPU time by eliminating the termination and rescheduling of resources. On a message GU call, if MODE=SNGL is specified on the TRANSACT macro and no more messages are available, IMS checks to see if other work needs to be done for the region. If no other work is available and the resources owned by the region are not needed by another region with work to do, the region becomes idle and waits until another message is queued (wait-for-input mode). It does not return a “QC” status code to the application program. If the new message is for the transaction that is scheduled in the region, IMS returns the new message to the application program without having to terminate and reschedule the resources.

If the next message that comes in is not for the transaction that is scheduled in the region, but it can be processed by the region, there is a delay before the new transaction can be processed. Because termination of the previous message is delayed until after the new message comes in, the region must go through termination before it can schedule the new transaction.

Recovery-related EXEC parameters for a message region

The TLIM parameter addresses a problem with an application program that causes an abnormal termination. Because the program can be scheduled many times into a region as a result of transactions in the queue, you need to be able to stop the operation of this region. The value for TLIM specifies the limiting number of abnormal terminations permitted. In the case where the application program has a

SPIE in effect, the SPIE option allows it to be left on during the DL/I call or for it to be turned off during the DL/I call and reinstated when returning to the application program. For performance reasons, it is not recommended to turn it on and off for each DL/I call. With PL/I Release 5 or later, you can use the PL/I SPIE facility without having IMS reset the SPIE on each DL/I call.

Security-related EXEC parameters for a message region

The parameter IMSID is related to security and to operations. You specify the 1- to four-character identifier for the name of the control region (the name given for the

IMSID parameter). If the value does not match the current IMSID of any operating control region, the message region is not scheduled.

EXEC parameters for IMS batch message processing regions

Using EXEC parameters, you can override definitions for IMS batch message processing regions that were initially set during IMS system definition, including definitions for databases, PSBs, data communications, region control and performance, recovery and restart, and security.

The PARM= positional parameters for the BMP region's EXEC statement are shown

in Table 33 on page 164. See the IMSBATCH procedure for more information.

Chapter 5. Specifying IMS execution parameters

163

164

System Definition

Table 33. Categories and purpose of BMP region parameters

Category Parameter Purpose

Database and PSB MBR

PSB

TEST

NBA

OBA

Data Communication IN

Region Control and

Performance

OUT

BMP

OPT

LOCKMAX

PRLD

Specify name of batch message program

Specify PSB name

Check validity of addresses in DL/I calls

Specify Fast Path database buffers

Specify Fast Path overflow buffers

Specify an input transaction queue

Specify an output transaction queue

Specify BMP for batch message region

Security options

PREINIT

STIMER

CPUTIME

PARDLI

DIRCA

SSM

ALTID

APARM

Recovery and Restart SPIE

CKPTID

IMSID

Coordinate region start with control region

Specify locking limitations

Specify suffix of DFSMPLxx for program preload

Specify suffix of DFSINTxx for modules to be preinitialized in readiness for takeover

Specify timer options

Specify processing time limit

Specify region for DL/I processing

Specify interregion communication size

Specify access to DB2 for z/OS subsystems

Specify alternate IMS system

Specify application parameter

Allow user program SPIE during DL/I call

Specify checkpoint for program restart

Override IMS subsystem identifier

PSB-related EXEC parameters for a batch message region

The MBR= parameter specifies the name of the program and is often the same as the PSB name. This parameter is required. The BMP has flexibility in using a program and PSB combination. This allows you to test modifications of the BMP using a temporary program name. You can also use a different PSB with the same program.

The PSB= parameter is optional if it matches the MBR name. If an APPLCTN macro has been included for the PSB specifying BATCH as the program type, the

PSB parameter specifies one of several PSBs for a given BMP. You should identify application programs that can process several types of transactions. These programs often require larger amounts of virtual storage, and you can need to adjust the size of the region.

The TEST= parameter is required, but the IMSBATCH procedure generates a 0 to specify that validity checking of the addresses in a DL/I call is not performed. (An address is invalid if it is either lower than the lowest address not in the z/OS nucleus or higher than the highest address in virtual storage.) Adequate testing of

| the program and controls over the DL/I call parameter list coding should make the generated option (no validity checking) appropriate.

The NBA= parameter specifies the number of Fast Path database buffers to be made available, up to the maximum number available. If you are using the Fast

Path 64-bit buffer manager, it manages the segmentation of the buffer pool into subpools. Otherwise, the buffers are taken from the global pool that is defined by the DBBF= parameter in the IMS control region procedure.

The OBA= parameter specifies the number of additional database buffers to be made available in the CSA/ECSA, up to the maximum number available. Fast Path allocates only the largest OBA specification from all currently active regions. If you are using the Fast Path 64-bit buffer manager, be aware that multiple regions can access the overflow buffers concurrently.

Data communication EXEC parameters for a batch message region

For the batch message program, you have the flexibility of declaring that the input transaction queue is to be made available to the program at execution time. You do this by specifying one transaction code as the value for the IN parameter.

Some BMP programs do not access a message queue; however, they have a requirement to send output to a terminal or to generate transactions to be processed by other application programs. Specify the LTERM name or transaction code (as appropriate) on the OUT parameter.

A null value for the IN parameter prevents the program from accessing message queues. When you specify the transaction code for the IN parameter, the program has no restrictions on generated transactions or output messages.

Region control EXEC parameters for a batch message region

BMP= is the first positional parameter for a batch message processing region.

These regions are not scheduled automatically. They must be invoked by the operator.

The OPT= parameter helps you control the start of a batch message region. If the control region is not active when the BMP region is invoked, you must decide whether to wait for a control program, cancel the batch message region, or ask the operator to decide. This might occur when JCL in the z/OS job stream for the BMP region is scheduled before the control region has completed its initialization or when the control region is terminating. Specifying 'wait' is a risk because the z/OS resource is reserved until the control region resumes. If you are starting the region from the master terminal, the default generated for the IMSBATCH procedure is satisfactory.

The optional PRLD= parameter specifies the two-character suffix for the member

DFSMPLxx in IMS.PROCLIB, which lists the preloaded modules. It is optional; it is generated as a null parameter for the IMSBATCH procedure. For BMP regions that do not have other programs scheduled in them, PRLD provides no performance gain.

The PREINIT= parameter specifies the two-character suffix of the member

DFSINTxx in IMS.PROCLIB. This member contains the preinitialization modules that are to receive control.

Chapter 5. Specifying IMS execution parameters

165

166

System Definition

The optional STIMER= and CPUTIME= parameters are generated as null parameters, their defaults are no use of the timer.

The optional PCB= parameter specifies the size of the interregion communication area. This area holds the PCBs included in the PSBs that are used by programs executing in this region. If you do not specify this parameter, the default value is

000. This allocates an area large enough to hold all the PCBs for the largest PSB defined to this IMS system and present in the active IMS.ACBLIBA/B. To specify a value, enter the number of 1024-byte blocks as a three-digit number (for example,

001).

Use the SSM parameter to reference a member in IMS.PROCLIB that identifies the

DB2 for z/OS subsystems that can be accessed from this BMP region. To prevent any access to a DB2 for z/OS subsystem from the BMP, use the name of a null member (no entries). The default (a null value) allows the BMP region to attach to any of the subsystems declared to the IMS control region. If necessary, coordinate this parameter with the corresponding SSM parameter in the IMS procedure. Refer

to the description in “System control and performance EXEC parameters for the

IMS control region” on page 145.

The FMTO= parameter and the type of z/OS dump DD statements selected determine whether IMS dumps are formatted online or offline.

Use the APARM= parameter to specify execution-time parameters unique to this dependent region. This parameter specifies a character string for the application program or Data Capture exit routine. For batch or BMP regions, the APARM= can be used to specify the frequency for checkpoint calls. The APARM= can also be used to pass parameters to the Data Capture exit routine to indicate whether the data should be captured.

Recovery-related EXEC parameters for a batch message region

For a BMP program, you can use the CKPTID= parameter to set a restart position for the program processing. The IMSBATCH procedure generates CKPTID as a null parameter. To invoke restart, create a special version of the procedure containing the exact checkpoint identification. This can be a 1- to 8-character extended checkpoint identifier generated by the program itself, or a 14-character time stamp identifier issued within a message from checkpoint processing. Alternatively, you can specify the value

LAST

as the identifier to cause the last recorded checkpoint to be used. One technique is to restart BMPs with JCL entered in the system reader and have the CKPTID symbolic parameter coded for the EXEC statement.

Related reading:

For more details regarding BMP restart for BMPs that use extended checkpoint, refer to

IMS Version 11 Operations and Automation

In the case where the application program has a SPIE in effect, you specify whether the SPIE is to remain on or off during the DL/I call. Negated SPIEs are reinstated before returning to the application program.

Recommendation:

For performance reasons, do not turn the SPIE on and off for each DL/I call. With PL/I Release 5 and later releases, you can use the PL/I SPIE facility without having IMS reset the SPIE on each DL/I call.

Security-related EXEC parameters for a batch message region

The parameter IMSID is related to security and to operations. You specify the 1-to

4-character identifier for the name of the operational IMS DB/DC environment. If the value does not match the current IMSID of any operating control region, the batch message region is not scheduled.

Chapter 5. Specifying IMS execution parameters

167

168

System Definition