HP B6960-96035 User's Manual

HP Data Protector A.06.10
Concepts guide
B6960-96035
Part number: B6960-96035
First edition: November 2008
Legal and notice information
© Copyright 1999, 2008 Hewlett-Packard Development Company, L.P.
Confidential computer software. Valid license from HP required for possession, use or copying. Consistent
with FAR 12.211 and 12.212, Commercial Computer Software, Computer Software Documentation, and
Technical Data for Commercial Items are licensed to the U.S. Government under vendor's standard
commercial license.
The information contained herein is subject to change without notice. The only warranties for HP products
and services are set forth in the express warranty statements accompanying such products and services.
Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for
technical or editorial errors or omissions contained herein.
Intel, Itanium, Pentium, Intel Inside, and the Intel Inside logo are trademarks or registered trademarks of
Intel Corporation or its subsidiaries in the United States and other countries.
Microsoft, Windows, Windows XP, and Windows NT are U.S. registered trademarks of Microsoft
Corporation.
Adobe and Acrobat are trademarks of Adobe Systems Incorporated.
Java is a US trademark of Sun Microsystems, Inc.
Oracle is a registered US trademark of Oracle Corporation, Redwood City, California.
UNIX is a registered trademark of The Open Group.
Printed in the US
Contents
Publication history .............................................................. 21
About this guide ................................................................. 23
Intended audience .............................................................................................
Documentation set .............................................................................................
Guides ......................................................................................................
Online Help ...............................................................................................
Documentation map ....................................................................................
Abbreviations ......................................................................................
Map ...................................................................................................
Integrations .........................................................................................
Document conventions and symbols .....................................................................
Data Protector graphical user interface .................................................................
General information ..........................................................................................
HP technical support ..........................................................................................
Subscription service ...........................................................................................
HP websites ......................................................................................................
Documentation feedback ....................................................................................
23
23
23
26
27
27
28
29
31
32
33
33
34
34
34
1 About backup and Data Protector ...................................... 35
In this chapter ...................................................................................................
About Data Protector .........................................................................................
Introducing backups and restores .........................................................................
What is a backup? .....................................................................................
What is a restore? ......................................................................................
Backing up a network environment ................................................................
Direct backup .............................................................................................
Data Protector architecture ..................................................................................
Operations in the cell ..................................................................................
Backup sessions ..........................................................................................
Restore sessions .........................................................................................
Enterprise environments ......................................................................................
Splitting an environment into multiple cells ......................................................
Concepts guide
35
35
38
38
39
39
40
40
42
43
44
45
45
3
Media management ..........................................................................................
Backup devices .................................................................................................
User interfaces ..................................................................................................
Data Protector GUI ......................................................................................
Data Protector Java GUI ........................................................................
Overview of tasks to set up Data Protector ............................................................
48
49
50
50
52
55
2 Planning your backup strategy .......................................... 57
In this chapter ................................................................................................... 57
Backup strategy planning ................................................................................... 58
Defining the requirements of a backup strategy ............................................... 58
Factors influencing your backup strategy ........................................................ 60
Preparing a backup strategy plan ................................................................. 60
Planning cells .................................................................................................... 62
One cell or multiple cells? ............................................................................ 62
Installing and maintaining client systems ........................................................ 64
Creating cells in the UNIX environment .......................................................... 64
Creating cells in the Windows environment .................................................... 65
Windows domains ............................................................................... 65
Windows workgroups ........................................................................... 66
Creating cells in a mixed environment ........................................................... 66
Geographically remote cells ......................................................................... 66
Understanding and planning performance ............................................................ 67
The infrastructure ......................................................................................... 67
Network versus local backups ................................................................ 67
Network or server versus direct backups .................................................. 68
Devices ............................................................................................... 68
High performance hardware other than devices ........................................ 69
Advanced high performance configuration .............................................. 69
Using hardware in parallel .................................................................... 69
Configuring backups and restores ................................................................. 70
Software compression ........................................................................... 70
Hardware compression ......................................................................... 70
Full and incremental backups ................................................................. 71
Disk image versus filesystem backups ...................................................... 71
Object distribution to media ................................................................... 71
Disk performance ........................................................................................ 72
SAN performance ....................................................................................... 73
Online database application performance ..................................................... 73
Planning security ............................................................................................... 73
Cells ......................................................................................................... 74
Data Protector users accounts ....................................................................... 74
4
Data Protector user groups ........................................................................... 75
Data Protector user rights ............................................................................. 75
Visibility of backed up data .......................................................................... 76
Data encryption .......................................................................................... 76
How Data Protector AES 256-bit encryption works ................................... 76
How Data Protector drive-based encryption works ..................................... 77
Restore from encrypted backups ............................................................. 78
What is backup ownership? ......................................................................... 78
Clustering ......................................................................................................... 79
Cluster concepts ......................................................................................... 79
Cluster support ........................................................................................... 82
Example cluster environments ....................................................................... 83
Cell Manager installed outside a cluster .................................................. 83
Cell Manager installed outside a cluster, devices connected to the cluster
nodes ................................................................................................. 85
Cell Manager installed in a cluster, devices connected to the cluster
nodes ................................................................................................. 87
Full and incremental backups .............................................................................. 91
Full backups ............................................................................................... 92
Synthetic backup .................................................................................. 92
Incremental backups .................................................................................... 92
Conventional incremental backup ........................................................... 93
Enhanced incremental backup ................................................................ 93
Types of incremental backups ................................................................. 93
Considering restore ..................................................................................... 96
Keeping backed up data and information about the data ....................................... 99
Data protection ........................................................................................... 99
Catalog protection .................................................................................... 100
Logging level ........................................................................................... 100
Browsing files for restore ............................................................................ 100
Enabling the browsing of files and quick restore ..................................... 101
Enabling the restore of files, but not browsing ......................................... 101
Overwriting backed up files with new data ............................................ 101
Exporting media from a cell ................................................................. 102
Backing up data ............................................................................................. 102
Creating a backup specification .................................................................. 103
Selecting backup objects .......................................................................... 103
Backup sessions ........................................................................................ 105
Object mirrors .......................................................................................... 105
Media sets ............................................................................................... 105
Backup types and scheduled backups .......................................................... 105
Scheduling, backup configurations, and sessions ........................................... 106
Scheduling tips and tricks ........................................................................... 106
Concepts guide
5
When to schedule backups ..................................................................
Staggering full backups .......................................................................
Optimizing for restore .........................................................................
Automated or unattended operation ...................................................................
Considerations for unattended backups .......................................................
Duplicating backed up data .............................................................................
Copying objects .......................................................................................
Why use object copy? ........................................................................
Object mirroring .......................................................................................
Copying media ........................................................................................
Automated media copying ...................................................................
Smart media copying using VLS ...........................................................
Restoring data ................................................................................................
Restore duration ........................................................................................
Selection of the media set ..........................................................................
Selection of devices ...................................................................................
Operators are allowed to restore ................................................................
End users are allowed to restore .................................................................
Disaster recovery .............................................................................................
Disaster recovery methods ..........................................................................
Alternative disaster recovery methods ....................................................
107
107
107
110
110
112
113
116
119
122
124
124
125
125
126
126
127
128
128
130
131
3 Media management and devices .................................... 133
In this chapter .................................................................................................
Media management ........................................................................................
Media life cycle ..............................................................................................
Media pools ...................................................................................................
Free pools ................................................................................................
Media pool usage examples ......................................................................
Implementing a media rotation policy ..........................................................
Media rotation and Data Protector ........................................................
Media needed for rotation ...................................................................
Media management before backups begin .........................................................
Initializing or formatting media ...................................................................
Labeling Data Protector media ....................................................................
Location field ............................................................................................
Media management during backup sessions .......................................................
Selecting media for backups ......................................................................
Adding data to media during backup sessions ..............................................
Writing data to several media sets during backup .........................................
Calculating media condition .......................................................................
Media management after backup sessions ..........................................................
6
133
133
134
135
137
140
143
144
144
145
145
145
146
147
147
148
150
150
151
Vaulting ................................................................................................... 151
Restoring from media in a vault ................................................................... 153
Devices .......................................................................................................... 153
Device lists and load balancing .................................................................. 155
How load balancing works .................................................................. 156
Device streaming and concurrency .............................................................. 156
Segment size ............................................................................................ 157
Block size ................................................................................................ 158
Number of disk agent buffers ..................................................................... 159
Device locking and lock names ................................................................... 159
Standalone devices .......................................................................................... 160
Small magazine devices ................................................................................... 161
Large libraries ................................................................................................. 162
Handling of media .................................................................................... 162
Size of a library ........................................................................................ 162
Sharing a library with other applications ..................................................... 163
Enter / eject mail slots ............................................................................... 163
Barcode support ....................................................................................... 163
Cleaning tape support ............................................................................... 164
Sharing a library with multiple systems ........................................................ 164
Data Protector and Storage Area Networks ......................................................... 170
Storage Area Networks ............................................................................. 171
Fibre Channel .......................................................................................... 172
Point-to-point topology ......................................................................... 173
Loop topology .................................................................................... 173
Switched topology .............................................................................. 174
Device sharing in SAN .............................................................................. 175
Configuring multiple paths to physical devices ........................................ 175
Device locking ................................................................................... 177
Indirect and Direct Library Access ............................................................... 178
Indirect Library Access ........................................................................ 178
Direct Library Access .......................................................................... 179
Device sharing in clusters ........................................................................... 180
Static drives ....................................................................................... 180
Floating drives ................................................................................... 181
4 Users and user groups ................................................... 183
In this chapter .................................................................................................
Increased security for Data Protector users ..........................................................
Access to backed up data ..........................................................................
Users and user groups .....................................................................................
Using predefined user groups .....................................................................
Concepts guide
183
183
183
184
185
7
Data Protector user rights ........................................................................... 185
5 The Data Protector internal database ............................... 187
In this chapter ................................................................................................. 187
About the IDB ................................................................................................. 187
The IDB on the Windows Cell Manager ....................................................... 188
The IDB on the UNIX Cell Manager ............................................................. 189
The IDB in the Manager-of-Managers environment ......................................... 189
IDB architecture ............................................................................................... 189
Media Management Database (MMDB) ...................................................... 190
Catalog Database (CDB) ........................................................................... 191
Detail Catalog Binary Files (DCBF) .............................................................. 192
Session Messages Binary Files (SMBF) ......................................................... 193
Serverless Integrations Binary Files (SIBF) ...................................................... 194
IDB operation ................................................................................................. 194
During backup .......................................................................................... 194
During restore ........................................................................................... 195
During object copying or object consolidation .............................................. 195
Exporting media ....................................................................................... 195
Removing the detail catalog ...................................................................... 196
Filenames purge ....................................................................................... 196
File versions purge .................................................................................... 196
Overview of IDB management ........................................................................... 196
IDB growth and performance ............................................................................ 197
Key IDB growth and performance factors ..................................................... 197
IDB growth and performance: key tunable parameters ................................... 198
Logging level as an IDB key tunable parameter ...................................... 199
Catalog protection as an IDB key tunable parameter ............................... 201
Recommended usage of logging level and catalog protection .................. 201
IDB size estimation ................................................................................... 203
6 Service management ..................................................... 205
In this chapter .................................................................................................
Overview .......................................................................................................
Data Protector and service management ......................................................
Native Data Protector functionality .....................................................................
Application Response Measurement version 2.0 (ARM 2.0 API) .......................
Integration with HP Operations Manager software ........................................
SNMP traps .............................................................................................
The monitor ..............................................................................................
Reporting and notification ..........................................................................
Event logging and notification ....................................................................
8
205
205
206
207
208
210
210
210
211
212
Data Protector log files ........................................................................
Windows application log ...........................................................................
Java-based online reporting .......................................................................
Data Protector checking and maintenance mechanism ...................................
Central management, distributed environment ...............................................
Using the data provided by Data Protector ...................................................
Service management integrations ......................................................................
Data Protector OM-R integration .................................................................
Data Protector OM SIP ..............................................................................
213
213
213
214
214
214
215
215
217
7 How Data Protector operates .......................................... 219
In this chapter .................................................................................................
Data Protector processes or services ...................................................................
Backup sessions ..............................................................................................
Scheduled and interactive backup sessions ...................................................
Backup session data flow and processes ......................................................
Pre-exec and post-exec commands ..............................................................
Queuing of backup sessions .......................................................................
Mount requests in backup sessions ..............................................................
Backing up with disk discovery ...................................................................
Restore sessions ...............................................................................................
Restore session data flow and processes ......................................................
Queuing of restore sessions ........................................................................
Mount requests in a restore session ..............................................................
Parallel restores .........................................................................................
Fast multiple single file restore ....................................................................
Object copy sessions .......................................................................................
Automated and interactive object copy sessions ............................................
Object copy session data flow and processes ...............................................
Queuing of object copy sessions .................................................................
Mount requests in an object copy session .....................................................
Object consolidation sessions ...........................................................................
Automated and interactive object consolidation sessions ................................
Object consolidation session data flow and processes ...................................
Queuing of object consolidation sessions .....................................................
Mount requests in an object consolidation session .........................................
Media management sessions ............................................................................
Media management session data flow .........................................................
219
219
220
221
221
223
224
224
225
225
226
227
228
228
229
229
230
230
232
232
233
233
233
234
235
235
235
8 Integration with database applications ............................. 237
In this chapter ................................................................................................. 237
Overview of database operation ....................................................................... 237
Concepts guide
9
Filesystem backup of databases and applications ................................................ 239
Online backup of databases and applications .................................................... 239
9 Direct backup ............................................................... 243
In this chapter .................................................................................................
Overview .......................................................................................................
Direct backup ...........................................................................................
Direct backup benefits .........................................................................
How direct backup works ...........................................................................
Environment .......................................................................................
About resolve .....................................................................................
About XCopy .....................................................................................
XCopy + Resolve ................................................................................
Direct backup process flow .........................................................................
Backup stages for data files .................................................................
Restore ..............................................................................................
Requirements and support ................................................................................
Supported configurations ..................................................................................
Three hosts: CM, application, Resolve ..........................................................
Two Hosts: Cell Manager/Resolve Agent and application ..............................
Basic configuration: single host ...................................................................
243
243
244
245
245
246
247
247
248
248
248
249
249
250
250
251
251
10 Disk backup ................................................................ 253
In this chapter .................................................................................................
Overview .......................................................................................................
Disk backup benefits ........................................................................................
Data Protector disk-based devices ......................................................................
253
253
254
255
11 Synthetic backup .......................................................... 257
In this chapter .................................................................................................
Overview .......................................................................................................
Synthetic backup benefits .................................................................................
How Data Protector synthetic backup works ........................................................
Synthetic backup and media space consumption ...........................................
Restore and synthetic backup ............................................................................
How data protection periods affect restore from synthetic backup ....................
257
257
258
258
260
260
262
12 Split mirror concepts .................................................... 265
In this chapter ................................................................................................. 265
Overview ....................................................................................................... 265
Supported configurations .................................................................................. 269
10
Local mirror - dual host ..............................................................................
Local mirror - single host ............................................................................
Remote mirror ...........................................................................................
Local/remote mirror combination ................................................................
Other configurations .................................................................................
269
270
270
272
273
13 Snapshot concepts ....................................................... 275
In this chapter .................................................................................................
Overview .......................................................................................................
Storage virtualization .................................................................................
Snapshot concepts ....................................................................................
Snapshot backup types ..............................................................................
Instant recovery ........................................................................................
Replica set and replica set rotation ..............................................................
Types of snapshots ....................................................................................
Supported configurations ..................................................................................
Basic configuration: single disk array - dual host ...........................................
Other supported configurations ...................................................................
Other configurations .................................................................................
275
275
275
276
278
279
279
279
281
281
282
286
14 Microsoft Volume Shadow Copy service .......................... 289
In this chapter .................................................................................................
Overview .......................................................................................................
Data Protector Volume Shadow Copy integration .................................................
VSS filesystem backup and restore .....................................................................
289
289
293
295
A Backup scenarios .......................................................... 297
In this appendix ..............................................................................................
Considerations .........................................................................................
Company XYZ ................................................................................................
Environment .............................................................................................
Backup strategy requirements .....................................................................
Proposed solution ......................................................................................
Company ABC ................................................................................................
Environment .............................................................................................
Backup strategy requirements .....................................................................
Proposed solution ......................................................................................
297
297
299
299
302
303
313
313
315
317
B Further information ........................................................ 331
In this appendix .............................................................................................. 331
Backup generations ........................................................................................ 331
Concepts guide
11
Examples of automated media copying ..............................................................
Example 1: automated media copying of filesystem backups ...........................
Incr1 backup .....................................................................................
Full backup ........................................................................................
Example 2: automated media copying of Oracle database backups ................
Full backup ........................................................................................
Internationalization ..........................................................................................
Localization ..............................................................................................
File name handling ...................................................................................
Background .......................................................................................
File name handling during backup ........................................................
Browsing file names ............................................................................
File name handling during restore .........................................................
332
333
333
335
338
339
340
340
341
341
342
342
342
Glossary ......................................................................... 345
Index .............................................................................. 403
12
Figures
1 Data Protector graphical user interface ............................................... 33
2 Backup process ............................................................................... 39
3 Restore process ............................................................................... 39
4 Network backup .............................................................................. 40
5 The Data Protector cell (physical view and logical view) ....................... 41
6 Backup or restore operation .............................................................. 43
7 Backup session ................................................................................ 44
8 Restore session ................................................................................ 44
9 Large Data Protector enterprise environment ........................................ 45
10 Single-point management of multiple cells ........................................... 46
11 Manager-of-Managers environment .................................................... 47
12 How backup specifications, devices, and media pools are related ......... 49
13 Using the Data Protector user interface ............................................... 50
14 Original Data Protector GUI .............................................................. 52
15 Data Protector Java GUI ................................................................... 52
16 Data Protector Java GUI architecture .................................................. 53
17 Backup session with AES 256-bit encryption ........................................ 77
18 Backup session with drive-based encryption ........................................ 78
19 Typical cluster ................................................................................. 80
20 Cell Manager installed outside a cluster ............................................. 84
21 Cell Manager installed outside a cluster, devices connected to the cluster
nodes ............................................................................................ 86
22 Cell Manager installed in the cluster, devices connected to cluster
nodes ............................................................................................ 89
Concepts guide
13
23 Incremental backups ........................................................................ 95
24 Leveled incremental backups ............................................................. 95
25 Media needed to restore from simple and leveled incremental
backups ......................................................................................... 98
26 Media needed to restore from leveled incremental backups ................... 98
27 Backup session .............................................................................. 102
28 Full backup with daily simple incremental backups ............................. 108
29 Full backup with daily level 1 incremental backups ............................. 109
30 Full backup with mixed incremental backups ..................................... 110
31 Object copy concept .................................................................... 114
32 Freeing media .............................................................................. 117
33 Demultiplexing a medium ............................................................... 118
34 Disk staging concept ..................................................................... 119
35 Object mirroring ............................................................................ 121
36 Free pools .................................................................................... 138
37 A simple one device/one media pool relation ................................... 140
38 Configuration of media pools for large libraries ................................. 141
39 Multiple devices, single media pool ................................................. 142
40 Multiple devices, multiple media pools ............................................. 143
41 Multiple objects and sessions per medium, sequential writes ................ 149
42 Multiple objects and sessions per medium, concurrent writes ............... 149
43 Multiple media per session, multiple media per object ........................ 149
44 Each object written on a separate medium ........................................ 150
45 Data format .................................................................................. 158
46 Device locking and device names .................................................... 160
47 Connecting drives to multiple systems ............................................... 165
48 Sharing a SCSI library (robotics attached to a Data Protector Client
System) ........................................................................................ 168
49 Sharing a SCSI library (robotics attached to an NDMP Server) ............ 169
14
50 Sharing an ADIC/GRAU or StorageTek ACS library ........................... 170
51 Storage Area Network ................................................................... 172
52 Loop initialization protocol .............................................................. 174
53 Example multipath configuration ...................................................... 176
54 Indirect Library Access ................................................................... 179
55 Direct Library Access ...................................................................... 180
56 IDB parts ...................................................................................... 190
57 The influence of logging level and catalog protection on IDB growth .... 199
58 Service management information flow .............................................. 207
59 Example of an IT service provider environment with service management
access through the client portal ....................................................... 215
60 Data Protector Reporter example ..................................................... 216
61 Operational error status report ........................................................ 217
62 Direct SIP integration example ......................................................... 218
63 Backup session information flow (1) .................................................. 222
64 Backup session information flow - multiple sessions ............................. 223
65 Restore session information flow ...................................................... 227
66 Parallel restore session flow ............................................................. 229
67 Object copy session information flow .............................................. 232
68 Relational database ....................................................................... 238
69 Data Protector integration with databases ......................................... 240
70 Direct backup architecture .............................................................. 246
71 Basic three host configuration .......................................................... 251
72 Synthetic backup ........................................................................... 259
73 Virtual full backup .......................................................................... 260
74 Full and incremental backups .......................................................... 261
75 Synthetic backup ........................................................................... 261
76 Regular synthetic backup ................................................................ 262
77 Synthetic backup and object copy .................................................... 262
Concepts guide
15
78 Split mirror backup concept ............................................................ 266
79 Local mirror - dual host (full performance, Zero Downtime Backup) ....... 269
80 Split mirror - remote mirror (LAN-free remote backup - data HA) ........... 271
81 Local/remote mirror combination (disaster recovery integrated backup
[Service HA - HP-UX only]) .............................................................. 272
82 Snapshot backup ........................................................................... 277
83 Single disk array - dual host (full performance, Zero Downtime
Backup) ........................................................................................ 281
84 Multiple disk arrays - dual host ........................................................ 282
85 Multiple application hosts - single backup host .................................. 283
86 Disk array(s) - single host ................................................................ 284
87 LVM mirroring - HP StorageWorks Virtual Array only .......................... 285
88 Campus Cluster with LVM Mirroring - HP StorageWorks Virtual Array
only ............................................................................................. 286
89 Actors of the traditional backup model ............................................. 292
90 Actors of the VSS backup model ...................................................... 292
91 Current XYZ backup topology .......................................................... 300
92 Proposed XYZ backup topology ....................................................... 305
93 Input parameters ........................................................................... 306
94 Results .......................................................................................... 306
95 Current ABC Cape Town backup topology ........................................ 314
96 ABC enterprise environment ............................................................ 318
97 ABC Cape Town enterprise backup environment ................................ 321
98 Input parameters ........................................................................... 322
99 Results .......................................................................................... 322
100 Backup generations ....................................................................... 332
101 Incr1 backup and automated media copying .................................... 335
102 Full backup and automated media copying ....................................... 337
103 Overview of backup and automated media copy sessions ................... 338
104 Full database backup and automated media copying ......................... 339
16
105 Overview of backup and automated media copy sessions ................... 340
Concepts guide
17
Tables
1 Edition history ................................................................................. 21
2 Document conventions ...................................................................... 31
3 Backup behavior ............................................................................. 85
4 Backup behavior ............................................................................. 87
5 Backup behavior ............................................................................. 90
6 Comparison of full and incremental backup ........................................ 91
7 Relative referencing of backup runs .................................................... 95
8 The staggered approach ................................................................ 107
9 Data Protector data duplication methods .......................................... 112
10 Required Data Protector Media Agent for drive control ....................... 166
11 Required Data Protector Media Agent for robotic control ..................... 167
12 Data Protector predefined user groups .............................................. 185
13 ARM functionality .......................................................................... 209
14 Benefits of using VSS ..................................................................... 293
15 Hardware and software environment of XYZ ...................................... 299
16 Proposed environment .................................................................... 303
17 The staggering approach ................................................................ 308
18 Remote full backups to the HP DLT 4115 library .................................. 309
19 Size of backup environment ............................................................ 313
20 Maximum acceptable downtime for recovery .................................... 315
21 How long data should be kept ........................................................ 316
22 Amount of data to be backed up ..................................................... 316
23 Amount of data to be backed up in five years ................................... 317
24 ABC cell configuration ................................................................... 319
18
25 ABC’s Media Pool Usage ............................................................... 325
26 The Staggering Approach for ABC Cape Town .................................. 326
27 ABC’s backup specification configuration ......................................... 327
Concepts guide
19
20
Publication history
Guide updates may be issued between editions to correct errors or document product
changes. To ensure that you receive updated or new editions, subscribe to the
appropriate product support service. See your HP sales representative for details.
Table 1 Edition history
Part number
Guide edition
Product
B6960-90059
August 2002
Data Protector Release A.05.00
B6960-90080
May 2003
Data Protector Release A.05.10
B6960-90105
October 2004
Data Protector Release A.05.50
B6960-96001
August 2006
Data Protector Release A.06.00
B6960-96035
November 2008
Data Protector Release A.06.10
Concepts guide
21
22
Publication history
About this guide
This guide describes Data Protector concepts. Read this manual to fully understand
the fundamentals and the model of Data Protector.
Intended audience
This guide is intended for users interested in understanding the concepts of Data
Protector operation and for people who plan company backup strategies. Depending
on the required level of detail, you can also use this manual together with the Data
Protector online Help.
Documentation set
Other documents and online Help provide related information.
Guides
Data Protector guides are available in printed format and in PDF format. Install the
PDF files during the Data Protector setup procedure by selecting the English
Documentation & Help component on Windows or the OB2-DOCS component
on UNIX. Once installed, the guides reside in the Data_Protector_home\docs
directory on Windows and in the /opt/omni/doc/C directory on UNIX.
You can find these documents from the Manuals page of the HP Business Support
Center website:
http://www.hp.com/support/manuals
In the Storage section, click Storage Software and then select your product.
• HP Data Protector concepts guide
This guide describes Data Protector concepts and provides background information
on how Data Protector works. It is intended to be used with the task-oriented
online Help.
Concepts guide
23
• HP Data Protector installation and licensing guide
This guide describes how to install the Data Protector software, taking into account
the operating system and architecture of your environment. This guide also gives
details on how to upgrade Data Protector, as well as how to obtain the proper
licenses for your environment.
• HP Data Protector troubleshooting guide
This guide describes how to troubleshoot problems you may encounter when
using Data Protector.
• HP Data Protector disaster recovery guide
This guide describes how to plan, prepare for, test and perform a disaster
recovery.
• HP Data Protector integration guides
These guides describe how to configure and use Data Protector to back up and
restore various databases and applications. They are intended for backup
administrators or operators. There are four guides:
• HP Data Protector integration guide for Microsoft applications: SQL Server,
SharePoint Portal Server, Exchange Server, and Volume Shadow Copy Service
This guide describes the integrations of Data Protector with the following
Microsoft applications: Microsoft Exchange Server, Microsoft SQL Server, and
Volume Shadow Copy Service.
• HP Data Protector integration guide for Oracle and SAP
This guide describes the integrations of Data Protector with Oracle, SAP R3,
and SAP DB/MaxDB.
• HP Data Protector integration guide for IBM applications: Informix, DB2, and
Lotus Notes/Domino
This guide describes the integrations of Data Protector with the following IBM
applications: Informix Server, IBM DB2, and Lotus Notes/Domino Server.
• HP Data Protector integration guide for VMware Virtual Infrastructure, Sybase,
Network Node Manager, and Network Data Management Protocol Server
This guide describes the integrations of Data Protector with VMware Virtual
Infrastructure, Sybase, Network Node Manager, and Network Data
Management Protocol Server.
• HP Data Protector integration guide for HP Service Information Portal
This guide describes how to install, configure, and use the integration of Data
Protector with HP Service Information Portal. It is intended for backup
administrators. It discusses how to use the application for Data Protector service
management.
24
About this guide
• HP Data Protector integration guide for HP Reporter
This manual describes how to install, configure, and use the integration of Data
Protector with HP Reporter. It is intended for backup administrators. It discusses
how to use the application for Data Protector service management.
• HP Data Protector integration guide for HP Operations Manager for UNIX
This guide describes how to monitor and manage the health and performance of
the Data Protector environment with HP Operations Manager and HP Service
Navigator on UNIX.
• HP Data Protector integration guide for HP Operations Manager for Windows
This guide describes how to monitor and manage the health and performance of
the Data Protector environment with HP Operations Manager and HP Service
Navigator on Windows.
• HP Data Protector integration guide for HP Performance Manager and HP
Performance Agent
This guide provides information about how to monitor and manage the health
and performance of the Data Protector environment with HP Performance Manager
(PM) and HP Performance Agent (PA) on Windows, HP-UX, Solaris, and Linux.
• HP Data Protector zero downtime backup concepts guide
This guide describes Data Protector zero downtime backup and instant recovery
concepts and provides background information on how Data Protector works in
a zero downtime backup environment. It is intended to be used with the
task-oriented HP Data Protector zero downtime backup administrator's guide and
the HP Data Protector zero downtime backup integration guide.
• HP Data Protector zero downtime backup administrator's guide
This guide describes how to configure and use the integration of Data Protector
with HP StorageWorks Virtual Array, HP StorageWorks Enterprise Virtual Array,
EMC Symmetrix Remote Data Facility and TimeFinder, and HP StorageWorks
Disk Array XP. It is intended for backup administrators or operators. It covers the
zero downtime backup, instant recovery, and the restore of filesystems and disk
images.
• HP Data Protector zero downtime backup integration guide
This guide describes how to configure and use Data Protector to perform zero
downtime backup, instant recovery, and standard restore of Oracle, SAP R/3,
Microsoft Exchange Server, and Microsoft SQL Server databases. The guide also
describes how to configure and use Data Protector to perform backup and restore
using the Microsoft Volume Shadow Copy Service.
• HP Data Protector MPE/iX system user guide
Concepts guide
25
This guide describes how to configure MPE/iX clients and how to back up and
restore MPE/iX data.
• HP Data Protector Media Operations user's guide
This guide provides tracking and management of offline storage media. It describes
the tasks of installing and configuring the application, performing daily media
operations and producing reports.
• HP Data Protector product announcements, software notes, and references
This guide gives a description of new features of HP Data Protector A.06.10. It
also provides information on supported configurations (devices, platforms and
online database integrations, SAN, and ZDB), required patches, and limitations,
as well as known problems and workarounds. An updated version of the supported
configurations is available at http://www.hp.com/support/manuals.
• HP Data Protector product announcements, software notes, and references for
integrations to HP Operations Manager, HP Reporter, HP Performance Manager,
HP Performance Agent, and HP Service Information Portal
This guide fulfills a similar function for the listed integrations.
• HP Data Protector Media Operations product announcements, software notes,
and references
This guide fulfills a similar function for Media Operations.
• HP Data Protector command line interface reference
This guide describes the Data Protector command-line interface, command options
and their usage as well as provides some basic command-line examples.
Online Help
Data Protector provides context-sensitive (F1) Help and Help Topics for Windows
and UNIX platforms.
You can access the online Help from the top-level directory on the installation
DVD-ROM without installing Data Protector:
• Windows: Unzip DP_help.zip and open DP_help.chm.
• UNIX: Unpack the zipped tar file DP_help.tar.gz, and access the online Help
system through DP_help.htm.
26
About this guide
Documentation map
Abbreviations
Abbreviations in the documentation map that follows are explained below. The guide
titles are all preceded by the words “HP Data Protector”.
Abbreviation
Guide
CLI
Command line interface reference
Concepts
Concepts guide
DR
Disaster recovery guide
GS
Getting started guide
Help
Online Help
IG-IBM
Integration guide for IBM applications: Informix, DB2, and Lotus
Notes/Domino
IG-MS
Integration guide for Microsoft applications: SQL Server, SharePoint
Portal Server, Exchange Server, and Volume Shadow Copy Service
IG-O/S
Integration guide for Oracle and SAP
IG-OMU
Integration guide for HP Operations Manager for UNIX
IG-OMW
Integration guide for HP Operations Manager for Windows
IG-PM/PA
Integration guide for HP Performance Manager and HP Performance
Agent
IG-Report
Integration guide for HP Reporter
IG-SIP
Integration guide for HP Service Information Portal
IG-Var
Integration guide for VMware Virtual Infrastructure, Sybase, Network
Node Manager, and Network Data Management Protocol Server
Concepts guide
27
Abbreviation
Guide
Install
Installation and licensing guide
MO GS
Media Operations getting started guide
MO RN
Media Operations product announcements, software notes, and
references
MO UG
Media Operations user guide
MPE/iX
MPE/iX system user guide
PA
Product announcements, software notes, and references
Trouble
Troubleshooting guide
ZDB Admin
ZDB administrator's guide
ZDB Concept
ZDB concepts guide
ZDB IG
ZDB integration guide
Map
The following table shows where to find information of different kinds. Shaded squares
are a good place to look first.
28
About this guide
Integrations
Look in these guides for details of the following integrations:
Integration
Guide
HP Operations Manager for UNIX/for Windows
IG-OMU, IG-OMW
HP Performance Manager
IG-PM/PA
HP Performance Agent
IG-PM/PA
Concepts guide
29
30
Integration
Guide
HP Reporter
IG-R
HP Service Information Portal
IG-SIP
HP StorageWorks Disk Array XP
all ZDB
HP StorageWorks Enterprise Virtual Array (EVA)
all ZDB
HP StorageWorks Virtual Array (VA)
all ZDB
IBM DB2 UDB
IG-IBM
Informix
IG-IBM
Lotus Notes/Domino
IG-IBM
Media Operations
MO User
MPE/iX system
MPE/iX
Microsoft Exchange Server
IG-MS, ZDB IG
Microsoft Exchange Single Mailbox
IG-MS
Microsoft SQL Server
IG-MS, ZDB IG
Microsoft Volume Shadow Copy Service (VSS)
IG-MS, ZDB IG
NDMP Server
IG-Var
Network Node Manager (NNM)
IG-Var
Oracle
IG-O/S
Oracle ZDB
ZDB IG
SAP DB
IG-O/S
SAP R/3
IG-O/S, ZDB IG
About this guide
Integration
Guide
Sybase
IG-Var
EMC Symmetrix
all ZDB
VMware
IG-Var
Document conventions and symbols
Table 2 Document conventions
Convention
Element
Blue text: Table 2 on page 31
Cross-reference links and e-mail addresses
Blue, underlined text: http://www.hp.com
website addresses
Italic text
Text emphasis
• File and directory names
• System output
Monospace text
• Code
• Commands, their arguments, and
argument values
Monospace, italic text
text
• Code variables
• Command variables
Emphasized monospace text
CAUTION:
Indicates that failure to follow directions could result in damage to equipment or data.
IMPORTANT:
Provides clarifying information or specific instructions.
Concepts guide
31
NOTE:
Provides additional information.
TIP:
Provides helpful hints and shortcuts.
Data Protector graphical user interface
Data Protector provides a cross-platform (Windows and UNIX) graphical user
interface. You can use the original Data Protector GUI (Windows only) or the Data
Protector Java GUI. For information about the Data Protector graphical user interface,
see the online Help.
32
About this guide
Figure 1 Data Protector graphical user interface
General information
General information about Data Protector can be found at http://www.hp.com/go/
dataprotector.
HP technical support
For worldwide technical support information, see the HP support website:
http://www.hp.com/support
Before contacting HP, collect the following information:
•
•
•
•
•
•
Product model names and numbers
Technical support registration number (if applicable)
Product serial numbers
Error messages
Operating system type and revision level
Detailed questions
Concepts guide
33
Subscription service
HP recommends that you register your product at the Subscriber's Choice for Business
website:
http://www.hp.com/go/e-updates
After registering, you will receive e-mail notification of product enhancements, new
driver versions, firmware updates, and other product resources.
HP websites
For additional information, see the following HP websites:
•
•
•
•
http://www.hp.com
http://www.hp.com/go/software
http://www.hp.com/support/manuals
http://www.hp.com/support/downloads
Documentation feedback
HP welcomes your feedback.
To make comments and suggestions about product documentation, please send a
message to DP.DocFeedback@hp.com. All submissions become the property of HP.
34
About this guide
1 About backup and Data
Protector
In this chapter
This chapter provides an overview of backup and restore concepts. It introduces Data
Protector architecture, media management, user interfaces, backup devices, and
other features. The chapter concludes with an overview of Data Protector configuration
and other tasks needed to set up Data Protector.
It is organized as follows:
“About Data Protector” on page 35
“Introducing backups and restores” on page 38
“Data Protector architecture” on page 40
“Enterprise environments” on page 45
“Media management” on page 48
“Backup devices” on page 49
“User interfaces” on page 50
“Overview of tasks to set up Data Protector” on page 55
About Data Protector
HP Data Protector is a backup solution that provides reliable data protection and
high accessibility for your fast growing business data. Data Protector offers
comprehensive backup and restore functionality specifically tailored for enterprise-wide
and distributed environments. The following list describes major Data Protector
features:
• Scalable and Highly Flexible Architecture
Concepts guide
35
Data Protector can be used in environments ranging from a single system to
thousands of systems on several sites. Due to the network component concept of
Data Protector, elements of the backup infrastructure can be placed in the topology
according to user requirements. The numerous backup options and alternatives
to setting up a backup infrastructure allow the implementation of virtually any
configuration you want. Data Protector also enables the use of advanced backup
concepts, such as synthetic backup and disk staging.
• Easy Central Administration
Through its easy-to-use graphical user interface (GUI), Data Protector allows you
to administer your complete backup environment from a single system. To ease
operation, the GUI can be installed on various systems to allow multiple
administrators to access Data Protector via their locally installed consoles. Even
multiple backup environments can be managed from a single system. The Data
Protector command-line interface allows you to manage Data Protector using
scripts.
• High Performance Backup
Data Protector enables you to perform backup to several hundred backup devices
simultaneously. It supports high-end devices in very large libraries. Various backup
possibilities, such as local backup, network backup, online backup, disk image
backup, synthetic backup, backup with object mirroring, and built-in support for
parallel data streams allow you to tune your backups to best fit your requirements.
• Data security
To enhance the security of your data, Data Protector lets you encrypt your backups
so that they become protected from others. Data Protector offers two data
encryption techniques: software-based and drive-based.
• Supporting Mixed Environments
As Data Protector supports heterogeneous environments, most features are common
to the UNIX and Windows platforms. The UNIX and Windows Cell Managers
can control all supported client platforms (UNIX, Windows, and Novell NetWare).
The Data Protector user interface can access the entire Data Protector functionality
on all supported platforms.
• Easy Installation for Mixed Environments
The Installation Server concept simplifies the installation and upgrade procedures.
To remotely install UNIX clients, you need an Installation Server for UNIX. To
remotely install Windows clients, you need an Installation Server for Windows.
The remote installation can be performed from any client with an installed Data
Protector GUI. For supported platforms for the Installation Server, see the HP Data
Protector product announcements, software notes, and references.
• High Availability Support
36
About backup and Data Protector
Data Protector enables you to meet the needs for continued business operations
around the clock. In today's globally distributed business environment,
company-wide information resources and customer service applications must
always be available. Data Protector enables you to meet high availability needs
by:
• Integrating with clusters to ensure fail-safe operation with the ability to back
up virtual nodes. For a list of supported clusters, see the HP Data Protector
product announcements, software notes, and references.
• Enabling the Data Protector Cell Manager itself to run on a cluster.
• Supporting all popular online database Application Programming Interfaces.
• Integrating with advanced high availability solutions like EMC Symmetrix, HP
StorageWorks Disk Array XP, HP StorageWorks Virtual Array, or HP
StorageWorks Enterprise Virtual Array.
• Providing various disaster recovery methods for Windows and UNIX platforms.
• Offering methods of duplicating backed up data during and after the backup
to improve fault tolerance of backups or for redundancy purposes.
• Easy Restore
Data Protector includes an internal database that keeps track of data such as
which files from which system are kept on a particular medium. In order to restore
any part of a system, simply browse the files and directories. This provides fast
and convenient access to the data to be restored.
• Automated or Unattended Operation
With the internal database, Data Protector keeps information about each Data
Protector medium and the data on it. Data Protector provides sophisticated media
management functionality. For example, it keeps track of how long a particular
backup needs to remain available for restoring, and which media can be (re)used
for backups.
The support of very large libraries complements this, allowing for unattended
operation over several days or weeks (automated media rotation). Additionally,
when new disks are connected to systems, Data Protector can automatically detect
(or discover) the disks and back them up. This eliminates the need to adjust backup
configurations manually.
• Service Management
Data Protector is the first backup and restore management solution to support
service management. The integration with Application Response Management
(ARM) and Data Source Integration (DSI) enables powerful support of Service
Level Management (SLM) and Service Level Agreements (SLA) concepts by
providing relevant data to management and planning systems.
Concepts guide
37
The DSI integration provides a set of scripts and configuration files from which
users are able to see how to add their own queries using Data Protector reporting
capabilities.
• Monitoring, Reporting and Notification
Superior web reporting and notification capabilities allow you to easily view the
backup status, monitor active backup operations, and customize reports. Reports
can be generated using the Data Protector GUI, or using the omnirpt command
on systems running UNIX or Windows, as well as using Java-based online
generated web reports.
You can schedule reports to be issued at a specific time or to be attached to a
predefined set of events, such as the end of a backup session or a mount request.
In addition, the Data Protector auditing functionality enables you to collect a
subset of backup session information and provides an overview of backup
operations. Backup session information is recorded to the audit log files.
• Integration with Online Database Applications
Data Protector provides online backup of Microsoft Exchange Server, Microsoft
SQL Server, Oracle, Informix Server, SAP R/3, Lotus Notes/Domino Server, IBM
DB2 UDB, Sybase database objects, and VMware Virtual Infrastructure objects.
For a list of supported versions for a particular operating system, see the HP Data
Protector product announcements, software notes, and references.
• Integration with Other Products
Additionally, Data Protector integrates with EMC Symmetrix, Microsoft Cluster
Server, MC/ServiceGuard and other products.
For detailed documentation describing the features of Data Protector, including
integrations, as well as the latest platform and integration support information, consult
the HP Data Protector home page at http://www.hp.com/support/manuals.
Introducing backups and restores
This section explains basic backup and restore concepts.
What is a backup?
A backup is a process that creates a copy of data on backup media. This copy is
stored and kept for future use in case the original is destroyed or corrupted.
A high-level presentation of a backup is shown in Figure 2 on page 39.
38
About backup and Data Protector
Figure 2 Backup process
In most cases, the source is data on a disk, such as files, directories, databases, and
applications. If the backup is expected to be used for disaster recovery, it needs to
be consistent.
Software that actually copies data to the destination is a backup application. The
destination is a backup device, such as a tape drive, with media to which a copy of
the data is written.
What is a restore?
A restore is a process that recreates the original data from a backup copy. This
process consists of the preparation and actual restore of data, and some post-restore
actions that make that data ready for use.
Figure 3 Restore process
The source is a backup copy. A restore application is software that actually writes
data to a destination. The destination is usually a disk to which the original data is
written.
Backing up a network environment
During backups in a network environment, data is transferred over the network from
systems to be backed up to media on systems with backup devices, where the data
is stored.
Concepts guide
39
Figure 4 Network backup
To accomplish backup of a network environment you need an application that allows
you to:
• Attach backup devices to any system in the network
This enables local backups of systems with large volumes of data and network
backups in order to reduce backup device costs.
• Route backup data flow to any network path
• Route backup data away from the LAN and onto a SAN when data volume or
network traffic makes LAN transfer inefficient
• Manage backup activities from any system
• Integrate into the IT management framework
• Support many different types of systems to be backed up
Direct backup
A direct backup is one in which you can send data directly from disk to tape in the
SAN without involving a dedicated backup server for the data movement.
The use of a filesystem-independent ability to resolve data is fully integrated with the
industry-standard XCOPY functionality that is embedded in supported disk arrays
and bridges, removing the need for a separate data mover appliance.
Data Protector architecture
The Data Protector cell, shown in Figure 5 on page 41, is a network environment
that has a Cell Manager, client systems, and devices. The Cell Manager is the central
control point where Data Protector software is installed. After installing Data Protector
software, you can add systems to be backed up. These systems become Data Protector
client systems that are part of the cell. When Data Protector backs up files, it saves
them to media in backup devices.
40
About backup and Data Protector
The Data Protector internal database (IDB) keeps track of the files you back up so
that you can browse and easily recover the entire system or single files.
Data Protector facilitates backup and restore jobs. You can do an immediate (or
interactive) backup using the Data Protector user interface. You can also schedule
your backups to run unattended.
Figure 5 The Data Protector cell (physical view and logical view)
NOTE:
The GUI and the Cell Manager systems can run on UNIX and Windows operating systems;
they do not have to run the same operating system. For a list of supported operating
systems for a particular Data Protector component, see the HP Data Protector product
announcements, software notes, and references.
Cell Manager
The Cell Manager is the main system in the cell. The Cell Manager:
• Manages the cell from a central point
• Contains the IDB
The IDB contains information about backup details such as, backup durations,
media IDs, and session IDs
• Runs core Data Protector software
• Runs Session Managers that start and stop backup and restore sessions and write
session information to the IDB
Concepts guide
41
Systems to be backed up
Client systems you want to back up must have the Data Protector Disk Agent (DA),
also called Backup Agent, installed. To back up online database integrations, install
the Application Agent. In the rest of the manual, the term Disk Agent will be used for
both agents. The Disk Agent reads or writes data from a disk on the system and sends
or receives data from a Media Agent. The Disk Agent is also installed on the Cell
Manager, thus allowing you to back up data on the Cell Manager, the Data Protector
configuration, and the IDB.
Systems with backup devices
Client systems with connected backup devices must have a Data Protector Media
Agent (MA) installed. Such client systems are also called Drive Servers. A backup
device can be connected to any system and not only to the Cell Manager. A Media
Agent reads or writes data from or to media in the device and sends or receives data
from the Disk Agent.
Systems with a user interface
You can manage Data Protector from any system on the network on which the Data
Protector graphical user interface (GUI) is installed. Therefore, you can have the Cell
Manager system in a computer room while managing Data Protector from your
desktop system.
Installation Server
The Installation Server holds a repository of the Data Protector software packages
for a specific architecture. The Cell Manager is by default also an Installation Server.
At least two Installation Servers are needed for mixed environments: one for UNIX
systems and one for Windows systems.
Operations in the cell
The Data Protector Cell Manager controls backup and restore sessions, which perform
all the required actions for a backup or restore, respectively, as shown in Figure
6 on page 43.
42
About backup and Data Protector
Figure 6 Backup or restore operation
Backup sessions
What is a backup session?
A backup session, shown in Figure 7 on page 44, is a process that creates a copy
of data on storage media. It is started either interactively by an operator using the
Data Protector user interface, or unattended using the Data Protector Scheduler.
How does it work?
The Backup Session Manager process starts Media Agent(s) and Disk Agent(s),
controls the session, and stores generated messages to the IDB. Data is read by the
Disk Agent and sent to a Media Agent, which saves it to media.
Concepts guide
43
Figure 7 Backup session
A typical backup session is more complex than the one shown in Figure 7 on page 44.
A number of Disk Agents read data from multiple disks in parallel and send data to
one or more Media Agents. For more information on complex backup sessions, see
Chapter 7 on page 219.
Restore sessions
What is a restore session?
A restore session, shown in Figure 8 on page 44, is a process that restores data from
previous backups to a disk. The restore session is interactively started by an operator
using the Data Protector user interface.
How does it work?
After you have selected the files to be restored from a previous backup, you invoke
the actual restore. The Restore Session Manager process starts the needed Media
Agent(s) and Disk Agent(s), controls the session, and stores messages in the IDB.
Data is read by a Media Agent and sent to the Disk Agent, which writes it to disks.
Figure 8 Restore session
A restore session may be more complex than the one shown in Figure 8 on page 44.
For more information on restore sessions, see Chapter 7 on page 219.
44
About backup and Data Protector
Enterprise environments
What is an enterprise environment?
A typical enterprise network environment, shown in Figure 9 on page 45, consists
of a number of systems from different vendors with different operating systems. The
systems may be located in different geographical areas and time zones. All the
systems are connected with LAN or WAN networks operating at various
communication speeds.
When to use an enterprise environment
This solution can be used when several geographically separated sites require common
backup policies to be used. It can also be used when all departments at the same
site want to share the same set of backup devices.
Figure 9 Large Data Protector enterprise environment
Configuring and managing backups of such a heterogeneous environment is
challenging. Data Protector functionality has been designed to highly simplify this
task. For information about the Manager of Managers (MoM), see MoM on page 46.
Splitting an environment into multiple cells
You may decide to split large environments into multiple cells for a number of reasons:
Concepts guide
45
Why split large environments into multiple cells?
•
•
•
•
•
Geographical grouping of systems.
Logical grouping of systems, for example, departments.
Slow network connection between some systems.
Performance considerations.
Separate administrative control.
For a list of considerations in planning your environment, see Chapter 2 on page 57.
Data Protector allows you to manage multiple cells from a single point.
Figure 10 Single-point management of multiple cells
MoM
Data Protector provides the Manager-of-Managers to manage large environments
with multiple cells. The MoM allows you to group multiple cells into a larger unit,
called a MoM environment that can be managed from a single point, as shown in
Figure 10 on page 46. The MoM enables virtually unlimited growth of your backup
environment. New cells can be added or existing ones split.
A MoM environment does not require a reliable network connection from Data
Protector cells to the central MoM cell, because only the controls are sent over the
long distance connections, however the backups are performed locally within each
Data Protector cell. Nevertheless, this is based on the assumption that each cell has
its own Media Management Database.
46
About backup and Data Protector
Figure 11 Manager-of-Managers environment
Manager-of-Managers provides the following features:
• Centralized licensing repository
This enables simplified license management. This is optional but useful for very
large environments.
• Centralized Media Management Database (CMMDB)
The CMMDB allows you to share devices and media across several cells in a
MoM environment. This makes devices of one cell (using the CMMDB) accessible
to other cells that use the CMMDB. The CMMDB, if used, must reside in the MoM
cell. In this case, a reliable network connection is required between the MoM cell
and the other Data Protector cells. Note that it is optional to centralize the Media
Management Database.
• Sharing libraries
With the CMMDB, you can share high-end devices between cells in the multi-cell
environment. One cell can control the robotics, serving several devices that are
connected to systems in different cells. Even the Disk Agent to Media Agent data
path can go across cell boundaries.
• Enterprise reporting
The Data Protector Manager-of-Managers can generate reports on a single-cell
basis as well as for the entire enterprise environment.
Concepts guide
47
Media management
Data Protector provides you with powerful media management, which lets you easily
and efficiently manage large numbers of media in your environment in the following
ways:
Media management functionality
• Grouping media into logical groups, called media pools, which allows you to
think about large sets of media without having to worry about each medium
individually.
• Data Protector keeps track of all media and the status of each medium, data
protection expiration time, availability of media for backup, and a catalog of
what has been backed up to each medium.
• Fully automated operation. If Data Protector controls enough media in the library
devices, the media management functionality lets you run the backup sessions
without operator intervention.
• Automated media rotation policies that allow media selection for backups to be
performed automatically.
• Recognition and support of barcodes on large library devices and silo devices
with barcode support.
• Recognition, tracking, viewing, and handling of media used by Data Protector
in large library devices and silo devices.
• The possibility of having information about the media in a central place and the
sharing of this information among several Data Protector cells.
• Interactive or automated creation of additional copies of the data on the media.
• Support for media vaulting.
What is a media pool?
Data Protector uses media pools to manage large numbers of media. A media pool
is a logical collection of media of the same physical type with common usage policies
(properties). Usage is based on the data on the media. The structure and quantity of
the pools, as well as which pool contains what type of data on its media, depend
entirely on your preferences.
When a device is configured, a default media pool is specified. This media pool is
used if no other media pool is defined in the backup specification.
48
About backup and Data Protector
Backup devices
Data Protector defines and models each device as a physical device with its own
usage properties, such as the default pool. This device concept is used because it
allows you to easily and flexibly configure devices and use them in conjunction with
backup specifications. The definition of the devices is stored in the Data Protector
Media Management Database.
Figure 12 How backup specifications, devices, and media pools are
related
Figure 12 on page 49 shows the relationship among the backup specification,
devices, and media pools. The devices are referred to in the backup specification.
Each device is linked to a media pool; this media pool can be changed in the backup
specification. For example, backup specification 2 references the pool Dept_X
instead of the default pool.
Data Protector supports various devices. For more information, see the HP Data
Protector product announcements, software notes, and references.
Concepts guide
49
User interfaces
Data Protector provides easy access to all configuration and administration tasks
using the Data Protector GUI on Windows and UNIX platforms. You can use the
original Data Protector GUI (on Windows) or the Data Protector Java GUI (on
Windows and UNIX). Both user interfaces can run simultaneously on the same
computer. Additionally, a command-line interface is available on Windows and UNIX
platforms.
The Data Protector architecture allows you to flexibly install and use the Data Protector
user interface. The user interface does not have to be used from the Cell Manager
system; you can install it on your desktop system. As depicted in Figure
13 on page 50, the user interface also allows you to transparently manage Data
Protector cells with Cell Managers on all supported platforms.
Figure 13 Using the Data Protector user interface
TIP:
In a typical mixed environment, install the Data Protector user interface on several systems
in the environment, thus providing access to Data Protector from several systems.
Data Protector GUI
Both, the original Data Protector GUI, depicted in Figure 14 on page 52, as well as
the Data Protector Java GUI, depicted in Figure 15 on page 52, are easy-to-use,
powerful interfaces providing the following functionalities:
50
About backup and Data Protector
• A Results Tab with all the configuration wizards, properties and lists.
• Easy configuration and management of the backup of online database applications
that run in Windows environments, such as Microsoft SQL Server, Microsoft
Exchange Server, SAP R/3, and Oracle or those that run in the UNIX environments,
such as SAP R/3, Oracle, and Informix Server.
• A comprehensive online Help system called the Help Topics, and context-sensitive
Help called the Help Navigator.
Concepts guide
51
Figure 14 Original Data Protector GUI
Figure 15 Data Protector Java GUI
Data Protector Java GUI
The Data Protector Java GUI is a Java-based graphical user interface with a
52
About backup and Data Protector
client-server architecture. It enables backup management with the same look and feel
as the original Data Protector GUI.
The Java GUI consists of two components: Java GUI Server and Java GUI Client.
Figure 16 on page 53 shows the relationship between these components.
Figure 16 Data Protector Java GUI architecture
The Java GUI Server is installed on the Data Protector Cell Manager system. The Java
GUI Server receives requests from the Java GUI Client, processes them and then
sends the responses back to the Java GUI Client. The communication is done through
Hypertext Transfer Protocol (HTTP) on port 5556.
The Java GUI Client contains only user interface related functionalities and requires
connection to the Java GUI Server to function.
Concepts guide
53
Benefits of Java GUI
The Data Protector Java GUI has the following advantages over the original Data
Protector GUI:
• Portability
The Data Protector Java GUI architecture enables you to install Java GUI Clients
on all platforms that support Java Runtime Environment (JRE).
• Easy firewall configuration
The Java GUI Client uses port 5556 to connect to the Java GUI Server. It is easier
to configure Java GUI in a firewall environment because only one port needs to
be opened. The communication between the Java GUI Client and the Java GUI
Server is done through HTTP, which is also firewall friendlier.
For details, see the Data Protector support matrices under specifications at http:/
/www.hp.com/support/manuals.
• Improved localization and internationalization
Only one installation package is needed for all locales. The Java GUI enables
better display in all locales, since controls are automatically resized to match the
size of the text.
• Non-blocking behavior
The Java GUI Server transmits only data for the current context, which reduces
the network traffic between the Java GUI Server and the Java GUI Client. Due to
its non-blocking behavior, you can work on different contexts while Java GUI
Server processes your requests in the background.
Differences from the Original Data Protector GUI
Due to the different underlying technologies used, there are also some visual and
minor functional differences between the two GUIs. These differences do not have
an important impact on the functionality of Data Protector.
For example, in the Clients context, if you view the Security tab in a client’s properties,
browsing the network behaves differently depending on the GUI used:
• The original Data Protector GUI (on Windows systems only) displays the network
neighborhood of the GUI client.
• The Data Protector Java GUI displays the network neighborhood of the Cell
Manager and not of the GUI client. Browsing is available only with a Windows
Cell Manager; however, it makes no difference if the GUI runs on a Windows
or UNIX system.
54
About backup and Data Protector
Overview of tasks to set up Data Protector
This section provides an overview of global tasks to set up your Data Protector backup
environment. Depending on the size and complexity of your environment, you may
not need to go through all these steps.
1.
Analyze your network and organizational structure. Decide which systems need
to be backed up.
2. Check if there are any special applications and databases which you want to
back up, such as Microsoft Exchange, Oracle, IBM DB2 UDB, SAP R/3, or
others. Data Protector provides specific integrations with these products.
3. Decide on the configuration of your Data Protector cell, such as:
• the system to be your Cell Manager
• systems on which you want to install the user interface
• local backup versus network backup
• systems to control backup devices and libraries
• type of connections, LAN and/or SAN
4. Purchase the required Data Protector licenses for your setup. This way you obtain
the passwords you will need to install.
Alternatively, you can operate Data Protector using an instant-on password.
However, this is valid only for 60 days from the date of installation. For details,
see the HP Data Protector installation and licensing guide.
5. Consider security aspects:
• Analyze security considerations. See the HP Data Protector installation and
licensing guide.
• Consider which user groups you need to configure.
• Enhance security by writing data to media in an encrypted format.
6. Decide how you want to structure your backups:
•
•
•
•
Which media pools do you want to have, and how will they be used?
Which devices will be used, and how?
How many copies of each backup do you want?
How many backup specifications do you need, and how should they be
grouped?
• If you are planning to back up to disk, consider advanced backup strategies
such as synthetic backup and disk staging.
Concepts guide
55
7.
Install and configure your Data Protector environment.
• Install the Data Protector Cell Manager system and use the Data Protector
user interface to distribute Data Protector components to other systems.
• Connect devices (tape drives) to the systems that will control them.
• Configure backup devices.
• Configure media pools and prepare the media.
• Configure backup specifications, including backup of the IDB.
• Configure reports, if needed.
8. Become familiar with tasks such as:
•
•
•
•
•
56
Handling failed backups
Performing restores
Duplicating backed up data and vaulting media
Preparing for disaster recovery
Maintaining the IDB
About backup and Data Protector
2 Planning your backup
strategy
In this chapter
This chapter describes backup strategy planning. It focuses on planning Data Protector
cells, performance, and security, as well as backing up and restoring data. The
chapter also discusses basic backup types, automated backup operation, clustering,
and disaster recovery.
It is organized as follows:
“Backup strategy planning” on page 58
“Planning cells” on page 62
“Understanding and planning performance” on page 67
“Planning security” on page 73
“Clustering” on page 79
“Full and incremental backups” on page 91
“Keeping backed up data and information about the data” on page 99
“Backing up data” on page 102
“Automated or unattended operation” on page 110
“Duplicating backed up data” on page 112
“Restoring data” on page 125
“Disaster recovery” on page 128
Concepts guide
57
Backup strategy planning
Data Protector is simple to configure and administer. However, if you work in a large
environment with diverse client systems and huge amounts of data to back up, plan
in advance. Planning simplifies subsequent configuration steps.
What is backup strategy planning?
Backup strategy planning is a process that includes the following steps:
1.
Defining the requirements and constraints for backups, for example, how often
your data needs to be backed up or whether you need additional copies of the
backed up data on additional media sets.
2. Understanding the factors that influence your backup solution, such as the
sustained data transfer rates of the network and of backup devices. These factors
can affect how you configure Data Protector and the kind of backup -- network
or direct, for example -- that you choose. For instance, if you back up to disk,
you can take advantage of advanced backup strategies such as synthetic backup
and disk staging.
3. Preparing the backup strategy that shows your backup concept and how it is
implemented.
This section provides detailed information on the preceding steps. The rest of this
guide provides important information and considerations that help you plan your
backup solution.
Defining the requirements of a backup strategy
Defining objectives and constraints of your backup strategy includes answering
questions, such as:
• What are your organizational policies regarding backups and restores?
Some organizations already have defined policies on archiving and storing data.
Your backup strategy should comply with these policies.
• What types of data need to be backed up?
List all types of data existing in your network, such as user files, system files, Web
servers, and large relational databases.
• How long is the maximum downtime for recovery?
58
Planning your backup strategy
The allowed downtime has a significant impact on the investments into network
infrastructure and equipment needed for backups. For each type of data, list the
maximum acceptable downtime for recovery, that is, how long specific data can
be unavailable before recovered from a backup. For example, user files may be
restored in two days, while some business data in a large database would need
to be recovered in two hours.
Recovery time consists mainly of the time needed to access the media and the
time required to actually restore data to disks. A full system recovery takes more
time, because some additional steps are required. For more information, see
“Disaster recovery” on page 128.
• How long should specific types of data be kept?
For each type of data, list how long the data must be kept. For example, you may
only need to keep user files for three weeks, while information about company
employees may be kept for five years.
• How should media with backed up data be stored and maintained?
For each type of data, list how long the media with data must be kept in a vault,
a safe, external location, if you use one. For example, user files may not be stored
in a vault at all, while order information may be kept for five years, with
verification of each medium after two years.
• To how many media sets should the data be written during backup?
Consider writing critical data to several media sets during backup to improve the
fault tolerance of such backups, or to enable multi-site vaulting. Object mirroring
increases the time needed for backup.
• How much data needs to be backed up?
List the estimated amount of data to be backed up, for each type of data. This
influences the time needed for backup and helps you to choose the right backup
devices and media for backup.
• What is the projected future growth of the amount of data?
Estimate future growth, for each type of data. This will help you to come up with
backup solutions that will not be quickly outdated. For example, if your company
plans to hire 100 new employees, the amount of users’ data and client systems’
data will grow accordingly.
• How long can a backup take?
Estimate the time needed for each backup. This directly affects the amount of time
data is available for use. User files can be backed up at any time when the users
are not working on them, while some transactional databases may only have a
few hours available for backup.
Concepts guide
59
The time needed for backup depends on the type of backup, full or incremental.
For more information, see “Full and incremental backups” on page 91. Data
Protector also backs up some popular online database applications. For more
information, see the HP Data Protector integration guide.
If you back up to disk, you can take advantage of synthetic backup and disk
staging. These advanced backup strategies significantly reduce the time needed
for backup. For more information, see Chapter 11 on page 257 and Disk
staging on page 118.
When there is a very fast and large disk to be backed up on a slower device,
consider the possibility of backing up one hard disk through multiple concurrent
Disk Agents. Starting multiple Disk Agents on the same disk speeds up the backup
performance considerably.
Also, if there is a large volume of information to be backed up and the available
time to complete the backup is limited, consider the possibility of using direct
backup to take advantage of SAN speed, reduced network traffic, and the lack
of the backup-server bottleneck.
• How often does data need to be backed up?
For each type of data, list how often the data needs to be backed up. For example,
user working files may be backed up on a daily basis, system data on a weekly
basis, and some database transactions twice a day.
Factors influencing your backup strategy
There are a number of factors that influence how your backup strategy is implemented.
Understand these factors before preparing your backup strategy.
•
•
•
•
Your company’s backup and storage policies and requirements.
Your company’s security policies and requirements.
Your physical network configuration.
Computer and human resources available at different sites of your company.
Preparing a backup strategy plan
The result of the planning is a backup strategy that must address the following areas:
• How critical system availability (and backup) is to the company
• The need to keep the backed up data at a remote location in case of a disaster.
• The level of business continuance
This includes the recovery and restore plan for all critical client systems.
• The security of backed up data
60
Planning your backup strategy
The need to guard premises to prevent unauthorized people from entering.
This also includes safeguarding all relevant data against unauthorized access,
using physical access prevention and electronic password protection.
• Types of data that need to be backed up
List the company’s types of data and how you want to combine them in backup
specifications, including the time frames available for backups. The company’s
data can be divided into categories like company business data, company
resource data, project data, and personal data, each with its own specific
requirements.
• Backup policy implementation
• How backups are done and the backup options that you use
This defines the frequency of full and incremental backups. It also defines the
backup options that are used and whether the backups are permanently
protected and the backup media stored at a security company.
• How the client systems are grouped into backup specifications
Consider how best to group backup specifications. This can be done on the
basis of departments, data types, or backup frequency.
• How the backups are scheduled
Consider using the staggered approach, whereby full backups are scheduled
for different clients (backup specifications) on different days to avoid network
load, device load, and time window issues.
• Retaining data on media, and information about backups
Consider protecting data from being overwritten by newer backups for a
specified amount of time. This protection, called data protection, is on a session
basis.
Define the period of time the Catalog Database should store information about
backup versions, the number of backed up files and directories, and messages
stored in the database. For as long as this catalog protection has not expired,
backed up data is easily accessible.
• Device configuration
Determine devices to use for backups, and the client systems they are connected
to. Connect the backup devices to client systems with the largest amount of data,
so that as much data as possible is backed up locally and not via the network.
This increases the backup speed.
If you need to back up large amounts of data:
• Consider using a library device.
Concepts guide
61
• Consider backing up to a disk-based device. Besides other benefits, backup
to disk reduces the time needed for backup and enables the use of advanced
backup strategies such as synthetic backup and disk staging.
• Consider configuring your system for direct backup by attaching a library
device to the SAN through a fibre channel bridge. This is a solution when the
network impairs the backup speed.
• Media management
Determine the type of media to use, how to group the media into media pools,
and how to place objects on the media.
Define how media are used for backup policies.
• Vaulting
Decide whether to store media at a safe place (a vault), where they are kept for
a specific period of time. Consider duplicating backed up data during or after
the backup for this purpose.
• Backup administrators and operators
Determine the rights of users that can administer and operate your storage product.
Planning cells
One of the most important decisions in planning your backup strategy is whether
you want to have a single or multiple cell environment. This section describes the
following:
•
•
•
•
Factors you should
How cells relate to
How cells relate to
How cells relate to
consider when planning cells
a typical network environment
Windows domains
Windows workgroup environments
One cell or multiple cells?
When deciding whether to have a single cell or multiple cells in your environment,
consider the following items:
• Backup administration issues
The use of multiple cells gives you higher administration freedom within each cell.
You can apply completely independent media management policies for each
cell. If you have several administrative groups, you may, for security reasons, not
want a cell to span across these groups. A disadvantage of having multiple cells
62
Planning your backup strategy
is that it can require more administrative work or might even require a separate
administrator for each cell.
• Size of each cell
The size of a Data Protector cell affects backup performance and the ability to
manage the cell. The recommended maximum size for a Data Protector cell is
100 client systems. Cells with more than 200 client systems are less manageable.
• Network considerations
All client systems of a cell should be on the same LAN for maximum performance.
For more information about other network considerations such as your network
configuration, see the sections that follow.
• Geographical location
If the client systems you want to back up are geographically dispersed, it may
be difficult to manage them from a single cell and there may be networking
problems between the client systems. Additionally, the security of data may be
an issue.
• Time Zones
Each cell should be within one time zone.
• Security of data
Data Protector provides cell level based security. All Data Protector administrative
work is done in the context of a single cell: media, backup devices, and backed
up data belong to one cell. Note that Data Protector lets you share devices or
move media between cells, so physical access to media must be limited to
authorized personnel.
• Mixed environments
Data Protector allows you to back up client systems of diverse platforms in a single
cell. However, it may be convenient to group client systems in a cell based on
the platforms. For example, you may have one cell with the Windows client
systems and one with the UNIX client systems. This is especially useful if you have
separate administrators and policies for the UNIX and Windows environments.
• Departments and sites
You can group each department or site in a separate cell. For example, you may
have one cell for the accounting, one for the IT, and one for the manufacturing
department. Even if you choose to have several cells, Data Protector allows you
to easily configure common policies among the cells.
Concepts guide
63
Installing and maintaining client systems
If you have several UNIX and Windows client systems, an efficient mechanism for
the installation of Data Protector becomes important. Local installation on every client
is not feasible in large environments.
Installation Servers and the Cell Manager
The main system in a Data Protector cell is the Cell Manager. To conveniently distribute
(push) Data Protector components to client systems from a central location, a system
holding the Data Protector software repository is needed. This system is called the
Data Protector Installation Server. The Cell Manager is by default also an Installation
Server.
Each time you perform a remote installation, you access the Installation Server. The
advantage of using Installation Servers is that the time required for remote installation,
update, upgrade, and removal of Data Protector software is greatly reduced,
especially in enterprise environments.
There are certain hardware and software requirements that need to be met by
Installation Servers and Cell Managers before you start installing the software. A
dedicated port, generally port 5555, needs to be available throughout the cell. For
details, see the HP Data Protector installation and licensing guide.
The Cell Manager and Installation Servers are installed directly from the CD. After
you have installed the Cell Manager and Installation Servers you can then install the
components on various client systems using the Data Protector Installation GUI.
When you install Data Protector for the first time, it runs with an instant-on license,
valid for 60 days, that lets you use Data Protector before you acquire a permanent
license. During this time, purchase any required licenses.
Also during this time, you should set up and configure your Data Protector environment
and request your permanent license. To request a permanent password string, you
need to know which client systems belong in which Data Protector cell, the number
of devices connected to the client systems, and whether you need to use any of the
Data Protector integrations.
Creating cells in the UNIX environment
Creating cells in the UNIX environment is easy. Based on the considerations given
in this manual, decide which client systems you want to add to the cell and define
the Cell Manager system. During installation, root access is required to every client
system. An important prerequisite is to have a clean node name resolving setup, such
64
Planning your backup strategy
that each client system is accessible from every other client system using the same
fully qualified node name.
Creating cells in the Windows environment
Due to the different possible configurations (domain versus workgroup), the various
levels of support for Windows Administrators may have some impact on the setup
of Data Protector during installation. An important prerequisite is to have a clean
node name resolving setup, so that each client system is accessible from every other
client system using the same fully qualified node name.
Windows domains
A Windows domain can easily be mapped to a Data Protector cell. In a single
Windows domain, use a one-to-one mapping if the size of the domain does not
exceed the recommended size of the Data Protector cell. Otherwise, split it into two
or more cells and manage these cells using the Data Protector Manager-of-Managers.
Mapping a Data Protector cell into a Windows domain
Mapping a Data Protector cell into a Windows domain also eases administration
within Data Protector itself. To ease administration, distribute the software such that
all the client systems can be installed using a central Windows account in a domain
organization. Other operations, however, are not limited to a Windows domain
organization since all operations and security verifications are performed by the
Data Protector internal protocol and not by the Windows Security.
In general, there are no limitations on how and where Data Protector can be installed.
However, because of the structure of Windows and the most common configurations
that are domain environments, some operations are easier when Data Protector is
mapped to a single domain or a multiple domain model, where one of the domains
is a master domain, to allow a single user to manage all the client systems within the
environment (Software Distribution and User Configuration).
In a multiple cell environment with a Manager-of-Managers, this issue is more
significant because all the cells that are configured require a central administrator
that has access to the entire backup environment. When a single domain or multiple
domains with a master domain are configured, the same global master domain user
can be the administrator of all the cells and the Manager-of-Managers environment.
If multiple independent domains are used, you need to configure multiple users to
administer the environment.
Concepts guide
65
Windows workgroups
Some of the configuration tasks require more steps in some cases, because there are
no global users as in a domain. Software distribution requires a unique logon for
every client system that you install the software on. This means that to install 100
client systems in a workgroup environment, you are required to enter 100 logons.
In such cases, use a domain environment, since installation and many other non-Data
Protector related administration tasks are much easier for a large-scale environment.
Using MoM in such an environment requires you to configure the administrators
separately for each cell, to manage the MoM environment from any of the cells.
Again, Data Protector is not limited to a Windows domain organization. However,
it takes advantage of and simplifies the administration procedures in the areas where
user authentication is required (Installation, User Management).
Creating cells in a mixed environment
In a mixed environment, take into account the factors described in “Creating cells in
the UNIX environment” on page 64. The more the environment is broken into multiple
domains and multiple workgroups, the more accounts and steps need to be considered
to distribute the software and to prepare the environment for administration.
Geographically remote cells
Data Protector allows you to easily administer geographically remote cells. For more
information, see “Splitting an environment into multiple cells” on page 45.
Considerations for geographically remote cells
When configuring geographically remote cells, remember the following:
• Data is not sent over a WAN.
The devices and the client systems that you are backing up are configured locally.
• The cells are configured in a MoM.
To manage geographically remote cells centrally, you need to configure the cells
in a MoM environment.
• Consider user configurations.
All the considerations that are mentioned regarding single domain, multiple
domain, and workgroup configurations need to be taken into account.
66
Planning your backup strategy
You can configure a single cell over geographically remote locations. In this case,
you need to ensure that data transfer from each client system to the corresponding
device is not done over a WAN. Because a WAN network is not a stable connection,
it is possible that connections are lost.
MoM environment
A MoM environment does not require a reliable network connection from cells to the
central MoM cell, because only controls are sent over the long distance connections,
and backups are performed locally within each Data Protector cell. However, this is
based on the assumption that each cell has its own media management database.
In such a case, use the Data Protector Reconnect broken connections backup option
so that connections are reestablished after they are broken.
Understanding and planning performance
In business-critical environments, it is a key requirement to minimize the time needed
for data recovery in case of a corrupt database or a disk crash. Therefore,
understanding and planning backup performance is extremely important. Optimizing
the time required for the backup of a number of client systems and large databases
that are all connected on different networks and different platforms is a challenging
task.
The following sections give an overview of the most common backup performance
factors. Due to the high number of variables it is not possible to give distinct
recommendations that fit all user requirements.
The infrastructure
The infrastructure has a high impact on the backup and restore performance. The
most important aspects are the parallelism of data paths and the use of high-speed
equipment.
Network versus local backups
Sending data over a network introduces additional overhead, as the network becomes
a component of performance consideration. Data Protector handles the data stream
differently for the following cases:
• Network datastream: Disk to Memory of Source System to Network to Memory
of Destination System to Device
• Local datastream: Disk to Memory to Device
Concepts guide
67
To maximize performance, use local backup configurations for high volume
datastreams.
Network or server versus direct backups
Sending data over a network and through a server introduces additional overhead,
as the network and the server become performance considerations. Data Protector
handles the datastream differently for the following cases:
• Network datastream: Disk to Memory of Source System to Network to Memory
of Destination System to Device
• Direct datastream: Disk to Device
To maximize performance, use direct backup configurations for high volume
datastreams.
Devices
Device performance
Device types and models impact performance because of the sustained speed at
which devices can write data to a tape (or read data from it).
Data transfer rates also depend on the use of hardware compression. The achievable
compression ratio depends on the nature of the data being backed up. In most cases,
using high speed devices with hardware compression improves performance. This
is true, however, only if the devices stream.
At the start and at the end of a backup session backup devices require some time
for operations such as rewinding media and mount or unmount media.
Libraries offer additional advantages because of their fast and automated access to
a large number of media. At backup time, loading new or reusable media is needed,
and at restore time the media which contain the data to be restored need to be
accessed quickly.
Data in disk based-devices is accessed faster than that in conventional devices, as
there is no need to load and unload media. This reduces the amount of time spent
for backup and restore. Additionally, disk-based devices enable the use of advanced
backup strategies such as synthetic backup and disk staging, which also reduce the
backup and restore time.
68
Planning your backup strategy
High performance hardware other than devices
Performance of computer systems
The speed of computer systems themselves directly impacts performance. The systems
are loaded during backups by reading the disks, handling software compression,
and so on.
The disk read data rate and CPU usage are important performance criteria for the
systems themselves, in addition to I/O performance and network types.
Advanced high performance configuration
Data Protector Zero Downtime Backup solution provides a means of shortening the
application downtime or backup mode time and reduces the network overhead by
using locally attached backup devices instead of network backup devices. The
application downtime or backup mode time is limited to the time needed to create
a replica of data, which is then backed up on a backup system to a locally attached
device.
For more information on Zero Downtime Backup, see the HP Data Protector zero
downtime backup concepts guide.
Using hardware in parallel
Using several datapaths in parallel is a fundamental and efficient method to improve
performance. This includes the network infrastructure. Parallelism boosts performance
in the following situations:
When to use parallelism
• Several client systems can be backed up locally, that is, with the disk(s) and the
related devices connected on the same client system.
• Several client systems can be backed up over the network. Here the network
traffic routing needs to be such that datapaths do not overlap, otherwise the
performance is reduced.
• Several objects (disks) can be backed up to one or several (tape) devices.
• An object (disk or files) can be directly backed up to several (tape) devices using
several XCOPY engines.
• Several dedicated network links between certain client systems can be used. For
example, if system_A has 6 objects (disks) to be backed up, and system_B has
Concepts guide
69
3 fast tape devices, consider using 3 dedicated network links between system_A
and system_B.
• Load Balancing
Using this Data Protector feature, Data Protector dynamically determines which
object (disk) should be backed up to which device. Enable this feature, especially
to back up a large number of filesystems in a dynamic environment. For more
information, see “How load balancing works” on page 156.
Note that you cannot predict to which media a particular object is written.
Configuring backups and restores
Any given infrastructure must be used efficiently to maximize performance. Data
Protector offers high flexibility to adapt to the environment and the desired way to
operate backups and restores.
Software compression
Software compression is done by the client CPU when reading data from a disk. This
reduces the data that is sent over the network, but it requires significant CPU resources
from the client.
By default, software compression is disabled. Use software compression only for
backups of many machines over a slow network, where data can be compressed
before sending it over the network. If software compression is used, hardware
compression should be disabled since trying to compress data twice actually expands
the data.
Hardware compression
Hardware compression is done by a device that receives original data from a Drive
Server and writes it to media in the compressed mode. Hardware compression
increases the speed at which a tape drive can receive data, because less data is
written to the tape.
By default, hardware compression is enabled. On HP-UX systems, enable hardware
compression by selecting a hardware compression device file. On Windows systems,
enable hardware compression during device configuration. Use hardware compression
with caution, because media written in compressed mode cannot be read using a
device in uncompressed mode and vice-versa.
70
Planning your backup strategy
Full and incremental backups
A basic approach to improve performance is to reduce the amount of data to back
up. Carefully plan your full and incremental backups. Note that you may not need
to perform all the full backups of all the client systems at the same time.
If you back up to disk, you can use advanced backup strategies such as synthetic
backup and disk staging.
Disk image versus filesystem backups
It used to be more efficient to back up disk images (raw volumes) rather than
filesystems. This is still true in some cases, such as heavily-loaded systems or disks
containing large numbers of small files. The general recommendation is to use
filesystem backups.
Object distribution to media
The following are examples of object/media backup configurations provided by
Data Protector:
• One object (disk) goes to one medium
The advantage is a known fixed relationship between an object and a medium
where the object resides. This can be of benefit for the restore process, since only
one medium needs to be accessed.
The disadvantage in a network backup configuration is the likely performance
limitation due to the network, causing the device not to stream.
• Many objects go to a few media, each medium has data from several objects,
one object goes to one device
The advantage here is the flexibility of datastreams at backup time, helping to
optimize performance, especially in a network configuration.
The strategy is based on the assumption that the devices receive enough data to
be able to stream, since each device receives data from several sources
concurrently.
The disadvantage is that data (from other objects) has to be skipped during the
restore of a single object. Additionally, there is no precise prediction as to which
medium will receive data from which object.
For more information on device streaming and backup concurrency, see “Device
streaming and concurrency” on page 156.
Concepts guide
71
Disk performance
All data that Data Protector backs up resides on disks in your systems. Therefore, the
performance of disks directly influences backup performance. A disk is essentially a
sequential device, that is, you can read or write to it, but not both at the same time.
Also, you can read or write one stream of data at a time. Data Protector backs up
filesystems sequentially, to reduce disk head movements. It also restores files
sequentially.
Sometimes this is not visible because the operating system stores most frequently used
data in a cache memory.
Disk fragmentation
Data on a disk is not kept in the logical order that you see when browsing the files
and directories, but is fragmented in small blocks all over the physical disk. Therefore,
to read or write a file, a disk head must move around the whole disk area. Note that
this differs from one operating system to another.
TIP:
Backups are most efficient for large files with little fragmentation.
Compression
If data is compressed on a disk, the Windows operating system first decompresses
the data before sending it across the network. This reduces the backup speed and
uses CPU resources.
Disk image backups
Data Protector also allows you to back up UNIX disks as disk images. With a disk
image backup, a complete image of the disk is backed up without tracking the
filesystem structure. The disk head moves linearly across the surface. Thus a disk
image backup can be considerably faster than a filesystem backup.
Disk Agent performance on Windows systems
Disk Agent performance of Windows filesystem backup can be improved by enabling
asynchronous reading. Asynchronous reading improves performance of the Disk
Agent when backing up data on disk arrays, especially if large files are backed up.
It is recommended to perform test backups to establish if asynchronous reading will
72
Planning your backup strategy
improve performance in your specific environment and determine the optimum
asynchronous reading settings.
SAN performance
If large volumes of data need to be backed up in one session, the time needed to
transfer the data becomes significant. This consists of the time required to move the
data over a connection (LAN, local, or SAN) to a backup device.
Online database application performance
When you back up databases and applications, such as Oracle, SAP R/3, Sybase,
and Informix Server, the performance of the backups also depends on the applications.
Database online backups are provided so that backups can occur while the database
application remains online. This helps to maximize database up time but may impact
application performance. Data Protector integrates with all popular online database
applications to optimize backup performance.
For more information on how Data Protector integrates with various applications and
for tips on how to improve backup performance, see the HP Data Protector integration
guide.
Also see the documentation that comes with your online database application for
more information on how to improve backup performance.
Planning security
When you plan your backup environment, consider security. A well thought out,
implemented, and updated security plan prevents the unauthorized access, duplication,
or modification of data.
What is security?
Security in the backup context typically refers to:
•
•
•
•
Who
Who
Who
Who
can
can
can
can
administer or operate a backup application (Data Protector).
physically access client systems and backup media.
restore data.
view information about backed up data.
Data Protector provides security solutions on all these levels.
Concepts guide
73
Data Protector security features
The following features allow and restrict access to Data Protector and the backed up
data. The items in this list are described in detail in the following sections.
•
•
•
•
•
Cells
Data Protector user accounts
Data Protector user groups
Data Protector user rights
Visibility and access to backed up data
Cells
Starting sessions
Data Protector security is based on cells. Backup and restore sessions can only be
started from the Cell Manager unless you have the Data Protector
Manager-of-Managers functionality. This ensures that users from other cells cannot
back up and restore data from systems in your local cell.
Access from a specific Cell Manager
Additionally, Data Protector allows you to explicitly configure from which Cell
Manager a client system can be accessed, that is, configuring a trusted peer.
Restrict pre- and post-execution
For security reasons, various levels of restrictions can be configured for pre-exec and
post-exec scripts. These optional scripts allow a client system to be prepared for the
backup by, for example, shutting down an application to obtain a consistent backup.
Data Protector users accounts
Anyone using any Data Protector functionality, administering Data Protector, or
restoring personal data, must have a Data Protector user account. This restricts
unauthorized access to Data Protector and backed up data.
Who defines user accounts?
An administrator creates this account specifying a user login name, systems from
which a user can log in, and the Data Protector user group membership that defines
the user rights.
74
Planning your backup strategy
When is the account checked?
When a user starts the Data Protector user interface, Data Protector checks user rights.
User rights are also checked when specific tasks are performed by a user.
For more information, see Chapter 4 on page 183.
Data Protector user groups
What are user groups?
When a new user account is created, the user becomes a member of the specified
user group. Each user group contains defined Data Protector user rights. All the
members of the group have the user rights set for the group.
Why use user groups?
Data Protector user groups simplify user configuration. The administrator groups users
according to the access they need. For example, an end-user group could allow
members to restore personal data to a local system only, while the operator group
allows the starting and monitoring of backups, but not the creating of backups.
For more information, see Chapter 4 on page 183.
Data Protector user rights
What are user rights?
Data Protector user rights define the actions that a user can perform with Data
Protector. They are applied on the Data Protector user group level and not to each
user individually. Users added to a user group automatically gain the user rights
assigned to this user group.
Why use user rights?
Data Protector provides flexible user and user group functionality, which allows the
administrator to selectively define who can use a particular Data Protector functionality.
It is important to carefully apply the Data Protector user rights: backing up and
restoring data is essentially the same as copying data.
For more information, see Chapter 4 on page 183.
Concepts guide
75
Visibility of backed up data
Backing up data means creating a new copy. Therefore, when you deal with
confidential information, it is important to restrict access to both the original data
and to the backup copy itself.
Hiding data from other users
When configuring a backup, you can decide whether during a restore the data is
visible to everyone (public) or only to the owner of the backup (private). For more
information about backup owners, see “What is backup ownership?” on page 78.
Data encryption
Open systems and public networking make data security in large enterprises essential.
Data Protector lets you encrypt backed-up data so that it becomes protected from
others. Data Protector offers two data encryption techniques: software-based and
drive-based.
Data Protector software encryption, referred to as AES 256-bit encryption, is based
on the AES-CTR (Advanced Encryption Standard in Counter Mode) encryption
algorithm that uses random keys of 256-bit length. The same key is used for both
encryption and decryption. With AES 256-bit encryption, data is encrypted before
it is transferred over a network and before it is written to media.
Data Protector drive-based encryption uses the encryption functionality of the drive.
The actual implementation and encryption strength depend on the drive's firmware.
Data Protector only turns on the feature and manages encryption keys.
The key management functionality is provided by the Key Management Server (KMS),
which is located on the Cell Manager. All encryption keys are stored centrally in the
key store file on the Cell Manager and administered by the KMS.
You can encrypt all or selected objects in a backup specification and also combine
encrypted and unencrypted sessions on the same medium.
In addition to the encryption functionality, Data Protector also offers the encoding
functionality that uses a keyless, built-in algorithm for this purpose.
How Data Protector AES 256-bit encryption works
The Backup Session Manager (BSM) reads the backup specification in which the AES
256-bit encryption option is selected and requests an active encryption key from the
Key Management Server (KMS). The key is transferred to the Disk Agent (DA), which
76
Planning your backup strategy
encrypts the data. Thus the backed up data is encrypted before it is transferred over
the network and written to media.
Figure 17 on page 77 shows a basic interaction during an encrypted backup session
with the AES 256-bit encryption option selected.
Figure 17 Backup session with AES 256-bit encryption
How Data Protector drive-based encryption works
The BSM reads the backup specification in which the Drive-based encryption option
is selected and requests an active encryption key from the KMS. The key is transferred
to the Media Agent (MA), which configures the drive for encryption and sets the
encryption key into the drive. The drive encrypts both the data and the meta-data
that is written to the medium.
In an object copy or object consolidation operation from an encrypted backup, the
data is decrypted by the source drives, transferred over the network and encrypted
by the destination drives.
Concepts guide
77
If a source medium involved in an automatic media copy session stores encrypted
as well as non-encrypted data, all data written to the corresponding target medium
will be either encrypted or non-encrypted, depending on current settings for
drive-based encryption.
Figure 18 on page 78 shows a basic interaction during an encrypted backup session
with the Drive-based encryption option selected.
Figure 18 Backup session with drive-based encryption
Restore from encrypted backups
No additional encryption related preparations are needed for restore of encrypted
backups, as Data Protector automatically obtains the appropriate decryption keys.
What is backup ownership?
Who owns a backup session?
Each backup session and all the data backed up within it is assigned an owner. The
owner can be the user who starts an interactive backup, the account under which
78
Planning your backup strategy
the CRS process is running, or the user specified as the owner in the backup
specification options. For instructions on how to specify a backup owner, see the
online Help index: "ownership".
Backup ownership and restore
Backup ownership affects the ability of users to see and restore data. Unless the
object is marked as Public, only the owner of the media set or an administrator can
see the data saved in the media set. The right to see and restore private objects can
be granted to groups other than admin as well. For instructions on who can see and
restore a private object and how this can be applied, see the online Help index:
"ownership".
Clustering
Cluster concepts
What is a cluster?
A cluster is a group of two or more computers that appear on the network as a single
system. This group of computers is managed as a single system and is designed to:
• Ensure that mission-critical applications and resources are as highly-available as
possible
• Tolerate component failures
• Support either the addition or subtraction of components
For clustering purposes, Data Protector integrates with Microsoft Cluster Server for
Windows Server, with MC/Service Guard for HP-UX, with Veritas Cluster for Solaris
and with Novell NetWare Cluster Services. For a list of supported clusters, see the
HP Data Protector product announcements, software notes, and references.
Concepts guide
79
Figure 19 Typical cluster
Components:
• Cluster nodes (two or more)
• Local disks
• Shared disks (shared between nodes)
Cluster nodes
Cluster nodes are computers that compose a cluster. They are physically connected
to one or more shared disks.
Shared disks
The shared disks volumes (MSCS, Novell NetWare Cluster Services) or shared volume
groups (MC/SG, Veritas Cluster) contain mission-critical application data as well as
specific cluster data needed to run the cluster. In MSCS clusters, a shared disk is
exclusively active on only one cluster node at a time.
Cluster network
A cluster network is a private network that connects all cluster nodes. It transfers the
internal cluster data called heartbeat of the cluster. The heartbeat is a data packet
with a time stamp that is distributed among all cluster nodes. Each cluster node
compares this packet and determines the cluster node that is still operational so that
80
Planning your backup strategy
you can make an appropriate determination of the ownership of the package
(MC/SG, Veritas Cluster) or group (MSCS).
What is a package or group?
A package (MC/SG, Veritas Cluster) or a group (MSCS) is a collection of resources
that are needed to run a specific cluster-aware application. Each cluster-aware
application declares its own critical resources. The following resources must be
defined in each group or package:
•
•
•
•
•
Shared disk volumes (MSCS, Novell NetWare Cluster Services)
Shared volume groups (MC/SG, Veritas Cluster)
Network IP names
Network IP addresses
Cluster-aware application services
What is a virtual server?
Disk volumes and volume groups represent shared physical disks. A network IP name
and a network IP address are resources that define a virtual server of a cluster-aware
application. Its IP name and address are cached by the cluster software and mapped
to the cluster node where the specific package or group is currently running. Since
the group or package can switch from one node to another, the virtual server can
reside on different machines in different time frames.
What is a failover?
Each package or group has its own “preferred” node where it normally runs. Such
a node is called a primary node. A package or group can be moved to another
cluster node (one of the secondary nodes). The process of transferring a package or
group from the primary cluster node to the secondary is called failover or switchover.
The secondary node accepts the package or group in case of failure of the primary
node. A failover can occur for many different reasons:
• Software failures on the primary node
• Hardware failures on the primary node
• The administrator intentionally transfers the ownership because of maintenance
on the primary node
In a cluster environment there can be more than one secondary node but only one
can be the primary.
Concepts guide
81
A cluster-aware Data Protector Cell Manager that is responsible for running the IDB
and managing backup and restore operations has many major benefits over
non-cluster versions:
High availability of the Data Protector Cell Manager
All Cell Manager operations are always available since Data Protector services are
defined as cluster resources within the cluster and are automatically restarted when
a failover occurs.
Automatic restart of backups
Data Protector backup specifications that define the backup procedure can easily be
configured to be restarted in case of a failover of the Data Protector Cell Manager.
Restart parameters can be defined using the Data Protector GUI.
Load balancing at failover
A special command-line utility is provided for operations that allow backup sessions
to be aborted in case applications other than Data Protector perform a failover. The
Data Protector Cell Manager allows you to define what should happen in such
situations. If the backup is less important than the application, Data Protector can
abort running sessions. If the backup is more important or is just ending, Data Protector
can continue the sessions. For more information on how to define the criteria, see
the online Help index: "cluster, managing backups".
Cluster support
The Data Protector cluster support means the following:
• The Data Protector Cell Manager is installed in a cluster. Such a Cell Manager
is fault tolerant and can restart operations in the cell automatically after the failover.
NOTE:
If the Cell Manager is installed in the cluster, its cluster critical resources need to
be configured in the same cluster package or group as the application being
backed up, in order to automatically restart failed backup sessions that failed
due to a failover. Otherwise, the failed backup sessions must be restarted
manually.
82
Planning your backup strategy
• The Data Protector client is installed in a cluster. The Cell Manager (if not installed
in the cluster) in such a case is not fault tolerant; the operations in the cell must
be restarted manually.
The behavior of the Cell Manager after the failover is configurable as far as the
backup session (failed due to the failover) is concerned - the failed session can be:
• restarted as a whole
• restarted only for the failed objects
• not restarted at all
For more information on backup session behavior options on failover of the Data
Protector Cell Manager, see the online Help index: "cluster, backup specification
options".
Example cluster environments
This section gives three example cluster configurations.
Cell Manager installed outside a cluster
In the environment depicted below:
• The Cell Manager installed outside a cluster
• A backup device connected to the Cell Manager or one of the (non-clustered)
clients
Concepts guide
83
Figure 20 Cell Manager installed outside a cluster
When creating a backup specification, you can see three or more systems that can
be backed up in the cluster.
• Physical Node A
• Physical Node B
• Virtual Server
Virtual server backup
If you select the virtual server in the backup specification, then the backup session
will back up the selected active virtual host/server regardless of the physical node
the package or group is currently running on.
For more information on how to define these options, see the online Help index:
"cluster, backup specification options".
84
Planning your backup strategy
The following is the expected backup behavior under this configuration.
Table 3 Backup behavior
Condition
Result
Failover of the node before
a backup starts
Successful backup
Failover of the node during
backup activity
Filesystem/disk image backup: The backup session fails. The
completed objects from the session can be used for restore,
the failed (running and pending) objects need to be backed
up again by restarting the session manually.
Application backup: The backup session fails. The session
needs to be restarted manually.
Cell Manager installed outside a cluster, devices connected to the cluster nodes
In the environment depicted below:
• The Cell Manager installed outside a cluster
• Backup devices connected to the nodes in the cluster
Concepts guide
85
Figure 21 Cell Manager installed outside a cluster, devices connected
to the cluster nodes
When creating a backup specification, you can see three or more systems that can
be backed up in the cluster.
• Physical Node A
• Physical Node B
• Virtual Server
Virtual server backup
If you select the virtual server in the backup specification, then the backup session
will back up the selected active virtual host/server regardless of the physical node
the package or group is currently running on.
86
Planning your backup strategy
NOTE:
The difference with the previous example is that each of the cluster nodes has a Data
Protector Media Agent installed. Additionally, you need to use the Data Protector load
balancing functionality. Include both devices in the backup specification. With load
balancing set to min=1 and max=1, Data Protector will only use the first available
device.
The following is the expected backup behavior under this configuration.
Table 4 Backup behavior
Condition
Result
Failover of the node before a backup starts
Successful backup due to automatic device
switching (load balancing)
Failover of the node during backup activity
Filesystem/disk image backup: The backup
session fails. The completed objects from the
session can be used for restore, the failed
(running and pending) objects need to be
backed up again by restarting the session
manually.
Application backup: The backup session fails.
The session needs to be restarted manually.
IMPORTANT:
If a failover during backup activity occurs in such a configuration, the MA may not be
able to properly abort the session. This results in the corruption of the medium.
Cell Manager installed in a cluster, devices connected to the cluster nodes
In the environment depicted below:
• The Cell Manager installed in a cluster.
With regard to the Data Protector application integrations, there are two possible
ways of configuring Data Protector and an application in such a configuration:
• The Data Protector Cell Manager is configured to run (both during the normal
operation and during the failover) on the same node as the application - the
Concepts guide
87
Data Protector cluster critical resources are defined in the same package
(MC/ServiceGuard) or group (Microsoft Cluster Server) as the application
cluster critical resources.
IMPORTANT:
Only in such a configuration, it is possible to define the automated action
concerning the Data Protector sessions aborted during the failover.
• The Data Protector Cell Manager is configured to run (both during the normal
operation and during the failover) on nodes other than the application node
- the Data Protector cluster critical resources are defined in some other package
(MC/ServiceGuard) or group (Microsoft Cluster Server) as the application
cluster critical resources.
• Backup device(s) connected to the cluster shared Fibre Channel bus via an
FC/SCSI MUX.
88
Planning your backup strategy
Figure 22 Cell Manager installed in the cluster, devices connected to
cluster nodes
When creating a backup specification, you can see three or more systems that can
be backed up in the cluster.
• Physical Node A
• Physical Node B
• Virtual Server
Virtual server backup
If you select the virtual server in the backup specification, then the backup session
will back up the selected active virtual host/server regardless of the physical node
the package or group is currently running on.
Concepts guide
89
NOTE:
Clusters do not support a SCSI bus with shared tapes. To bring high availability also to
Media Agents, the Fibre Channel technology can be used as an interface to the device.
The device itself is not highly-available in this configuration.
This configuration allows the following features:
• Customizable automatic restart of backups in case of failover of the Cell Manager.
The Data Protector backup specifications can be configured to be restarted in
case of failover of the Cell Manager. Restart parameters can be defined using
the Data Protector GUI.
• System load control at failover.
Sophisticated control is provided to define Data Protector behavior at failover. A
special command, omniclus, is provided for this purpose. The Cell Manager
allows the administrator to define what should happen in such situations.
• If the backup is less important than the application that just switched to the
backup system, Data Protector can abort the running sessions.
• If the backup is more important or it is just pending, Data Protector continues
the sessions.
The following is the expected backup behavior under this configuration.
Table 5 Backup behavior
Condition
Result
Failover before a
backup starts
Successful backup
Failover of the
application and the Cell
Manager during backup
activity (Cell Manager
runs on the same node
as the application).
Filesystem/disk image backup The
backup session fails. The completed
objects from the session can be used
for restore, the failed (running and
pending) objects are backed up again
by restarting the session automatically.
Application backup The backup
session fails. The session is restarted
automatically.
90
Planning your backup strategy
IMPORTANT
To restart the session,
the appropriate Data
Protector option must be
selected. For information
on defining all possible
Data Protector actions in
case of failover of the
Cell Manager, see the
online Help index:
"cluster, managing
backups".
Condition
Result
Failover of the
application during
backup activity without
Cell Manager failover
(Cell Manager runs on
other node than the
application).
Filesystem/disk image backup The backup session fails at failover
of the node where the filesystem is installed. The completed
objects from the session can be used for restore, the failed
(running and pending) objects need to be backed up again by
restarting the session manually.
Application backup The backup session fails. The session needs
to be restarted manually.
IMPORTANT:
If a failover during backup activity occurs in such a configuration, the MA may not be
able to properly abort the session. This results in the corruption of the medium.
Additionally, the Data Protector cluster Cell Manager/client can be integrated with
the EMC Symmetrix or HP StorageWorks Disk Array XP environment, producing a
very highly-available backup environment. For more information, see the HP Data
Protector zero downtime backup administrator's guide.
Full and incremental backups
Data Protector provides two basic types of filesystem backups: full and incremental.
A full backup saves all the files selected for backup in a filesystem. An incremental
backup saves only those files that have changed since the last full or incremental
backup. This section gives hints on how to choose the backup type and how this
influences your backup strategy.
Table 6 Comparison of full and incremental backup
Full backup
Incremental backup
Resources
Takes more time to complete than
incremental backup and requires
more media space.
Backs up only changes made since
a previous backup, which requires
less time and media space.
Device handling
If you use a standalone device with
a single drive, you need to change
the media manually if a backup
does not fit on a single medium.
It is less likely that the backup will
require additional media.
Concepts guide
91
Full backup
Incremental backup
Restore
Enables simple and quick restore.
A restore takes more time because
of the number of media needed.
IDB impact
Occupies more space in the IDB.
Occupies less space in the IDB.
Data Protector can also make incremental backups of online database applications.
These vary from application to application. On Sybase, for instance, this type of
backup is referred to as a transaction backup (a backup of transaction logs modified
since the last backup).
Note that the incremental backup concept is not related to the log level concept,
which defines the amount of information written to the IDB.
NOTE:
A number of additional backup types (such as direct backup, split mirror backup, snapshot
backup, and data mover backup) are available with Data Protector application
integrations. For more information, see the respective HP Data Protector integration
guides for more information.
Full backups
Full backups always back up all selected objects, even if there are no changes since
the previous backup.
Synthetic backup
Synthetic backup is an advanced backup solution that eliminates the need to run
regular full backups. Instead, incremental backups are run, and subsequently merged
with the full backup into a new, synthetic full backup. For more information, see
Chapter 11 on page 257.
Incremental backups
Incremental backups back up changes from a previous still protected (full or
incremental) backup. A full backup of an object (with identical client name, mount
point, and description) must exist before an incremental backup of this object is
possible.
92
Planning your backup strategy
Incremental backups depend on the last full backup. If you specify an incremental
backup and there is no protected full backup, a full backup is performed instead.
Conventional incremental backup
Before running an incremental backup of a specific backup object, Data Protector
compares the trees in the backup object with the trees in the valid restore chain of
this object. If the trees do not match (for example, an additional directory in the
backup object was selected for backup since the last backup or multiple backup
specifications with the same backup object and different trees exist), a full backup
is automatically performed. This ensures that all files that have changed since the last
relevant backup are backed up.
With conventional incremental backup, the main criterion for determining whether
a file has changed or not since a previous backup is the file's modification time.
However, if a file has been renamed, moved to a new location, or if some of its
attributes have changed, its modification time does not change. Consequently, the
file is not always backed up in a conventional incremental backup. Such files are
backed up in the next full backup.
Enhanced incremental backup
Enhanced incremental backup reliably detects and backs up also renamed and moved
files, as well as files with changes in attributes.
Enhanced incremental backup also eliminates unnecessary full backups of an entire
backup object when some of the trees selected for backup change. For example, if
an additional directory is selected for backup since the last backup, a full backup of
this directory (tree) is performed, whereas the backup of the rest is incremental.
Using enhanced incremental backup is a prerequisite for synthetic backup.
You can also perform enhanced incremental backup using the Windows NTFS Change
Log Provider. Change Log Provider queries the Windows Change Journal for a list
of changed files rather than performs a time-consuming file tree walk. As the Change
Journal detects and records all changes made to the files and directories on an NTFS
volume, Data Protector can use it as a tracking mechanism to generate a list of files
modified since the last full backup. This improves the incremental backup speed,
especially in environments containing millions of files only a few of which have
changed, and allows to eliminate unnecessary full backups.
Types of incremental backups
Data Protector provides incremental backups of different types:
Concepts guide
93
94
Incr
A simple incremental backup, shown in Figure
23 on page 95, is based on the last backup that is still
protected, which can be a full backup or an incremental
backup.
Incr1-9
A leveled incremental backup, shown in Figure
24 on page 95, depends on the last backup of the next lower
level that is still protected. For example, an Incr1 backup
saves all changes since the last full backup, while an Incr5
backup saves all changes since the last Incr4 backup. An
Incr1-9 backup never references an existing Incr backup.
Planning your backup strategy
Figure 23 Incremental backups
Figure 24 Leveled incremental backups
Table 7 on page 95 shows the relative referencing of backup runs with various
backup types. See the text following the table for a full explanation.
Table 7 Relative referencing of backup runs
1
Full
<----
Incr1
2
Full
<----
<----
<----
Incr2
3
Full
<----
Incr1
<----
Incr2
4
Full
<----
Incr
5
Full
<----
Incr1
<----
Incr
6
Full
<----
Incr1
<----
Incr2
<----
Incr
7
Full
<----
Incr1
<----
Incr
<----
Incr
Concepts guide
95
8
Full
<----
Incr1
<----
Incr3
9
Full
<----
Incr1
<----
Incr2
<----
Incr3
10
Full
<----
<----
<----
Incr2
<----
Incr3
11
Full
<----
<----
<----
<----
<----
Incr3
How to read Table 7 on page 95
• The rows in Table 7 on page 95 are independent of each other and show different
situations.
• The age of the backups increases from right to left, so that the far left is the oldest
and the far right is the most recent backup.
• The full and IncrX represent still protected objects of the same owner. Any existing
IncrX that is not protected can be used for restore, but is not considered for
referencing on subsequent backup runs.
Examples
• In the second row, there is a full, still protected backup and an Incr2 is running.
There is no Incr1, so the backup is executed as an Incr1.
• In the fifth row, there is a full backup, an Incr1 and another incremental is running.
Data Protector references the currently running backup to the previous incremental,
that is Incr1.
• In the eighth row, the Incr3 is executed as Incr2, and in the eleventh row, the
Incr3 is executed as Incr1.
Considering restore
To restore the latest data, you need media from your last full backup and subsequent
incremental backups. Therefore, the more incremental backups you have, the more
media you need to handle. This is inconvenient if you use standalone devices, and
the restore can last long.
Using simple and leveled incremental backups, as indicated in Figure 25 on page 98,
will require access to all five previously completed media sets, up to and including
the full backup. The space needed on the media is lowest here, but the restore is
rather complex. The series of required media sets is also called a restore chain.
96
Planning your backup strategy
TIP:
Use the Data Protector Appendable on Incrementals Only option to keep data from full
and incremental backups (of the same backup specification) on the same media set.
Another common use of the incremental backup concept is indicated in Figure
26 on page 98. Here the required space on the media is slightly larger. Only two
media sets need to be accessed to restore to the desired point in time. Note that
there is no dependency on any previous Incr1 media set for this restore, unless the
desired restore point in time would be moved.
Concepts guide
97
Figure 25 Media needed to restore from simple and leveled incremental
backups
Figure 26 Media needed to restore from leveled incremental backups
Note that you must set the appropriate data protection in order to get all needed full
and incremental backups for restore. If the data protection is not properly set, you
can get a broken restore chain. For more information, see Appendix B on page 331.
98
Planning your backup strategy
Keeping backed up data and information about the
data
Data Protector lets you specify how long to keep your backed up data on the media
itself (data protection), how long to keep information about the backed up data in
the IDB (catalog protection), and what level of information to keep in the IDB (logging
level).
You can set the protection independently for backed up data and for backup
information about this data in the IDB. When copying media, you can specify a
different protection period for the copies than the protection of the original media.
Data Protector Internal Database
Restore performance depends, in part, on how fast the media required for a restore
can be found. By default, this information is stored in the IDB to enable the highest
restore performance as well as the convenience of being able to browse the files and
directories to be restored. However, putting all file names of all backups in the IDB
and keeping them for a long time can cause the IDB to grow to unmanageable levels.
Data Protector allows you to trade off IDB growth with the convenience of restore,
by letting you specify catalog protection independently of data protection. For example
you can implement a policy that enables an easy and fast restore within four weeks
after the backup, by setting catalog protection to four weeks. From then on restores
can still be done in a less convenient way until the data protection expires, say after
one year. This would considerably reduce the space requirements in the IDB.
Data protection
What is data protection?
Data Protector allows you to specify the amount of time data on media is protected
from being overwritten by Data Protector. You can specify the protection in absolute
or relative dates.
You can specify data protection in different parts of Data Protector. For details, see
the online Help index: "data protection".
If you do not change the Data Protection backup option when configuring a backup,
it is permanently protected. Note that if you do not change this protection, the number
of media needed for backup grows constantly.
Concepts guide
99
Catalog protection
What is catalog protection?
Data Protector saves information about backed up data in the IDB. Since the
information about the backed up data is written to the IDB each time a backup is
done, the IDB grows with the number and the size of backups. Catalog protection
tells Data Protector how long the information about backed up data is available to
users browsing data during restore. Once catalog protection has expired, Data
Protector will overwrite this information in the IDB (not on the media) in one of the
subsequent backups.
You can specify the protection using absolute or relative dates.
If you do not change the Catalog Protection backup option when configuring your
backup, information about backed up data has the same protection duration as data
protection. Note that if you do not change this, the IDB grows constantly as new
information is added with each backup.
For more information on how catalog protection settings influence the IDB growth
and performance, see “Catalog protection as an IDB key tunable
parameter” on page 201.
The protection model used by Data Protector can be mapped to the concept of backup
generations, which is elaborated in Appendix B on page 331
Logging level
What is logging level?
Logging level determines the amount of details on files and directories written to the
IDB during backup. You can always restore your data, regardless of the logging
level used during the backup.
Data Protector provides four logging levels that control the amount of details on files
and directories written to the IDB. For more information, see “Logging level as an
IDB key tunable parameter” on page 199.
Browsing files for restore
The IDB keeps information about the backed up data. This information allows you to
browse, select and start the restore of files using the Data Protector user interface.
You can also restore data without this information as long as the media are still
100
Planning your backup strategy
available, but you must know which media to use and what needs to be restored,
for example, the exact file name.
The IDB also keeps information on how long the actual data on the media will not
be overwritten.
Data protection, catalog protection and logging level policies influence the availability
of data and access time to data during restore.
Enabling the browsing of files and quick restore
To restore files quickly, both information about backed up data in the catalog and
protected data on the media, must exist. Information in the catalog allows you to
browse, select, and start the restore of files using the Data Protector user interface
and allows Data Protector to quickly locate data on backup media.
Enabling the restore of files, but not browsing
Once catalog protection has expired and data protection is still valid, you cannot
browse files in the Data Protector user interface, but you can still restore data if you
know the file name and the media. The restore is slower as Data Protector does not
know where on the media the desired data is located. You can also import the media
back into the IDB, thus re-establishing the information about backed up data in the
catalog, and then start restoring.
Overwriting backed up files with new data
Once data protection has expired, data on the media is overwritten in one of the
subsequent backups. Before this happens, you can still restore the data from the
media.
TIP:
Set data protection to the amount of time that you must keep the data, for example, one
year.
Set the catalog protection to the amount of time you want to be able to browse, select,
and restore files quickly using the Data Protector user interface.
Concepts guide
101
Exporting media from a cell
Exporting media from a Data Protector cell removes all the information about backed
up data on the media and the media themselves from the IDB. You cannot browse,
select or restore files from exported media using the Data Protector user interface.
You need to re-read (or add) the media back into the Data Protector cell. This
functionality is needed to move media to a different cell.
Backing up data
Backing up your data consists of some or all of the following steps:
•
•
•
•
•
Selecting what to back up, from which client system - the source of data.
Selecting where to back up - the destination.
Selecting to write the same data to additional media sets - mirroring.
Selecting how to back up - backup options.
Scheduling a backup for automated operation.
You can specify all these when creating a backup specification.
Figure 27 Backup session
At the specified time, Data Protector starts a backup session based upon a backup
specification. The source of data is specified as a list of objects (such as a filesystem
102
Planning your backup strategy
on UNIX or disk drives on Windows systems) and the destinations are specified (tape)
devices. During the backup session, Data Protector reads the objects, transfers data
through the network, and writes it to the media residing in the devices. The backup
specification names the devices to use. It also can specify a media pool. If no media
pool is specified, the default media pool is used. A backup specification can be a
simple definition of the backup of a disk to a standalone DDS drive, or a complex
definition of the backup of 40 large servers to a Silo tape library with eight drives.
Creating a backup specification
What is a backup specification?
A backup specification allows you to group objects that you want to back up in a
group with common characteristics, such as scheduling, used devices, type of backup,
and backup session options.
How to create a backup specification
You configure a backup specification using the Data Protector user interface. You
need to know what you want to back up, how many mirrors you want to create,
which media and which devices you want to use for the backup, and optionally,
some desired specific behavior for the backup. Data Protector provides default
behavior that is suitable for most cases. You can customize backup behavior using
Data Protector backup options.
Data Protector can back up a client with all the disks connected to it by discovering
the disks at backup time. See “Backing up with disk discovery” on page 225.
Selecting backup objects
What is a backup object?
Data Protector uses the term backup object for a backup unit that contains all items
selected for backup from one disk volume (logical disk or mount point). The selected
items can be any number of files, directories, or the entire disk or mount point.
Additionally, a backup object can be a database entity or a disk image (rawdisk).
A backup object is defined by:
• Client name: a hostname of the Data Protector client where the backup object
resides.
• Mount point: an access point in a directory structure (drive on Windows and
mount point on UNIX) on the client where the backup object is located.
Concepts guide
103
• Description: uniquely defines the backup objects with identical client name and
mount point.
• Type: backup object type, for example filesystem or Oracle.
The way in which a backup object is defined is important to understand how
incremental backups are done. For example, if the description of a backup object
changes, it is considered as a new backup object, therefore a full backup will be
automatically performed instead of incremental.
Examples of backup options
You can customize the backup behavior for each individual backup object by
specifying the backup options for this object. The following are examples of the
backup options you can specify:
• Logging level of information going to the IDB.
Data Protector provides four levels that control the amount of details on files and
directories stored in the IDB:
• Log All
• Log Files
• Log Directories
• No Log
Note that changing the level of stored information affects the ability to browse
the files using the Data Protector user interface when restoring. For more
information on logging levels, see “Logging level as an IDB key tunable
parameter” on page 199.
• Automatic load balancing
Dynamic device allocation from a specified list. For more information, see “How
load balancing works” on page 156.
Data Protector dynamically determines which object (disk) should be backed up
to which device.
• Pre-exec and post-exec scripts
Processing to prepare a client for a consistent backup. For more information, see
“Pre-exec and post-exec commands” on page 223.
You can also specify the directories to exclude from a backup, or back up specific
directories only. You can also back up disks as they are added. Thus, your backup
is fully configurable and dynamic.
104
Planning your backup strategy
Backup sessions
What is a backup session?
A backup session is a process that backs up data from a client system to media. A
backup session always runs on the Cell Manager system. A backup session is based
on a backup specification and is started when a backup is run.
During a backup session, Data Protector backs up data using default or customized
behavior.
For advanced information on backup sessions, and how to control sessions, see
Chapter 7 on page 219.
Object mirrors
What is an object mirror?
An object mirror is an additional copy of a backup object created during a backup
session. When creating a backup specification, you can choose to create one or
several mirrors of specific objects. The use of object mirroring improves the fault
tolerance of backups and enables multi-site vaulting. However, object mirroring
during a backup session increases the time needed for backup.
For more information, see “Object mirroring” on page 119.
Media sets
What is a media set?
The result of a backup session is backed up data on a medium or a media set. Each
backup session results in one or several media sets, depending on whether you
perform backup with object mirroring. Depending on the pool usage, several sessions
can share the same media. When you restore data, you need to know the media
from which to restore. Data Protector keeps this information in the Catalog Database.
Backup types and scheduled backups
A scheduling policy defines when backups start and the backup types (full or
incremental). Consider the differences between full and incremental backups. See
Table 6 on page 91.
Concepts guide
105
You can combine full and incremental backups when you configure scheduled
backups. For example, you may run a full backup on Sundays and incremental
backups every working day. To back up a large amount of data and avoid the high
volume peak for the full backups, use the staggered approach. See “Staggering full
backups” on page 107.
Scheduling, backup configurations, and sessions
Backup configuration
When you schedule a backup, all the objects specified in that backup specification
are backed up in the scheduled backup session(s).
For each individual or periodic scheduled backup, you can specify the following
options: Backup type (full or incremental), Network load, and Backup protection.
With split mirror or snapshot backup, in the case of ZDB to disk or ZDB to disk+tape
(instant recovery enabled), you specify the Split mirror/snapshot backup option. For
split mirror and snapshot backups, the backup type is ignored (a full backup is
performed).
Within one backup specification, you can schedule both ZDB to disk and ZDB to
disk+tape, and specify a different data protection period for each individual or
periodic scheduled backup.
Backup session
When a backup session is started, Data Protector tries to allocate all needed resources,
such as devices. The session is queued for as long as the required minimum resources
are not yet available. Data Protector tries to allocate the resources for a specific
period of time, the timeout. Timeout is user configurable. If the resources are still
unavailable after the timeout, the session is aborted.
Optimizing backup performance
To optimize the load on the Cell Manager, Data Protector by default starts five backup
sessions at the same time. If more are scheduled at the same time, the excessive
sessions are queued and started subsequently as the others are finished.
Scheduling tips and tricks
The sections “Full and incremental backups” on page 91 and “Keeping backed up
data and information about the data” on page 99 describe the concept of backup
generations, data protection, and catalog protection.
106
Planning your backup strategy
This section combines all these concepts by giving some examples of backup schedules
and some tips for efficient scheduling.
When to schedule backups
Typically, you schedule backups to run during lowest user activity, usually at night.
Full backups take the most time, so schedule them at weekends.
Consider scheduling full backups for different clients (backup specifications) on
different days, as shown in “Staggering full backups” on page 107.
NOTE:
Data Protector offers reports that show available time slots from a device-usage point of
view. This allows you to pick a time where the devices to use are not likely to be occupied
by serving already existing backups.
Staggering full backups
Performing a full backup of all systems during the same day may cause network load
and time window problems. To avoid these problems, use the staggered approach
for full backups.
Table 8 The staggered approach
Mon
Tue
Wed
...
system_grp_a
FULL
Incr1
Incr1
...
system_grp_b
Incr1
FULL
Incr1
...
system_grp_c
Incr1
Incr1
FULL
...
Optimizing for restore
The combination of your scheduling policy with full and incremental backups highly
influences the time needed to restore your data. This is illustrated in three examples
in this section.
For a point-in-time restore, you need a full backup plus all the incremental backups
to the desired point in time. Since full and incremental backups are typically not on
the same media, you may need to load different media for the full and each
Concepts guide
107
incremental backup. For more information on how Data Protector selects media for
backups, see Selecting media for backups .
Example 1
Figure 28 on page 108 depicts a scheduling policy based on a full backup plus simple
incremental backups.
Figure 28 Full backup with daily simple incremental backups
This policy reduces the media space and time needed for backing up, because you
only back up changes from the previous day. However, to restore files from a Thursday
backup, you need to provide the media for the full and each of the incremental
backups until Thursday, that is five media sets. This complicates and slows down the
restore.
Example 2
Figure 29 on page 109 depicts a scheduling policy based on a full backup plus level
one incremental backups.
108
Planning your backup strategy
Figure 29 Full backup with daily level 1 incremental backups
This policy requires slightly more time for backups and also requires a little more
media since you back up all the changes from the last full backup every day. To
restore files from Thursday’s backup, you need to provide media for the full and for
Thursday’s incremental backup, that is, two media sets only. This considerably
simplifies and speeds up the restore.
Example 3
Depending on your environment and requirements, the best solution could lie
somewhere in between. For example, you may have the following scheduling policy:
Concepts guide
109
Figure 30 Full backup with mixed incremental backups
This policy takes into account the fact that there are not many changes during
weekends. Data is backed up using a combination of simple incremental backups
and Incr1 (differential) backups to optimize backup performance. To restore files
from Thursday’s backup, you need to provide media from the full backup and the
second Incr1 backup, that is, two media sets.
Automated or unattended operation
To simplify operation and the operator’s involvement in the backup process, Data
Protector provides extensive functionality supporting unattended or automatic backup
during lights-out time. This section describes how to plan your scheduling policies,
how these policies influence the behavior of backup, and provides examples of
scheduling policies. This section focuses on longer periods of unattended operation
spanning from several days to weeks, rather than the unattended operation during
a single backup.
Considerations for unattended backups
Data Protector provides simple ways of scheduling your backups. Since the
effectiveness of scheduling policies depends on your environment, you need to plan
before finding the best scheduling policy.
110
Planning your backup strategy
• When is the lowest system usage and user activity?
Typically, this is at night and most backups are scheduled to run during the night.
Data Protector can generate reports about devices used for backup.
• What kind of data do you have and how often do you want to schedule backups
of this data?
Data that changes often and is important to the company, such as user files,
transactions, and databases must be backed up regularly. System-specific data,
such as program files that do not change often, do not need to be backed up so
often.
• How much do you want to simplify restore?
Depending on how you schedule your full and incremental backups, you will
need media from the full and incremental backups to restore the latest version of
files. This may take longer or even require manual media handling if you do not
have an automatic library device.
• How much data do you need to back up?
Full backups take longer than incremental backups. Backups must typically be
done in a limited time-frame.
• How many media are required?
Define a media rotation policy. See “Implementing a media rotation
policy” on page 143. This will show if you can keep enough media inside the
planned library to operate for the desired period without having to handle media
manually.
• What about mount prompt handling?
Consider whether to use one or several libraries. This enables automatic operation,
since Data Protector can have access to all or most of the media, hence
significantly reducing the need to manually handle media. If the data volume is
too large for a library, then consider using more libraries. or more information,
see “Large libraries” on page 162.
• How do I handle unavailable devices?
Use dynamic load balancing or device chaining, and provide several devices
when creating a backup specification. This way you avoid the failure of a backup
if a device is not turned on or the system to which the device is connected is not
functioning.
• How long can a backup of all data take?
Since backups must finish during a period of low network usage and when users
do not use their systems, consider scheduling backups appropriately to distribute
the network load caused by the backups, and to maximize the efficiency of backup
sessions. This may require using the staggered approach.
Concepts guide
111
If you need to back up large amounts of data and the backup window presents
a problem, consider backing up to disk-based devices and using advanced backup
strategies such as synthetic backup and disk staging.
• How can I prepare running applications for backups? Many applications keep
files open, so running a backup would produce an inconsistent backup. This can
be avoided by using pre-exec and post-exec scripts that can be used to synchronize
the status of applications with the backup activities.
Duplicating backed up data
Duplicating backed up data brings several benefits. You can copy data to improve
its security and availability, or for operational reasons.
Data Protector provides the following methods of duplicating backed up data: object
copy, object mirror, and media copy. See Table 9 on page 112 for an overview of
the main characteristics of these methods.
Table 9 Data Protector data duplication methods
Object copy
Object mirror
Media copy
Any
combination of
object versions
from one or
several backup
sessions
A set of objects
from a backup
session
An entire
medium
Time of
duplication
Any time after
the completion
of a backup
During backup
Any time after
the completion
of a backup
Any time after
the completion
of a backup
Media type of
source and
target media
Can be different
Can be different
Must be the
same
Are different as
disk-based
storage is
combined with
tape-based
storage
Size of source
and target
media
Can be different
Can be different
Must be the
same
Must be the
same
What is
duplicated
112
Planning your backup strategy
Smart Media
Copy
An entire
medium
1
Object copy
Object mirror
Media copy
Smart Media
Copy
Appendability
of target media
Yes
Yes
No
No
2
3
Result of the
operation
Media
containing the
selected object
versions
Media
containing the
selected object
versions
Media identical
to the source
media
Media identical
to the source
media
1
Source media are located on virtual tapes stored on disk arrays and target media are located on a
physical tape library attached to the VLS.
2
You can use only unformatted media, empty media, or media with expired protection as target media.
After the operation, both the source and the target media become non-appendable.
3
You can use only unformatted media, empty media, or media with expired protection as target media.
After the operation, both the source and the target media become non-appendable.
Copying objects
What is object copy?
The Data Protector object copy functionality enables you to copy selected object
versions to a specific media set. You can select object versions from one or several
backup sessions or object consolidation sessions. During the object copy session,
Data Protector reads the backed up data from the source media, transfers the data,
and writes it to the target media.
The result of an object copy session is a media set that contains copies of the object
versions you specified.
Figure 31 on page 114 shows how data backed up at a specific point in time can be
copied afterwards. You can copy any backup object from a medium containing a
backup or a medium containing a copy of the object.
Concepts guide
113
Figure 31 Object copy concept
In the figure, there is an object version resulting from a backup of object A, version
1, and two additional copies of the same object version. Version 1-1 has been
obtained by copying the object version resulting from the backup, and version 1-1-1
by copying a copy of the object version. Any of these object versions can be used
for a restore of the same object version.
Start of object copy session
You can start an object copy session interactively or specify an automated start of
the session. Data Protector offers two types of automated object copying: post-backup
object copying and scheduled object copying.
Post-backup object copying
Post-backup object copying takes place after the completion of a session that is
specified in the automated object copy specification. It copies objects selected
according to the automated object copy specification that were written in that
particular session.
Scheduled object copying
Scheduled object copying takes place at a user-defined time. Objects from different
sessions can be copied in a single scheduled object copy session.
114
Planning your backup strategy
Selection of devices
You need separate devices to be used with the source media and the target media.
The destination devices can have a larger block size than the source devices.
However, to avoid impact on performance, it is recommended that the devices have
the same block size and are connected to the same system or to a SAN environment.
Object copying is load balanced by default. Data Protector makes optimum use of
the available devices by utilizing as many devices as possible.
If you do not specify the source devices to be used in the object copy specification,
Data Protector uses the default devices. By default, the devices that were used for
writing the objects are used as source devices. If destination devices are not specified
per object, Data Protector selects them automatically from those you selected in the
object copy specification according to the following criteria in the order of priority:
• destination devices of the same block size as source devices are selected before
those with a different block size
• locally attached devices are selected before network attached devices
Devices are locked at the beginning of the session. Devices that are not available at
that time cannot be used in the session, as device locking after the beginning of the
session is not possible. If a media error occurs, the device with errors will be avoided
within that copy session.
Selection of the media set to copy from
If an object version that you want to copy exists on more than one media set, which
has been created using one of the Data Protector data duplication methods, any of
the media sets can be used as a source for copying. You can influence the media
set selection by specifying the media location priority.
The overall process of media selection is the same as for restore. For details, see
“Selection of the media set” on page 126.
Object copy session performance
An impact on object copy performance can be caused by factors such as device
block sizes and the connection of devices. If the devices used in the object copy
session have different block sizes, the data will be repackaged during the session,
which takes additional time and resources. If the data is transferred over the network,
there will be additional network load and time consumption. This impact can be
minimized if the operation is load balanced.
Concepts guide
115
Why use object copy?
Additional copies of backed up data are created for multiple purposes:
• Vaulting
You can make copies of backed up objects and keep them in several locations.
• Freeing media
To keep only protected object versions on media, you can copy such object
versions, and then leave the medium for overwriting.
• Demultiplexing of media
You can copy objects to eliminate interleaving of data.
• Consolidating a restore chain
You can copy all object versions needed for a restore to one media set.
• Migration to another media type
You can copy your backups to media of a different type.
• Support of advanced backup concepts
You can use backup concepts such as disk staging.
Vaulting
Vaulting is a process of storing media in a safe place, often called a vault, where
they are kept for a specific period of time. For details, see “Vaulting” on page 151.
It is recommended to keep a copy of the backed up data on site for restore purposes.
To obtain additional copies, you can use the object copy, object mirror, or media
copy functionality, depending on your needs.
Freeing media
You can minimize the media space consumption by keeping only protected backups
and overwriting unprotected ones. As a single medium may contain both, you can
copy protected objects to a new media set and leave the medium for overwriting.
See Figure 32 on page 117.
116
Planning your backup strategy
Figure 32 Freeing media
Demultiplexing of media
Multiplexed media contain interleaved data of multiple objects. Such media may
arise from backup sessions with the device concurrency more than 1. Multiplexed
media may compromise the privacy of backups and require more time for restore.
Data Protector offers a possibility of demultiplexing of media. Objects from a
multiplexed medium are copied to several media that you specify. See Figure
33 on page 118.
Concepts guide
117
Figure 33 Demultiplexing a medium
Consolidating a restore chain
You can copy a restore chain (all backups that are necessary for a restore) of an
object version to a new media set. A restore from such a media set is faster and more
convenient, as there is no need to load several media and seek for the needed object
versions.
Migration to another media type
You can migrate backed up data to another media type. For example, you can copy
objects from file devices to LTO devices or from DLT devices to LTO devices.
Disk staging
The concept of disk staging is based on backing up data in several stages to improve
the performance of backups and restores, reduce costs of storing the backed up data,
and increase the data availability and accessibility for restore.
The backup stages consist of backing up data to media of one type and later moving
the data to media of a different type. The data is backed up to media with high
performance and accessibility, but limited capacity (for example, system disks). These
backups are usually kept accessible for restore for a period of time when a restore
is the most probable. After a certain period of time, the data is moved to media with
lower performance and accessibility, but high capacity for storage, using the object
copy functionality. See Figure 34 on page 119.
118
Planning your backup strategy
Figure 34 Disk staging concept
Disk staging also eliminates the need for frequent backups of numerous small objects
to tape. Such backups are inconvenient due to frequent loading and unloading of
media. The use of disk staging reduces backup time and prevents media deterioration.
Object mirroring
What is object mirroring?
The Data Protector object mirror functionality enables writing the same data to several
media sets simultaneously during a backup session. You can mirror all or some
backup objects to one or more additional media sets.
Concepts guide
119
The result of a successful backup session with object mirroring is one media set
containing the backed up objects and additional media sets containing the mirrored
objects. The mirrored objects on these media sets are treated as object copies.
Benefits of object mirroring
The use of the object mirror functionality serves the following purposes:
• It increases the availability of backed up data due to the existence of multiple
copies.
• It enables easy multi-site vaulting, as the backed up data can be mirrored to
remote sites.
• It improves the fault tolerance of backups, as the same data is written to several
media. A media failure on one medium does not affect the creation of the other
mirrors.
Object mirror operation
In a backup session with object mirroring, each selected object is backed up and at
the same time mirrored as many times as specified in the backup specification. See
Figure 35 on page 121.
Let us take Object 3 in the figure as an example. The Disk Agent reads a block of
data from the disk and sends it to the Media Agent that is responsible for the backup
of the object. This Media Agent then writes the data to the medium in Drive 2 and
forwards it to the Media Agent that is responsible for mirror 1. This Media Agent in
turn writes the data to the medium in Drive 4 and forwards it to the Media Agent
that is responsible for mirror 2. This Media Agent writes the data to the medium in
Drive 5. At the end of the session, Object 3 is available on three media.
120
Planning your backup strategy
Figure 35 Object mirroring
Selection of devices
Object mirroring is load balanced by default. Data Protector makes optimum use of
the available devices by utilizing as many devices as possible. Devices are selected
according to the following criteria in the order of priority:
• devices of the same block size are selected, if available
• locally attached devices are selected before network attached devices
When you perform an object mirror operation from the command line, load balancing
is not available.
Backup performance
Object mirroring has an impact on backup performance. On the Cell Manager and
Media Agent clients, the impact of writing mirrors is the same as if additional objects
were backed up. On these systems, the backup performance will decrease depending
on the number of mirrors.
On the Disk Agent clients, there is no impact caused by mirroring, as backup objects
are read only once.
Concepts guide
121
Backup performance also depends on factors such as device block sizes and the
connection of devices. If the devices used for backup and object mirroring have
different block sizes, the mirrored data will be repackaged during the session, which
takes additional time and resources. If the data is transferred over the network, there
will be additional network load and time consumption.
Copying media
What is media copying?
The Data Protector media copy functionality enables you to copy media after a
backup has been performed. Media copying is a process that creates an exact copy
of a medium containing a backup. You can use it to duplicate media for archiving
or vaulting purposes. After the media have been copied, you can move either the
original media or the copies to an off-site vault.
Besides manually started media copying, Data Protector also offers automated media
copying. For more information, see “Automated media copying” on page 124.
How to copy media
You need two devices of the same media type, one for the source medium and one
for the target medium. The source medium is the medium being copied while the
target medium is the medium to which data is copied.
When you copy media within a library that has multiple drives, you can use one
drive for the source and one for the copy.
What is the result?
The result of copying media is two identical sets of media, the original media set and
the copy. Either of them can be used for restore.
After the source medium has been copied, Data Protector marks it as non-appendable
to prevent appending new backups (this would result in the original being different
from its copy.) The copy is also marked as non-appendable. The default protection
of the copy is the same as for the original.
You can make multiple copies of the original media. You cannot, however, make
copies of copies, also known as second generation copies.
122
Planning your backup strategy
Automated media copying
What is automated media copying?
Automated media copying is an automated process that creates copies of the media
containing backups. This functionality is available with library devices.
Data Protector offers two types of automated media copying: post-backup media
copying and scheduled media copying.
Post-backup media copying
Post-backup media copying takes place after the completion of a backup session. It
copies the media used in that particular session.
Scheduled media copying
Scheduled media copying takes place at a user-defined time. Media used in different
backup specifications can be copied in a single session. You create an automated
media copy specification to define which media will be copied.
How does automated media copying operate?
First you create an automated media copy specification. When the automated media
copy session begins, Data Protector generates a list of media, referred to as source
media, based on the parameters specified in the automated media copy specification.
For each source medium, a target medium is selected to which the data will be
copied. The target media are selected from the same media pool as the source media,
from a free pool, or from the blank media in a library.
For each source medium, Data Protector selects a pair of devices from the devices
that you specified in the automated media copy specification. The automated media
copy functionality provides its own load balancing. Data Protector tries to make
optimum use of the available devices by utilizing as many devices as possible and
selecting local devices, if they are available.
The automated media copy functionality does not handle mount or cleanme requests.
If a mount request is received, the media pair concerned is aborted, but the session
continues.
For examples of use, see “Examples of automated media copying” on page 332.
Concepts guide
123
Smart media copying using VLS
What is smart media copying?
In smart media copying, the data is first backed up to a virtual tape library (VTL)
configured on the Virtual Library System (VLS). Then, a copy of a virtual tape
containing a backup is made to the physical library attached to the VLS in a process
called automigration. Data Protector initiates the copy process, which is then
performed by the VLS. The data is transferred to a physical library in a smart copy
operation, which allows Data Protector to distinguish between the source and the
target media thus enabling media management. The smart copy media follow the
Data Protector format and can thus be inserted in any compatible tape drive and
read by Data Protector. The result of smart copying is two identical sets of media,
the source medium located on the VLS' virtual tape and the target medium (a smart
copy) located on a physical tape library attached to the VLS. Either of these copies
can be used for restore, thus increasing the security and availability of the backed
up data. You can also keep smart media copies for archiving or vaulting purposes.
Data Protector offers two types of smart media copying: automated smart media
copying and interactive smart media copying.
Automated smart media copying
You can create automated smart media copying of the following types:
• Post-backup smart media copying, which takes place after the completion of a
backup session and copies the media used in that particular session.
• Scheduled smart media copying, which takes place at a specific time or at regular
intervals.
Interactive smart media copying
Interactive smart media copying creates a copy of a medium containing the backed
up data and can be started on demand at any point in time.
What happens after the backup?
After the backup data has been moved to a physical tape, it is still available for the
Data Protector restore. However, since the destination library is not visible to Data
Protector, the restore cannot be performed directly from this library but from any tape
drive or library that is controlled by Data Protector.
For more information about VLS smart copies, see the online Help index: "smart
media copying" and the VLS documentation.
124
Planning your backup strategy
Restoring data
Policies for restoring data are an essential part of the overall backup strategy in the
company. Keep the following in mind:
• Backing up and restoring files is essentially the same as copying files. Therefore,
ensure that only authorized people have the rights to restore confidential data.
• Ensure that unauthorized people cannot restore files of other people.
This section describes some possible implementations of the restore policy using Data
Protector. You can restore your filesystem data by browsing through restore objects
or restore sessions. By default, data is restored to its original location. However, you
can specify any location to be the destination of restored data.
Restore duration
After data loss, access to data is possible only after the recovery process is finished.
It is often critical to minimize restore duration so that users can do their regular work.
Therefore, plan for the time needed to restore specific data.
Factors affecting restore duration
The restore duration depends on a number of factors, such as:
• The amount of data to be restored. This also directly influences all the following
items.
• A combination of full and incremental backups. For more information, see “Full
and incremental backups” on page 91.
• Media and devices used for backup. For more information, see Chapter
3 on page 133.
• Speed of networks and systems. For more information, see “Understanding and
planning performance” on page 67.
• The application you are recovering, for example, Oracle database files. For more
information, see the appropriate HP Data Protector integration guide.
• The use of parallel restore. Several objects can be restored with a single read
operation, depending on how the data was backed up. See “Parallel
restores” on page 228.
• Speed and ease of selecting the data to be restored, which depends on the
logging level settings used during the backup and on catalog protection time.
See “Logging level as an IDB key tunable parameter” on page 199.
Concepts guide
125
Selection of the media set
If an object version that you want to restore exists on more than one media set, which
has been created using one of the Data Protector data duplication methods, any of
the media sets can be used for the restore. By default, Data Protector automatically
selects the media set that will be used. You can influence the media set selection by
specifying the media location priority. You can also manually select the media set
you want to use for the restore, except when restoring integration objects.
Media set selection algorithm
By default, Data Protector selects the media set with the best availability and quality.
For example, Data Protector avoids media sets with missing media or poor media;
it considers the completion status of the objects, the availability and locality of the
device to be used with a certain media set, and so on. A media set located in a
library is used before one in a standalone device.
Selection of restore chain
If you use synthetic backup, there is often more than one restore chain for the same
point in time of an object. By default, Data Protector selects the most convenient
restore chain and the most appropriate media within the selected restore chain.
Media location priority
To influence the selection of the media set, specify the media location priority. This
is important if you use the concept of multi-site storage. If you keep media at different
sites, you can specify which location is preferable for a specific restore. Data Protector
will use the media set with the highest priority if more than one media set matches
the conditions of the selection algorithm.
You can set the media location priority globally or for a specific restore session.
Selection of devices
By default, Data Protector restores selected data with the same devices that were
used during backup. Optionally, you can restore the data with some other devices
of the same type. You can also specify what Data Protector should do if the selected
devices are not available, for example, if they are disabled or already in use:
• Original device selection:
Data Protector should wait for the devices to become available. This is the
preferred option for Data Protector SAP DB/MaxDB, IBM UDB DB2, Microsoft
126
Planning your backup strategy
SQL Server, and Microsoft SharePoint Portal Server integrations. Such databases
are usually backed up with interdependent data streams and, consequently,
restore must be started with the same number of devices as used during backup.
• Automatic device selection (default):
Data Protector should automatically replace unavailable devices with available
devices of a compatible type. You can define which devices are compatible by
giving them the same device subtype name during configuration. Only devices
(drives) that are of the same media type and from the same library should have
the same device subtype name. The restore can be started with fewer devices
than were used during backup.
Operators are allowed to restore
A popular restore policy is that only dedicated backup operators or network
administrators have the right to restore files or perform disaster recovery.
When to use this policy
Use this policy in the following cases:
• In a large network environment where it is best to have a dedicated person to do
such jobs.
• In an environment where end users do not have the necessary computer knowledge
to restore files, operators can be trusted to restore sensitive data.
What needs to be done
You need to do the following, to implement this policy:
• Add the backup operators or network administrators that will restore data for
other people to the Data Protector operators or admin user group.
You do not need to add other people (such as users who want to perform restores
to their own systems) to any Data Protector user group.
• During installation, do not install the Data Protector user interface on end-user
systems. Install the Disk Agent that allows Data Protector to back up these systems.
• Establish a policy of handling requests for restore. This policy should cover how
end users request the restore of files, for example, via email containing all the
details necessary for the operator to locate and restore the files back to the
end-user system. The end users should also have a way of knowing when the files
have been restored.
Concepts guide
127
End users are allowed to restore
Another possible restore policy is to allow all or just selected end users to restore
their own data. This policy provides sufficient security and may relieve the backup
operator from doing a number of restore operations.
When to use this policy
Use this policy in the following cases:
• When the end users have sufficient knowledge to handle restores. You may need
to provide some training for the users on basic backup concepts and restore
operations.
• You use library backup devices with media of most recent backups. The end
user Data Protector user group, by default, does not allow end users to handle
mount requests for needed media. The end users will still need the assistance of
the backup operator in case of mount requests. This can be avoided by using
large libraries.
What needs to be done
You need to do the following to implement this policy:
• Add the end users that are allowed to restore their own data to the Data Protector
end users user group. For additional security, you may limit the Data Protector
access of these users, to a specific system only.
• Install the Data Protector user interface on the systems the end users are using.
Data Protector automatically checks the user rights and allows restore functionality
only.
• When you configure backups of the end-user systems, make backups visible to
the end users by setting the Data Protector public option.
Disaster recovery
This section provides only a short overview of the disaster recovery concepts. Detailed
disaster recovery concepts, planning, preparation, and procedures are described in
the HP Data Protector disaster recovery guide.
A computer disaster refers to any event that renders a computer system unbootable,
whether due to human error, hardware or software failure, natural disaster, etc. In
these cases it is most likely that the boot or system partition of the system is not
available and the environment needs to be recovered before the standard restore
128
Planning your backup strategy
operation can begin. This includes repartitioning and/or reformatting the boot partition
and recovery of the operating system with all the configuration information that
defines the environment. This has to be completed in order to recover other user data.
After a computer disaster has occurred, the system (referred as target system) is
typically in a non-bootable state and the goal of Data Protector disaster recovery is
to restore this system to the original system configuration. The difference between the
crashed and the target system is that the target system has all faulty hardware
replaced.
A disaster is always serious, however the following factors can exacerbate the
situation:
• The system needs to be returned to online status as quickly and efficiently as
possible.
• Administrators are not familiar with the required steps to perform the disaster
recovery procedure.
• The available personnel to perform the recovery have only fundamental system
knowledge.
Disaster recovery is a complex task that involves extensive planning and preparation
before execution. You need to have a well-defined, step-by-step process in place to
prepare for, and recover from, disastrous situations.
The disaster recovery process consists of 4 phases:
1.
Phase 0 (planning/preparation) is the prerequisite for a successful disaster
recovery.
CAUTION:
It is too late to prepare for a disaster recovery once a disaster has occurred.
2. In Phase 1, DR OS is installed and configured, which usually includes
repartitioning and reformatting of the boot partition, since the boot or system
partition of the system are not always available and the environment needs to
be recovered before normal restore operations can resume.
3. In Phase 2, the operating system with all the configuration information that defines
the environment with Data Protector (as it was) is restored.
4. Only after phase 2 is completed, is the restore of applications and user data
possible (Phase 3). A well-defined, step-by-step process has to be followed to
ensure a fast and efficient restore.
Concepts guide
129
Disaster recovery methods
Data Protector supports the following disaster recovery methods:
• Manual disaster recovery
This is a basic and very flexible disaster recovery method. You need to install
and configure the DR OS. Then use Data Protector to restore data (including the
operating system files), replacing the operating system files with the restored
operating system files.
• Automated disaster recovery
Automated System Recovery (ASR) is an automated system on Windows systems,
which reconfigures a disk to its original state (or resizes the partitions if the new
disk is larger than the original disk) in the case of a disaster. ASR thus enables
the Data Protector drstart.exe command to install the active DR OS that
provides Data Protector disk, network, tape and file system access.
• Disk delivery Disaster recovery
On Windows clients, the disk of the crashed system (or the replacement disk for
the physically damaged disk) is temporarily connected to a hosting system. After
being restored, it can be connected to the faulty system and booted. On UNIX
systems, the auxiliary disk with a minimal operating system, networking, and
Data Protector agent installed is used to perform Disk Delivery Disaster Recovery.
• Enhanced Automated Disaster Recovery (EADR)
Enhanced Automated Disaster Recovery (EADR) is a fully automated Data Protector
recovery method for Windows clients and Cell Manager, where user intervention
is reduced to minimum. The system is booted from the disaster recovery CD ISO
image and Data Protector automatically installs and configures DR OS, formats
and partitions the disks, and finally recovers the original system with Data Protector
as it was at the time of backup.
• One Button Disaster Recovery (OBDR) is a fully automated Data Protector recovery
method for Windows clients and Cell Manager, where user intervention is reduced
to a minimum. The system is booted from the OBDR tape and automatically
recovered.
For a list of supported disaster recovery methods for a particular operating system,
see the Support Matrices in the HP Data Protector product announcements, software
notes, and references or on the Web:
http://www.hp.com/support/manuals
130
Planning your backup strategy
Alternative disaster recovery methods
This section compares the Data Protector disaster recovery concept with concepts of
other vendors. This section points out only significant aspects of alternative recovery
concepts. Two alternative recovery approaches are discussed:
Recovery methods supported by operating system vendors
Most vendors provide their own methods, but when it comes to restore, they typically
require the following steps:
1.
Reinstall the operating system from scratch
2. Reinstall the application(s)
3. Restore application(s) data
Excessive manual reconfiguration and customization of the operating system and the
application(s) is required to reconstruct the status before the disaster. This is a very
complicated, time consuming, and error-prone process using different tools that are
not integrated with each other. It does not benefit from a backup of the operating
system, the application(s), and their configurations as a whole set.
Recovery using third-party tools (for Windows)
This often consists of a special tool that backs up the system partition as a snapshot,
which can be restored rapidly. The method conceptually requires the following steps:
1.
Restore the system partition (using the third-party tool)
2. Restore any other partition (perhaps selective) if required using the standard
backup tool
It is obvious that one has to work from two different backups with different tools. This
is a difficult task to perform on a regular basis. If this concept is implemented for a
large organization, the administrative overhead to manage the different versions
(weekly backup) for the data from two tools must be addressed.
Data Protector on the other hand represents a powerful all-in-one cross-platform
enterprise solution for fast and efficient disaster recovery that includes backup and
restore and supports clustering. It provides easy central administration, easy restore,
high availability support, monitoring, reporting and notifications to aid administration
of systems in a large organization.
Concepts guide
131
132
Planning your backup strategy
3 Media management and
devices
In this chapter
This chapter describes Data Protector concepts of media and device management.
It discusses media pools, devices, and large libraries.
It is organized as follows:
“Media management” on page 133
“Media life cycle” on page 134
“Media pools” on page 135
“Media management before backups begin” on page 145
“Media management during backup sessions” on page 147
“Media management after backup sessions” on page 151
“Devices” on page 153
“Standalone devices” on page 160
“Small magazine devices” on page 161
“Large libraries” on page 162
“Data Protector and Storage Area Networks” on page 170
Media management
Serious challenges can arise when administrating large quantities of media in an
enterprise environment. Data Protector media management functionality allows for
a flexible and efficient allocation of backup data to media. This can be done in many
ways by defining methods of automatic or strict media allocation.
Concepts guide
133
Media management functionality
Data Protector provides the following media management functionality that allows
simple and efficient management of a large number of media:
• Grouping media into logical groups, media pools, that enable you to think about
large sets of media without having to worry about each medium individually.
• Data Protector keeps track of all media and the state of each medium, the data
protection expiration time, the availability of media for backups, and a catalog
of what has been backed up to each medium.
• Automated media rotation policies so that you do not need to take care of tape
rotation manually.
• The possibility to explicitly define which media and which devices you want to
use for backup.
• Optimized media management for specific device types, such as standalone,
magazine, library devices and large silo devices.
• Fully automated operation. If Data Protector has control of enough media in the
library devices, the media management functionality enables the running of
backups without the need for an operator to handle media for weeks.
• Recognition and support of barcodes on large libraries with barcode support and
silo devices.
• Automatic recognition of Data Protector media format and other popular tape
formats.
• Data Protector only writes to blank media initialized (formatted) by Data Protector.
You cannot force Data Protector to overwrite foreign tape formats during a backup,
thus you avoid accidental overwrites of media that belong to other applications.
• Recognition, tracking, viewing, and handling of media used by Data Protector
and separating it from media used by other applications in library and silo devices.
• Keeping information about the media used in a central place and sharing this
information among several Data Protector cells.
• Support for media vaulting.
• Interactive or automated creation of additional copies of the data on the media.
This chapter describes the above functionality in more detail.
Media life cycle
A typical media life cycle consists of the following steps:
134
Media management and devices
1.
Preparing media for backup.
This includes initializing (formatting) media for use with Data Protector and
assigning media to media pools, which are used to track the media.
For more information, see “Media management before backups
begin” on page 145.
2. Using media for backup.
This defines how media are selected for backup, how the condition of the media
is checked, how new backups are added to the media, and when data on the
media is overwritten.
For more information, see “Media management during backup
sessions” on page 147.
3. Vaulting media for long-term data storage. You can use one of Data Protector’s
data duplication methods to make copies of the backed up data for vaulting
purposes.
For more information on vaulting, see “Media management after backup
sessions” on page 151.
4. Recycling media for new backups once the data on the media is no longer
needed.
5. Retiring media.
Once a medium has expired, it is marked poor and will no longer be used by
Data Protector.
See “Calculating media condition” on page 150.
Media pools
Data Protector media pools manage large numbers of media, hence reducing the
management effort for the administrators to a minimum.
What is a media pool?
A pool is a logical set, or group, of media with a common usage pattern and media
properties. It can only have media of the same physical type. DLT and DAT/DDS
media cannot be in the same pool for instance.
The current location of a medium has no influence on its relation to the pool. Whether
the medium is in a drive, in a repository slot of a library, in the vault or somewhere
Concepts guide
135
else, does not matter; it always belongs to its pool until it is recycled and exported
from the cell.
Several devices can use media from the same pool.
Media pool property examples
Examples of pool properties are:
• appendable
This allows Data Protector to append data to the media in this pool when
performing subsequent backup sessions.
If this option is not selected, then the media will contain data from a single session
only.
• append incrementals only
A backup session appends to a medium only if an incremental backup is
performed. This allows you to have a complete set of full and incremental backups
on the same medium, if there is enough space.
• media allocation policy
There are several levels of strictness as to which media can be used for backup.
They range from strict, where Data Protector requires a specific medium, to loose,
where Data Protector accepts any suitable medium in the pool, including new
(blank) media.
Every device is linked to a default pool. This pool can be changed in the backup
specification.
For information on other media pool properties, see the online Help index: “media
pools, properties of“.
Media pools and dcbf directories
Data Protector allows you to set a target dcbf directory for a media pool. This means
that information about all media from the media pool is stored in the specified dcbf
directory.
For information on the DCBF part of the IDB and dcbf directories, see “IDB
architecture” on page 189.
How to use media pools
The usage of pools depends mainly on your preferences. For example, pools can be
defined using criteria like:
136
Media management and devices
• system platform (one pool for UNIX systems, one for Windows 2000 systems,
and one for Windows XP systems)
• per system (every system has its own pool)
• organizational structure (all systems in department_A have a pool, and systems
in department_B have another pool)
• systems categories (running large databases, or business critical applications)
• backup type (all full backups use one pool, and all incremental backups use
another pool)
• combinations of the above criteria, and more.
A simplified way to think about media pools is to view them as a destination for your
backup while you look at the devices as a transfer mechanism between the data and
the media pools.
The relationship of a pool to a system category is defined by putting certain systems
into the same backup specification and also specifying the pool(s). The options used
(when defining the devices, pools, and backup specifications) determine how the
data of the objects will end up on the media.
Grouping such media used for a similar kind of backup to media pools allows you
to apply common media handling policies on a group level while not bothering with
each medium individually. All media in a pool are tracked as one set and have the
same media allocation policy.
Default media pools
Data Protector provides default media pools for various media types. These default
media pools allow you to quickly run backups without having to create your own
media pools. However, to efficiently manage your large environment, create different
media pools for specific needs. When you run a backup, specify which media pool
to use.
Free pools
If media allocated to a specific media pool run out, you cannot use the media in
another pool, even if the media are of the same type. This can result in unnecessary
mount requests and operator intervention. To solve this problem, you can use the
single pool model, at which all media are in the same pool. While this allows you
to share free media, it compromises the benefits of using media pools in the first
place: easier media management, separation of important from not so important
data, etc. To alleviate this drawback, free pools are used.
Concepts guide
137
What is a free pool?
A free pool is an auxiliary source of media of the same type (for example, DLT) for
use when all free media in a regular pool run out. It helps to avoid failed backups
due to missing (free) media.
Figure 36 Free pools
When is a free pool used?
Media are moved between regular and free pools on two events (Figure
36 on page 138):
• Allocation. Media are moved from a free pool to a regular pool
• Deallocation. Media are moved from a regular pool to a free pool. You can
specify in the GUI whether deallocation is done automatically. Media from the
PC backup pool in Figure 36 on page 138, for example, are not automatically
deallocated.
Protected (allocated, used) media belong to a specific regular pool (like the SAP
pool), while free Data Protector media can be (automatically) moved to a free pool.
This free pool is later used for allocation of free media for all pools that are configured
to use this free pool.
Some regular pools, for example the Private pool in Figure 36 on page 138, can also
be configured not to share any media with free pools.
Free pool benefits
A free pool has the following benefits:
• Sharing of free media between pools
138
Media management and devices
All free (unprotected, empty) media can be grouped in a free pool and shared
between all media pools that support free pool usage.
• Reduced operator intervention for backup
Assuming that all free media are shared, the need for mount requests is reduced.
Free pool properties
A free pool:
• can be created manually or automatically when you configure the use of one.
You cannot delete free pools if they are linked to a normal pool or are not empty.
• is different from a regular pool in that it does not provide allocation policy options.
• contains only Data Protector media (no unknown or blank media).
Media quality calculation
Media quality is calculated equally between pools. That means that medium condition
factors will be configurable for a free pool only and will be inherited by all pools
using the free pool.
Free pool limitations
Free pools have the following limitations:
• You cannot select different condition factors for each pool. Instead, all pools that
use a free pool use condition factors configured for this free pool.
• You cannot move protected media to a free pool and unprotected media to a
regular pool that has automatic deallocation configured.
• You cannot use some operations such as Import, Copy and Recycle on media in
a free pool.
• Pools with magazine support cannot use a free pool.
• You may experience some temporary inconsistencies in pools when using free
pools, for example, when there is an unprotected medium in a regular pool
waiting for the de-allocation process.
• If you change the protection of media after its expiry (for example to Permanent),
though the media may be in a free pool, they are not allocated for backup.
• When allocated from a free pool, media with different data format type can be
used and are automatically reformatted, for example NDMP media are reformatted
to normal media.
For further information on free pools, see the Data Protector online Help index: “free
pools, characteristics”.
Concepts guide
139
Media pool usage examples
The examples below show some configurations you may want to consider when
choosing the appropriate strategy for a particular backup environment.
Example 1
In the model shown in Figure 37 on page 140, all objects are backed up to the same
media pool. The backup specification does not reference a pool, so the default pool
is used, which is part of the device definition.
Figure 37 A simple one device/one media pool relation
Example 2
Large library devices contain a number of physical drives and media used by different
departments or applications. You can configure a media pool for each department,
as shown in Figure 38 on page 141, and decide which drive in the library will handle
the actual data transfer. The arrow pointing from a backup specification to a media
pool indicates that you defined a target media pool in a backup specification. If you
do not specify a media pool in the backup specification, the default pool, specified
in the device definition, is used.
For details about the relation between media pools and large library devices, see
“Large libraries” on page 162.
140
Media management and devices
Figure 38 Configuration of media pools for large libraries
Example 3
Figure 39 on page 142 shows an example when data is backed up to media in a
media pool with multiple devices simultaneously. Higher performance is achieved
due to the use of several devices in parallel, regardless of which pool is used.
For more information, see “Device lists and load balancing” on page 155.
Concepts guide
141
Figure 39 Multiple devices, single media pool
Example 4
Data is backed up to media in multiple media pools on multiple devices
simultaneously. If you want to use the same device with different pools, you need to
create several backup specifications. In the example below, a separate media pool
is dedicated to each database application.
142
Media management and devices
Figure 40 Multiple devices, multiple media pools
Implementing a media rotation policy
What is a media rotation policy?
A media rotation policy defines how media are used during backup, including the
following. In defining a media rotation policy, answer the following questions:
•
•
•
•
•
How many backup generations are needed?
Where are media stored?
How often media are used?
When can media be overwritten and re-used for new backups?
When are media old enough to be replaced?
Traditional backup strategies used with older backup tools required a thoroughly
planned and well defined media rotation policy controlled by the administrator rather
than a backup application. With Data Protector, you can implement a rotation policy
by specifying usage options such that media selection for subsequent backups is
done automatically.
Concepts guide
143
Media rotation and Data Protector
Automatic media rotation and media handling
Data Protector automates media rotation and media handling as follows:
• Because media are grouped into media pools, you no longer need to manage
single media. Data Protector automatically tracks and manages each single
medium in the media pools.
• You do not need to decide to which media the backed up data is to be written
to; Data Protector does that for you. You back up to a media pool.
• Data Protector automatically selects media from a media pool according to the
media allocation policy and usage options you specified. You can also disable
the automatic selection and perform manual media selection.
• The location of media is tracked and displayed in the Data Protector user interface
as long as the media are configured in Data Protector.
• Data Protector automatically tracks the number of overwrites on the media and
the age of the media and thus tracks the condition of the media.
• Data Protector provides a security mechanism so that media with protected data
do not get overwritten accidentally by Data Protector.
Media needed for rotation
Estimating the quantity of needed media
The following helps to estimate the quantity of media you might need for a full rotation:
• Determine if the media capacity can be used fully or if some media are
non-appendable and can only be used partially.
• Determine the systems that will be backed up and the media space required for
the related data. For example, you can use backup preview.
• Determine the backup frequency, such as the number of incremental backups
between two full backups.
• Determine the quantity of media needed for one backup generation, where a
backup generation contains a full backup and a sequence of incremental backups
up to the next full backup. Consider also hardware compression if you have
planned to use it with the devices.
• Determine for how long the media will remain protected.
• Calculate the number of backup generations that will have been created before
the first backup generation can be overwritten.
144
Media management and devices
By now you should be able to estimate the quantity of media required for a full media
rotation. Additional media will be required in case you:
• Assume 10% overhead added by Data Protector to the data on the media for
directory and file information. This information is already calculated in the backup
preview size.
• After the media no longer fulfill the usage criteria, they need to be replaced.
• Expect some growth in the volume of data to be backed up.
Media management before backups begin
Before you can use media for backup, media must be initialized, or formatted, for
use with Data Protector. You can either initialize (format) media manually, or you
can let Data Protector automatically initialize (format) media when the media are
selected for backup. See “Selecting media for backups ” on page 147.
Initializing or formatting media
What is initializing (formatting) media?
Before Data Protector uses media for backup, it initializes (formats) the media. This
saves the information about each medium (medium ID, description and location) in
the IDB and also writes this information on the medium itself (to the medium header).
When you initialize (format) media, you also specify to which media pool the media
belong.
If media are not initialized (formatted) before backup, Data Protector can initialize
(format) blank media during backup with the default labels, if the pool policy is set
accordingly. The first backup to such media will take more time. For more information,
see “Selecting media for backups ” on page 147.
Labeling Data Protector media
How Data Protector labels media?
When you add media for use with Data Protector by initializing (formatting) media,
you must specify the media label which helps you identify the media later. If a device
has a barcode reader, the barcode is automatically displayed as a prefix of the
medium description. A barcode provides a unique ID for each medium in the IDB.
You can optionally use the barcode as medium label during the initialization of the
medium.
Concepts guide
145
Data Protector also assigns each medium a media ID that uniquely identifies this
medium.
An ANSI X3.27 label is also written on the tape for identification on other systems.
Data Protector writes these labels with other information to a medium header and to
the IDB.
If you change the medium label, Data Protector modifies the medium label in the IDB
and not on the medium itself. Therefore, if you export and import media that have
not been updated, the medium label in the IDB is replaced with the medium label
from the media. The media label on the tape can be changed only by re-initializing
(formatting) the media.
How are labels used?
These labels identify the medium as a Data Protector medium. When loading a
medium for backup or restore, Data Protector checks the medium for the medium ID.
The media management system maintains the information about this medium, which
tells Data Protector whether the requested action is allowed for this medium. For
example, if you try to write a new backup to this medium, the media management
system checks whether the data protection for the data already contained on this
medium has expired. The user defined label is used to identify a specific medium.
Location field
Backup media are usually stored in different locations. For example, a backup needs
to be available on site for fast restore access, whereas a medium containing a copy
of the backed up data is often stored off-site for safety reasons.
Data Protector provides a location field for each medium, which can be used freely
by the operator(s). This field can help to track the location of the media. Examples
of meaningful location fields would be: In Library, off-site, and vault_1.
The media location setting is also useful if an object version that you want to restore
exists on more than one media set. You can set the media location priority, which
influences the selection of the media set that will be used for the restore. For more
information on the selection of media for restore, see “Selection of the media
set” on page 126.
146
Media management and devices
Media management during backup sessions
What happens during backup?
During a backup session, Data Protector automatically selects media for backup and
keeps track of which data is backed up to which media. This simplifies management
of media so that the operator does not need to know exactly which data was backed
up to which media. Backup objects that have been backed up within the same backup
session represent a media set.
This section provides the following information:
• How Data Protector selects media for backup
• How full and incremental backups are added to the media
• How the condition of media is calculated
For related information, see the following sections:
• “Full and incremental backups” on page 71
• “Media pools” on page 135
Selecting media for backups
Data Protector automatically selects media for backup based on media allocation
policies. This simplifies media management and media handling; a backup operator
does not need to manually administer the media for backup.
Media allocation policy
You can influence how media are selected for backup using the media allocation
policy. You can specify a loose policy, where any suitable medium is used for backup,
including new, blank media or a strict policy, where media must be available in a
predefined order to facilitate balanced media usage. Additionally, you can use a
pre-allocation list.
Pre-allocating media
Data Protector allows you to explicitly specify media from a media pool that you
want to use for a backup using a pre-allocation list. Combine this list with the strict
media allocation policy. In this case, the media are used in the exact order as
specified. If media are not found in this order, Data Protector issues a mount request.
Concepts guide
147
Media condition
The condition of the media also influences which media are selected for backup, for
example, media in good condition are used for backup before media in fair condition.
For more information, see “Calculating media condition” on page 150.
Adding data to media during backup sessions
To maximize space usage of media as well as backup and restore efficiency, you
can select how Data Protector treats the space on the medium left over from the
previous backup. This is defined with a media usage policy.
Media usage policy
The available media usage policies are listed below:
Appendable
A backup session starts writing data to the space
remaining on the last medium used from a
previous backup session. Subsequent media
needed in this session are written from the
beginning of the tape, hence only unprotected or
new tapes can be used. Appending media
conserves media space but can add complexity
to vaulting, because one medium can contain data
from several media sets.
Non Appendable
A backup session starts writing data at the
beginning of the first available medium for
backup. Each medium contains data from a single
session only. This simplifies vaulting.
Appendable of Incrementals
Only
A backup session appends to a medium only if
an incremental backup is performed. This allows
you to have a complete set of full and incremental
backups on the same medium, if there is enough
space.
Distributing objects over media
The following figures show some examples of how objects can be distributed over
media:
148
Media management and devices
Figure 41 Multiple objects and sessions per medium, sequential writes
Figure 41 on page 149 shows an example of eight sequential writes over four sessions,
using the appendable media usage policy. The data was written in four sessions,
one object at a time. The three media belong to the same media pool. Medium_A
and medium_B are already full, while medium_C has still some space left.
Figure 42 Multiple objects and sessions per medium, concurrent writes
Figure 42 on page 149 shows an example of eight objects that have been written
during four sessions with the concurrency settings that allow for simultaneous writes.
In this case, obj_1, obj_2, and obj_3 have been backed up concurrently in sess_1;
obj_4 and obj_5 have been backed up concurrently in sess_2, and so on. Obj_1
could come from system_A and obj_2 from system_B, or they could come from
different disks on the same system. The media usage policy is appendable.
Figure 43 Multiple media per session, multiple media per object
Figure 43 on page 149 shows an example of four backup objects that have been
backed up during two sessions, so that the first pair of backup objects has been
Concepts guide
149
concurrently written in sess_7 and the second one in sess_8. Note that one object
can be stretched over several media. The media usage policy is appendable.
Figure 44 Each object written on a separate medium
Figure 44 on page 150 shows an example of using one backup specification per
object with the non-appendable media usage policy. The result is higher media
consumption. You could combine this with the append incrementals only policy, to
get the incremental backups of the object on the same medium.
For more information on how full and incremental backup policies influence restore
performance and media usage, see “Full and incremental backups” on page 71.
Writing data to several media sets during backup
During a backup session, you can write all or some objects to several media sets
simultaneously, using the Data Protector object mirror functionality. For more
information, see “Object mirroring” on page 119.
Calculating media condition
Media condition factors
Data Protector calculates the state of used media using media condition factors. The
state of the poorest medium in a pool determines the state of the entire pool. For
example, as soon as the state of one medium in a media pool is poor, the state of
the pool becomes poor. When that particular medium is removed from the pool, the
state reverts to either fair or good.
Media can have three states: good, fair, or poor.
On a per medium basis, the following is used for calculating the condition:
• number of overwrites
The usage of a medium is defined as the number of overwrites from the beginning
of the medium. Once the medium has more than the threshold number of
overwrites, it is marked as poor
150
Media management and devices
• media age
The age of a medium is calculated as the number of months that have elapsed
since you formatted, or initialized, the medium. Once a medium is older than the
threshold number of months, it is marked as poor.
• device errors
Some device errors result in the medium being marked as poor. If a device fails
during a backup, the medium used for the backup in this device is marked as
poor.
Media management after backup sessions
Once the data is stored on the media, you must take the right precautions to protect
the media and the data on the media. Consider the following:
• Protecting media from overwrites.
You have specified this when you configured a backup of data, but you can
change this after the backup is done. For more information on data and catalog
protection, see “Keeping backed up data and information about the
data” on page 99.
• Protecting media from physical damage.
Media with permanent data may be stored to a safe place.
• Copying backed up data and keeping the copies at a safe place.
See “Duplicating backed up data” on page 112.
The following sections describe how to vault media and restore from such media.
Vaulting
What is vaulting?
Vaulting is a process of storing media with important information to a safe place,
where they are kept for a specific period of time. The safe place for media is often
called a vault.
Data Protector supports vaulting with the following features:
• Data protection and catalog protection policies.
• Easy selecting and ejecting of media from a library.
• The field media location tells you the physical location where the media are stored.
Concepts guide
151
• A report showing media used for backup within a specified time-frame.
• A report showing which backup specifications have used specified media during
the backup.
• A report showing media stored at a specific location with data protection expiring
in a specific time.
• Displaying a list of media needed for a restore and the physical locations where
the media are stored.
• Filtering of media from the media view based on specific criteria.
Implementing vaulting
The implementation of vaulting depends on your company’s backup strategy and
policies for handling data and media. Generally, it consists of the following steps:
1.
Specifying the desired data protection and catalog protection policies when
configuring backup specifications.
2. Configuring a vault in Data Protector. Essentially, this means specifying a name
for the vault you will use for media, for example: Vault_1.
3. Establishing the appropriate media maintenance policy for media in the vault.
4. Optionally, creating additional copies of the backed up data for vaulting
purposes, using the object mirror functionality during backup, or the object copy
or media copy functionality after backup.
5. Selecting the media you want to store in a vault, ejecting the media and storing
it in the vault.
6. Selecting the media with expired data which is in a vault and inserting the media
in a library.
Vaulting usage example
Your company backup policy, for example, says that you must back up data daily.
Each week a full backup must be stored in a vault where it must be available for the
next five years. You must be able to easily restore data from all the previous year’s
backups stored in the vault. After five years, media from the vault can be re-used.
This implies the following Data Protector settings: a full backup once a week with
daily incrementals. Data protection is set to five years. Catalog protection is set to
one year. Therefore, you will be able to simply browse and restore data for one year
and the data will be available for restore from media for five years. Media from the
full backup are copied and stored to a vault. After one year, Data Protector
automatically deletes detailed information from the IDB about the data on the media,
thus creating more space in the database for new information.
152
Media management and devices
Restoring from media in a vault
Restoring media from a vault is no different than restoring from any other media.
Depending on how your data and catalog protection policies are defined, you may
need to do some additional steps:
1.
Bring media from a vault and insert the media into a device.
2. If the catalog protection for the media is still valid, restore data simply by selecting
what you want to restore using the Data Protector user interface.
If the catalog protection for the media has expired, Data Protector does not have
detailed information about the backed up data. You must restore by manually
specifying the files or directories you want to restore. You can also restore the
complete object to a spare disk and then search for files and directories in the
restored filesystem.
TIP:
To re-read detailed information about the files and directories backed up on the media
once the catalog protection has expired, export the media and import them back. Then
specify that you want to read the detailed catalog data from those media. Now you will
be able to select files and directories in the Data Protector user interface again.
For more information on how data protection and catalog protection policies influence
restores, see “Keeping backed up data and information about the data” on page 99.
Devices
Data Protector supports a number of devices available on the market. For an up-to-date
list of supported devices, see the HP Data Protector product announcements, software
notes, and references.
Using devices with Data Protector
To use a device with Data Protector, you must configure the device in the Data
Protector cell. When you configure a device, you specify a name for the device,
some device specific options, such as barcode or cleaning tape support, and a media
pool. The process of configuring devices is simplified with a wizard that leads you
through all the steps and can even detect and configure devices automatically. The
same physical device can be defined multiple times with different usage properties
Concepts guide
153
in Data Protector using different (logical) device names, for example, one without
hardware data compression and another one with hardware data compression.
The following sections describe some specific device functionality and how Data
Protector operates with various devices.
Library management console support
Many modern tape libraries provide a management console that allows libraries to
be configured, managed, or monitored from a remote system. The scope of tasks
that can be performed remotely depends on the management console implementation,
which is independent of Data Protector.
Data Protector eases access to the library management console interface. The URL
(web address) of the management console can be specified during the library
configuration or re-configuration process. By selecting a dedicated menu item in the
GUI, a web browser is invoked and the console interface is automatically loaded
into it.
For a list of device types for which this feature is available, see the HP Data Protector
product announcements, software notes, and references.
IMPORTANT:
Before using the library management console, consider that some operations which you
can perform through the console may interfere with your media management operations
and/or your backup and restore sessions.
TapeAlert
TapeAlert is a tape device status monitoring and messaging utility that makes it easy
to detect problems that could have an impact on backup quality. From the use of
worn-out tapes to defects in the device hardware TapeAlert provides
easy-to-understand warnings or errors as they arise, and suggests a course of action
to remedy the problem.
Data Protector fully supports TapeAlert 2.0, as long as the connected device also
provides this functionality.
154
Media management and devices
Device lists and load balancing
Multiple devices for backup
When configuring a backup specification, you can specify several standalone devices
or multiple drives in a library device that will be used for the operation. In this case,
the operation is faster because data is backed up in parallel to multiple devices
(drives).
Balancing the use of devices
By default, Data Protector automatically balances the load (the usage) of devices so
that they are used evenly. This is called load balancing. Load balancing optimizes
the usage by balancing the number of the objects backed up to each device. Since
load balancing is done automatically during backup time, you do not have to manage
the allocation of objects to devices used in the session; you just specify the devices
to be used.
When to use load balancing
Use load balancing when:
•
•
•
•
•
You back up a large number of objects.
You use library (autochanger) devices with several drives.
You do not need to know on which media objects will be backed up.
You have a good network connection.
You want to increase the robustness of the backup. Data Protector automatically
redirects the backup operation from failed devices to other devices in a device
list.
When not to use load balancing
Do not use load balancing when:
• You want to back up a small number of large objects. In this case Data Protector
often cannot effectively balance the load among devices.
• You want to explicitly select to which device each object will be backed up.
Device chaining
Data Protector allows you to configure several standalone devices of the same type,
connected to the same system, as a device chain. When a medium in one device
Concepts guide
155
gets full, the backup automatically continues on the medium in the next device in the
device chain.
How load balancing works
For example, assume that there are 100 objects configured for backup to four devices
with concurrency set to three and with load balancing parameters MIN and MAX both
configured at two. If at least two devices are available, the session will start with
three objects being backed up in parallel to each of the first two available devices.
The other 94 objects will be pending and will not be assigned to a particular device
at that time.
Once a backup of a particular object is done, the next pending object is started and
assigned to the device that has less than three concurrent objects being backed up.
Load balancing ensures that the two devices are running in parallel as long as there
are still pending objects to be backed up. If a device fails during backup, one of the
two devices in reserve is used. The objects that were being backed up to the failed
device are aborted, while the next three pending objects are assigned to the new
device. This means that each failure of a device can cause a maximum of three
objects to be aborted, provided that other devices are available for the backup
session to continue.
Device streaming and concurrency
What is device streaming?
To maximize a device performance, it must be kept streaming. A device is streaming
if it can feed enough data to the medium to keep the medium moving forward
continuously. Otherwise, the medium tape has to be stopped while the device waits
for more data. In other words, if the rate at which data is written to the tape is less
than or equal to the rate which data can be delivered to the device by the computer
system, then the device is streaming. In network-focused backup infrastructures, this
deserves attention. For local backups, where disks and devices are connected to the
same system, a concurrency of 1 may suffice if your disks are fast enough.
How to configure device streaming
To allow the device to stream, a sufficient amount of data must be sent to the device.
Data Protector accomplishes this by starting multiple Disk Agents for each Media
Agent that writes data to the device.
156
Media management and devices
Disk agent concurrency
The number of Disk Agents started for each Media Agent is called Disk Agent (backup)
concurrency and can be modified using the Advanced options for the device or when
configuring a backup. Data Protector provides default numbers that are sufficient for
most cases. For example, on a standard DDS device, two Disk Agents send enough
data for the device to stream. For library devices with multiple drives where each
drive is controlled by one Media Agent, you can set the concurrency for each drive
independently.
Increased performance
If properly set, backup concurrency increases backup performance. For example, if
you have a library device with four drives, each controlled by a Media Agent and
each Media Agent receives data from two Disk Agents concurrently, data from eight
disks is backed up simultaneously.
Device streaming is also dependent on other factors, such as network load and the
block size of the data written to the device.
For related information, see “Backup sessions” on page 220.
Multiple data streams
Data Protector allows you to concurrently back up parts of a disk to multiple devices.
This feature is useful for backing up very large and fast disks to relatively slow devices.
Multiple Disk Agents read data from the disk in parallel and send the data to multiple
Media Agents. This method speeds up the backup, but requires that you take into
account the following:
If one mount point was backed up through many Disk Agents, data is contained in
multiple objects. To restore the whole mount point define all parts of the mount point
in a single backup specification and then restore the entire session.
Segment size
A medium is divided into data segments, catalog segments and a header segment.
Header information is stored in the header segment, which is the same size as the
block size. Data is stored in data blocks of data segments. Information about each
data segment is stored in the corresponding catalog segment. This information is first
stored in the Media Agent memory and then written to a catalog segment on the
medium as well as to the IDB. All segments are divided by file marks as shown in
Figure 45 on page 158.
Concepts guide
157
NOTE:
Some tape technologies place limitations on the number of file marks per medium. Ensure
that your segment size is not too low.
Figure 45 Data format
Segment size, measured in megabytes, is the maximum size of data segments. If you
back up a large number of small files, the actual segment size can be limited by the
maximum size of catalog segments. Segment size is user configurable for each device.
It affects the speed of a restore. A smaller segment size leaves less space on the
medium for data, because each segment has a file mark that takes up media space.
However, a larger number of file marks results in faster restores, because a Media
Agent can more quickly locate the segment containing the data to be restored.
Optimal segment size depends on the type of media used in the device and the kind
of data to be backed up. For example, by default the segment size for DLT medium
is 150 MB.
Block size
Segments are not written as a whole unit, but rather in smaller subunits called blocks.
The hardware of a device processes data in units of a device-type specific block size.
Data Protector allows you to adjust the size of the blocks it sends to the device. The
default block size value for all devices is 64 KB.
158
Media management and devices
Increasing the block size can improve performance. Changing the block size should
be done before formatting tapes. For example, a tape written with the default block
size cannot be appended to using a different block size.
NOTE:
Use the same block size for media that can be used with different device types. Data
Protector can only append data to media using the same block size.
Number of disk agent buffers
Data Protector Media Agents and Disk Agents use memory buffers to hold data
waiting to be transferred. This memory is divided into a number of buffer areas (one
for each Disk Agent, depending on device concurrency). Each buffer area consists
of 8 Disk Agent buffers (of the same size as the block size configured for the device).
You can change this value to be anything between 1 and 32, although this is rarely
necessary. There are two basic reasons to change this setting:
• Shortage of memory
The shared memory required for a Media Agent can be calculated as follows:
DAConcurrency*NumberOfBuffers*BlockSize
Reducing the number of buffers from 8 to 4, for instance, results in a 50% reduction
in memory consumption, with performance implications.
• Streaming
If the available network bandwidth varies significantly during backup, then it
becomes more important that a Media Agent has enough data ready for writing
to keep the device in the streaming mode. In this case, increase the number of
buffers.
Device locking and lock names
Device names
When configuring devices for use with Data Protector, you can configure the same
physical device many times with different characteristics simply by configuring the
same physical device in Data Protector with different device names. For example, a
simple standalone DDS device can be configured as a compressed device and then
as an uncompressed device, although this is not recommended.
Concepts guide
159
Physical device collision
When specifying a device used for backup, you may specify one device name in
one backup specification and another device name of the same physical device in
a different backup specification. Depending on the backup schedule, this may result
in Data Protector trying to use the same physical device in several backup sessions
at the same time, thus creating a collision.
Preventing collision
To prevent this collision, specify a virtual lockname in both device configurations.
Data Protector checks if the devices have the same lockname and prevents collision.
For example, a DDS device is configured as a compressed device named DDS_C,
and as a non-compressed device DDS_NC as shown in Figure 46 on page 160.
Specify the same lockname, DDS, for both devices.
Figure 46 Device locking and device names
Standalone devices
What are standalone devices?
Standalone devices are devices with one drive that reads/writes to one medium at
time.
Standalone devices are used for small scale backups or special backups. When the
medium is full, the operator must manually replace it with a new medium for the
backup to proceed.
Data Protector and standalone devices
Once you have connected a device to the system, you use the Data Protector user
interface to configure the device for use with Data Protector. To do this, you must first
160
Media management and devices
install a Data Protector Media Agent on the system with the device connected. Data
Protector can detect and automatically configure most standalone devices.
During a backup, Data Protector issues a mount request when the medium in a device
is full. The operator must replace the medium for the backup to continue.
What are device chains?
Data Protector allows you to configure multiple standalone devices to a device chain.
When a medium in one device gets full, the backup automatically continues on the
medium in the next device in the device chain.
Device chains allow running unattended backups using several standalone devices
without having to manually insert/eject media when the media are full.
Stacker devices
Stacker devices, similar to device chains, contain a number of media that are used
in a sequential order. When a medium gets full, the next medium is loaded and used
for backup.
Small magazine devices
What are magazine devices?
Magazine devices group a number of media into a single unit called a magazine.
Data Protector treats the magazine as if it were a single medium. A magazine has
a larger capacity than a single medium and is easier to handle than several single
media. For a list of supported devices, see the HP Data Protector product
announcements, software notes, and references.
Data Protector and magazine devices
Data Protector allows you to perform media management tasks on magazines as
sets, emulating single media by providing magazine and media views, or on a single
medium.
You can alternatively use magazine devices as normal libraries without using Data
Protector magazine support. Data Protector can detect and automatically configure
magazine devices.
Cleaning dirty drives
Using cleaning tapes, Data Protector can automatically clean magazines and other
devices when they get dirty.
Concepts guide
161
Large libraries
What are library devices?
Library devices are automated devices, also called autoloaders, exchangers or
jukeboxes. In Data Protector, most libraries are configured as SCSI libraries. They
contain a number of media cartridges in a device’s repository and can have multiple
drives writing to multiple media at a time.
A typical library device has a SCSI ID for each drive in the device and one for the
library robotic mechanism that moves media from slots to drives and back. For
example, a library with four drives has five SCSI IDs, four for the drives and one for
the robotic mechanism.
Data Protector also supports silo libraries, such as HP StorageWorks Libraries,
StorageTek/ACSLS and ADIC/GRAU AML. For a list of supported devices, see the
HP Data Protector product announcements, software notes, and references.
Handling of media
The Data Protector user interface provides a special library view, which simplifies
managing library devices.
Media in a large library device can all belong to one Data Protector media pool, or
they can be split into several pools.
Configuring a library
When configuring a device, you configure the slot range you want to assign to Data
Protector. This allows sharing of the library with the other application. The assigned
slots may contain blank (new) media, Data Protector or non-Data Protector media.
Data Protector checks the media in the slots and displays the information about the
media in the library view. This allows you to view all kinds of media, not just the
media used by Data Protector.
Size of a library
The following may help you estimate the size of the library you need:
• Determine if you need to distribute the media to several locations or keep them
in a central location.
162
Media management and devices
• Obtain the number of required media. See “Implementing a media rotation
policy” on page 143.
Sharing a library with other applications
A library device can be shared with other applications storing data to media in the
device.
You can decide which drives from the library you want to use with Data Protector.
For example, out of a four-drive library you may choose to use only two drives with
Data Protector.
You can decide which slots in the library you want to manage with Data Protector.
For example, out of the 60 slots library you might use slots 1-40 with Data Protector.
The remaining slots would then be used and controlled by a different application.
Sharing of the library with other applications is especially important with large HP
libraries and silo libraries, such as StorageTek/ACSLS or ADIC/GRAU AML devices.
Enter / eject mail slots
Library devices provide special enter/eject mail slots an operator uses to enter or
eject media to or from the device. Depending on the device, more than one enter/eject
slot can be provided. In case of a single mail slot, media are inserted one by one,
while in case of multiple mail slots, a particular number of slots can be used in one
enter/eject operation.
Data Protector allows you to enter/eject several media in one step. For example,
you can select 50 slots in the device and eject all media in one action. Data Protector
will automatically eject media in the correct order for the operator to remove the
media from the enter/eject mail slot.
For more information, see the documentation about your device.
Barcode support
Data Protector supports library devices with a barcode reader. In these devices, each
medium has a barcode that uniquely identifies media.
Advantages of barcodes
Barcodes enable Data Protector to significantly improve media recognition, labeling,
and cleaning tape detection.
Concepts guide
163
• Scanning the barcodes of the media in a device’s repository is faster, because
Data Protector does not need to actually load the media to a drive and read the
medium header.
• A barcode is automatically read by Data Protector and used to identify the media.
• A cleaning tape is automatically detected if it has a CLN barcode prefix.
• A barcode is a unique identifier for media in the IDB. You cannot have duplicate
barcodes in your environment.
TIP:
You can optionally use the barcode as medium label during the initialization of the
medium.
Cleaning tape support
HP Data Protector provides automatic cleaning for most devices using a cleaning
tape. This medium will be used automatically by Data Protector if a dirty drive event
from the device is detected.
• For SCSI libraries it is possible to define which slot holds a cleaning tape.
• For devices with a barcode reader, Data Protector recognizes cleaning tape
barcodes automatically if they have the CLN prefix.
• For devices without a cleaning tape, a dirty drive detection will cause a cleaning
request to be displayed on the session monitor window. The operator must clean
the device manually.
You cannot continue your backup without cleaning the drive, since the backup
may fail because data may not be correctly written and stored on the media.
Sharing a library with multiple systems
What is library sharing?
Device sharing allows you to connect different drives of a physical library to different
systems. These systems can then perform local backups to the library. The result is
significantly higher backup performance and less network traffic. To enable library
sharing, the drives in the library must have the possibility to connect to separate SCSI
buses. This is useful with high performance libraries to allow the drive to receive data
in a continuous stream from multiple systems, further enhancing performance. Data
164
Media management and devices
Protector internally redirects the robotic commands to the system that manages the
robotics.
Figure 47 Connecting drives to multiple systems
Control protocols and Data Protector Media Agents
The drives in the library must be able to physically connect to different systems that
have a Data Protector Media Agent (the General Media Agent or the NDMP Media
Agent) installed.
With Data Protector, there are two types of protocols used for drive control:
• SCSI—for SCSI or Fibre Channel connected drives.
This protocol is implemented in both the General Media Agent and in the NDMP
Media Agent.
• NDMP—for NDMP dedicated drives.
This protocol is implemented in the NDMP Media Agent only.
On the other hand, there are four types of protocols used for library robotic control:
• ADIC/GRAU—for ADIC/GRAU library robotics
• StorageTek ACS—for StorageTek ACS library robotics
Concepts guide
165
• SCSI—for robotics other libraries
• NDMP—for NDMP robotics
All four library robotic control protocols are implemented in both the General Media
Agent and in the NDMP Media Agent.
Drive control
Any Data Protector client system configured to control a drive in a library (regardless
of the drive control protocol and platform used) can communicate with any Data
Protector client system configured to control the robotics in the library (regardless of
the robotics control protocol and platform used). Thus, it is possible to share drives
in any supported library among Data Protector clients systems on various platforms
using various robotic and drive protocols. The NDMP Media Agent is needed only
on client systems controlling the backup of an NDMP server (on client systems
configured for NDMP dedicated drives). In all other cases the two Data Protector
Media Agents are interchangeable.
Table 10 on page 166 show the Data Protector Media Agent (the General Media
Agent or the NDMP Media Agent) required on client systems configured for drive
control of a library with drives shared among multiple client systems.
Table 10 Required Data Protector Media Agent for drive control
Drive control protocol
Robotic control
protocol
(ADIC/GRAU,
StorageTek ACS,
SCSI, or NDMP)
NDMP
SCSI
NDMP Media Agent
NDMP Media Agent or General
Media Agent
Robotic control
A Data Protector client system controlling the library robotics can have either the
General Media Agent or the NDMP Media Agent installed, regardless of the type
of drive protocol (NDMP or SCSI) used with the drives in the library.
166
Media management and devices
Table 11 on page 167 show the Data Protector Media Agent (the General Media
Agent or the NDMP Media Agent) required on a client system configured for robotic
control of a library with drives shared among multiple client systems.
Table 11 Required Data Protector Media Agent for robotic control
Robotic control protocol
Drive control protocol
(NDMP or SCSI)
ADIC/GRAU
StorageTek
ACS
SCSI
NDMP
NDMP Media
Agent or
General
Media Agent
NDMP Media
Agent or
General
Media Agent
NDMP Media
Agent or
General
Media Agent
NDMP Media
Agent or
General
Media Agent
Exemplary configurations
Figures Figure 48 on page 168 to Figure 50 on page 170 show exemplary
configurations of shared drives in libraries and Data Protector Media Agents
distributions in such configurations.
Concepts guide
167
Figure 48 Sharing a SCSI library (robotics attached to a Data Protector
Client System)
Figure 48 on page 168 shows a SCSI library, with its robotics attached to and
configured on the Data Protector client system with either the General Media Agent
or the NDMP Media Agent installed. The SCSI robotic control protocol is used by
the General Media Agent or the NDMP Media Agent on the client. The Data Protector
client system with the attached robotics can also have one or more drives attached.
The NDMP dedicated drive in the library is configured on the Data Protector client
system with the NDMP Media Agent installed. The NDMP drive control protocol is
used by the NDMP Media Agent on the client.
Another drive in the library is configured on and attached to the Data Protector client
system with either the General Media Agent or the NDMP Media Agent installed.
The SCSI drive control protocol is used by the General Media Agent or the NDMP
Media Agent on the client.
168
Media management and devices
Figure 49 Sharing a SCSI library (robotics attached to an NDMP Server)
Figure 49 on page 169 shows a SCSI library, with its robotics attached to an NDMP
Server and configured on the Data Protector client system with either the General
Media Agent or the NDMP Media Agent installed. The SCSI robotic control protocol
is used by the General Media Agent or the NDMP Media Agent on the client. The
NDMP Server with the attached robotics can also have one or more drives attached.
IMPORTANT:
If the NDMP Server with the attached robotics also have an NDMP dedicated drive
attached, the Data Protector client system on which the robotics and the NDMP dedicated
drive are configured, can only have the NDMP Media Agent installed, since the NDMP
drive control protocol is used for the NDMP dedicated drive.
The NDMP dedicated drive in the library is configured on the Data Protector client
system with the NDMP Media Agent installed. The NDMP drive control protocol is
used by the NDMP Media Agent on the client.
Another drive in the library is configured on and attached to the Data Protector client
system with either the General Media Agent or the NDMP Media Agent installed.
The SCSI drive control protocol is used by the General Media Agent or the NDMP
Media Agent on the client.
Concepts guide
169
Figure 50 Sharing an ADIC/GRAU or StorageTek ACS library
Figure 50 on page 170 shows an ADIC/GRAU or StorageTek ACS library, with its
robotics attached to an ADIC/GRAU or StorageTek ACS Server and configured on
the Data Protector client system with either the General Media Agent or the NDMP
Media Agent installed. The ADIC/GRAU robotic control protocol is used by the
General Media Agent or the NDMP Media Agent on the client. The ADIC/GRAU or
the StorageTek ACS Server can also have one or more drives attached.
The NDMP dedicated drive in the library is configured on the Data Protector client
system with the NDMP Media Agent installed. The NDMP drive control protocol is
used by the NDMP Media Agent on the client.
Another drive in the library is configured on and attached to the Data Protector client
system with either the General Media Agent or the NDMP Media Agent installed.
The SCSI drive control protocol is used by the General Media Agent or the NDMP
Media Agent on the client.
Data Protector and Storage Area Networks
Where and how you store data in your enterprise may have a serious impact on
your business. Information is becoming increasingly mission-critical to most companies.
Today, terabytes of data must be accessible to users across the network. The Data
Protector implementation of SAN-based Fibre Channel technology provides you with
the data storage solution you need.
170
Media management and devices
Storage Area Networks
A Storage Area Network (SAN), depicted in Figure 51 on page 172, is a new
approach to network storage that separates storage management from server
management with a network devoted to storage.
A SAN provides any-to-any connectivity for all network resources, thus enabling
device sharing between multiple client systems and increasing data traffic performance
as well as the availability of devices.
The SAN concept allows the exchange of information between multiple data storage
devices and servers. The servers can access data directly from any device and do
not need to transfer data over the conventional LAN. A SAN consists of servers,
backup devices, disk arrays, and other nodes, all connected with a fast network
connection, typically Fibre Channel. This additional network provides off-loading
storage operations from the conventional LAN to a separate network.
Data Protector’s direct backup functionality is a productive application of SAN and
fibre channel technology.
Concepts guide
171
Figure 51 Storage Area Network
Fibre Channel
Fibre Channel is an ANSI standard for high-speed computer interconnection. Using
either optical or copper cables, it allows the bidirectional transmission of large data
files at up to 4.25 gigabits per second, and can be deployed between sites within
a 30 kilometer range. Fibre Channel is the most reliable, highest performance solution
for information storage, transfer, and retrieval available today.
Fibre Channel connects nodes using three physical topologies that can have variants:
• Point-to-point
• Loop
• Switched
172
Media management and devices
Point-to-point, loop, and switched Fibre Channel topologies can be mixed to best suit
your connectivity and growth requirements.
For a list of supported configurations, see the HP Data Protector product
announcements, software notes, and references or http://www.hp.com/support/
manuals.
Point-to-point topology
This topology allows the connecting of two nodes, typically a server and a backup
device. It provides the basic benefit of improved performance and longer distances
between nodes.
Loop topology
The loop topology is based on the Fibre Channel Arbitrated Loop (FC-AL) standard,
which allows the connecting of up to 126 nodes. Nodes include servers, backup
devices, hubs, and switches. Any node in a loop can communicate with any other
node in the loop, and all nodes share the same bandwidth. An FC-AL loop is typically
implemented using an FC-AL hub with automatic port by-pass. Automatic port by-pass
allows the hot-plug of nodes into the loop.
LIP
A Loop Initialization Primitive (Protocol) (LIP) may be triggered by a number of causes,
most common being the introduction of a new device. The new device could be a
former participant that has been powered on or an active device that has been moved
from one switch port to another. A LIP occurrence can cause an undesirable disruption
of an ongoing process on the SAN, for example, a tape backup operation. It resets
the SCSI bus connecting the SCSI/FC Bridge and the node (SCSI device). See Figure
52 on page 174.
In the case of a backup or restore, a SCSI bus reset is registered as a write error.
Data Protector aborts all operations upon write errors. In the case of backups, it is
recommended to (copy the information already backed up on the medium and then)
reformat the medium and restart the backup.
Concepts guide
173
Figure 52 Loop initialization protocol
Switched topology
The switched topology provides any-to-any connectivity between all nodes connected
to a switch. Switches are easy to install and use, because the Fibre Channel protocol
provides self-configuration and self-management. Switches automatically detect what
is connected (nodes, FC-AL Hubs or other FC switches), and configure themselves
accordingly. Switches provide scaled bandwidth to connected nodes. The switched
topology provides real hot-plug of nodes.
NOTE:
Hot-plug refers to protocol capabilities such as reset, re-establish communication, and
so on. Take into account that ongoing data transfers are interrupted during hot-plug and
that some devices, such as tape devices, cannot handle this behavior. Connecting nodes
to or disconnecting nodes from a loop is likely to interrupt your backup or restore process
and cause the operation to fail. Connect or disconnect nodes from loops only when there
are no running backups or restores using the related hardware.
174
Media management and devices
Device sharing in SAN
Data Protector supports the SAN concept by enabling multiple systems to share
backup devices in the SAN environment. The same physical device can be accessed
from multiple systems. Thus, any system can perform a local backup on some device
or any other device. Because data is transferred over the SAN, backups do not need
any bandwidth on your conventional LAN. This type of backup is sometimes referred
to as a “LAN-free” backup. Backup performance is also improved, because
SAN-based Fibre Channel technology typically provides an order of magnitude
higher throughput than LAN technologies.
You need to prevent several computer-systems from writing to the same device at the
same time. This can become even more complex when devices are used from several
applications. Access to the devices needs to be synchronized between all systems
involved. This is done using locking mechanisms.
SAN technology provides an excellent way to manage the robotics of a library from
multiple systems. This allows the option to manage the robotics from one system
(classic) or allow each system that uses the library to access the robotics directly,
provided the requests to the robotics are synchronized between all the systems
involved.
Configuring multiple paths to physical devices
A device in a SAN environment is usually connected to several clients and can thus
be accessed through several paths, that is client names and SCSI addresses (device
files on UNIX). Data Protector can use any of these paths. You can configure all paths
to a physical device as a single logical device - multipath device.
For example, a device is connected to client1 and configured as /dev/rs1 and
/dev/rs2, on client2 as /dev/r1s1 and on client3 as scsi1:0:1:1. Thus,
it can be accessed through four different paths: client1:/dev/rs1,
client1:/dev/rs2, client2:/dev/r1s1 and client3:scsi1:0:1:1. A
multipath device therefore contains all four paths to this tape device.
Concepts guide
175
Figure 53 Example multipath configuration
Why use multiple paths
With previous versions of Data Protector, a device could be accessed from only one
client. To overcome this problem, several logical devices had to be configured for a
physical device using a lock name. Thus, if you were using lock names for configuring
access from different systems to a single physical device, you had to configure all
devices on every system. For example, if there were 10 clients which were connected
to a single device, you had to configure 10 devices with the same lock name. With
this version of Data Protector, you can simplify the configuration by configuring a
single multipath device for all paths.
Multipath devices increase system resilience. Data Protector will try to use the first
defined path. If all paths on a client are inaccessible, Data Protector will try to use
paths on the next client. Only when none of the listed paths is available, the session
aborts.
Path selection
During a backup session, the device paths are selected in the order defined during
the device configuration, except if a preferred client is selected in the backup
specification. In this case, the preferred client is used first.
176
Media management and devices
During a restore session, the device paths are selected in the following order:
1.
Paths that are on the client to which the objects are restored, if all objects are
restored to the same target client
2. Paths that were used for backup
3. Other available paths
If direct library access is enabled, local paths (paths on the destination client) are
used for library control first, regardless of the configured order.
Backward compatibility
Devices configured with previous versions of Data Protector are not reconfigured
during the upgrade and can be used as in previous releases of Data Protector without
any changes. To utilize the new multipath functionality, you must reconfigure devices
as multipath devices.
Device locking
Locking devices must cover the possibility of several applications using the same
device, as well as only Data Protector using a device by sending data and commands
to it from several systems. The purpose of locking is to ensure that only one system
at a time communicates with a device that is shared between several systems.
Device locking with multiple applications
If Data Protector and at least one other application want to use the same device from
several systems, the same (generic) device locking mechanism has to be used by
each application. This mechanism needs to work across several applications. This
mode is not currently supported by Data Protector. Should this be required, operational
rules must ensure exclusive access to all devices from only one application at a time.
Device locking within Data Protector
If Data Protector is the only application that uses a drive, but that same drive needs
to be used by several systems, Device Locking has to be used.
If Data Protector is the only application that uses a robotics control from several
systems, Data Protector handles this internally, provided the library control is in the
same cell as all the systems that need to control it. In such a case, all synchronization
of access to the device is managed by Data Protector internal control.
Concepts guide
177
Indirect and Direct Library Access
Upon configuring Data Protector with a SCSI Library device, there are two ways in
which client systems can access library robotics: Indirect Library Access and Direct
Library Access.
Indirect Library Access
This configuration can be used in SAN as well as conventional SCSI direct connect
environments. Several systems can access the library robotics by forwarding their
requests to a client system that has direct access to the library robotics. This is called
Indirect Library Access. In the example depicted in Figure 54 on page 179, two client
systems are attached to an HP StorageWorks DLT multidrive library. The client system
castor controls the robotics and the first drive, while the client system pollux
controls the second drive. A Data Protector Media Agent on pollux communicates
with a process running on castor to operate the robotics. This Data Protector library
sharing feature is used automatically when the hostnames of the library and drive
are different.
178
Media management and devices
Figure 54 Indirect Library Access
Note that you cannot use a shared library if the client system that controls the robotics,
castor, in our example, fails.
Direct Library Access
When the SAN concept is used, Data Protector can be configured with a SCSI Library
so that each client system has its own access to library robotics and drives. This is
called Direct Library Access
There is no single “controlling client system” for the robotics: a failure of the system
controlling the robotics does not exclude any other system from using the library. This
is performed without reconfiguration. Several client systems can be used to control
the robotics.
Figure 55 on page 180 shows an HP StorageWorks DLT multidrive library attached
via a SAN to two client systems. Both client systems have access to the library and
to both drives. The SCSI protocol is used for communication with the library.
Concepts guide
179
Figure 55 Direct Library Access
Device sharing in clusters
Clustering, which is often used in combination with the SAN concept, is based on
sharing network resources (for example network names, disks, and tapes devices)
between nodes.
Cluster-aware applications can at any time run on any node in a cluster (they run on
virtual hosts). To perform a local backup of such an application, you need to configure
devices with virtual hostnames instead of real node names. Configure as many devices
for each physical device as you need, using the Lock Name device locking
mechanism. For details, see “Device locking” on page 177.
Static drives
Static drives are devices that are configured on a real node in a cluster. They can
be used to back up data from systems with disks that are not shared. However, they
are not useful for backing up cluster-aware applications, because such application
can run on any node in the cluster.
180
Media management and devices
Floating drives
Floating drives are device that are configured on a virtual host, using virtual system
names. Floating drives should be configured for the backup of cluster-aware
applications. This ensures that no matter on which node in the cluster the application
is currently running, Data Protector always starts a Media Agent on that same node.
Concepts guide
181
182
Media management and devices
4 Users and user groups
In this chapter
This chapter discusses Data Protector security, users, user groups, and user rights.
It is organized as follows:
“Increased security for Data Protector users” on page 183
“Users and user groups” on page 184
Increased security for Data Protector users
Data Protector provides advanced security functionality that prevents unauthorized
backing up or restoring of data. Data Protector security involves hiding data from
unauthorized users, data encoding, and restricted grouping of users according to
their responsibilities.
This section describes security issues related to using Data Protector for backing up
data, restoring data, or monitoring the progress of backup sessions.
Access to backed up data
Backing up and then restoring data is essentially the same as copying data. Therefore,
it is important to restrict access to this data to authorized users only.
Data Protector provides the following user-related security:
• All users intent on using any of the Data Protector functionality must be configured
as Data Protector users.
Visibility of backed up data
• Backed up data is hidden from other users, except the backup owner. Other users
do not even see that data was backed up. For example, if the backup operator
has configured a backup, only the backup operator or the system administrator
Concepts guide
183
can see and restore the backed up data. You can make data visible to other users
using the Data Protector Public option. For instructions, see the Data Protector
online Help.
Users and user groups
To use Data Protector, you must be added to the Data Protector configuration as a
Data Protector user with certain privileges. Note that adding a new user is not a
prerequisite for backing up the system this user is using.
Users are grouped into user groups with specific user rights, for example, to monitor
sessions in the cell, configure backups, and restore files.
Predefined user groups
To simplify the configuration of your backup, Data Protector provides predefined
user groups with specific rights to access Data Protector functionality. For example,
only members of the admin user group can access all Data Protector functionality.
Operators can, by default, start and monitor backups.
TIP:
In small environments, only one person is required to perform all backup tasks. This
person must be a member of the Data Protector admin user group. In this case, there is
no need to add other users to the Data Protector configuration.
Depending on your environment, you may decide to use the default Data Protector
user groups, modify them, or create new ones.
Default administrators
During installation, the following users are automatically added to the Data Protector
admin user group:
• UNIX root user on the UNIX Cell Manager system
• User installing Data Protector on the Windows Cell Manager system
This allows them to configure and use the complete Data Protector functionality. For
more information, see the online Help index: “user groups, admin“.
184
Users and user groups
Using predefined user groups
The following default groups are provided by Data Protector:
Table 12 Data Protector predefined user groups
User group
Access rights
Admin
Allowed to configure Data Protector and perform backup, restore,
and all other available operations.
Operator
Allowed to start backups and respond to mount requests.
End-user
Allowed to perform restore of their own objects. In addition, users
can monitor and respond to mount requests for their own restore
sessions.
NOTE:
Admin capabilities are powerful. A member of the Data Protector admin user group has
system administrator privileges on all the clients in the Data Protector cell.
Data Protector user rights
Data Protector users have the Data Protector user rights of the user group they belong
to. For example, all members of the admin user group have the rights of the Data
Protector admin user group.
When configuring a user from the Windows domain in Data Protector running on
the UNIX Cell Manager, the user must be configured with the Domain Name or the
wildcard group "*".
For a detailed description of the Data Protector user rights for each user group, see
the online Help.
Concepts guide
185
186
Users and user groups
5 The Data Protector internal
database
In this chapter
This chapter describes the Data Protector internal database (IDB) architecture, as well
as its usage and operation. Explanations of the database parts and their records are
presented, along with recommendations on how to manage database growth and
performance, including formulas for calculating its size. This information is needed
to effectively administer the database configuration and maintenance.
It is organized as follows:
“About the IDB” on page 187
“IDB architecture” on page 189
“IDB operation” on page 194
“Overview of IDB management” on page 196
“IDB growth and performance” on page 197
About the IDB
What is the Data Protector Internal Database (IDB)?
The IDB is an embedded database, located on the Cell Manager, which keeps
information regarding what data is backed up, on which media it resides, the result
of backup, restore, copy, object consolidation, and media management sessions,
and what devices and libraries are configured.
Why is the IDB used?
There are three key reasons for using the IDB:
Concepts guide
187
• Fast and convenient restore The information stored in the IDB enables you to
quickly find the media required for a restore, and therefore makes the restore
much faster. It also offers you the convenience of being able to browse for files
and directories to be restored.
• Backup management The information stored in the IDB enables you to verify how
backups were done. You can also configure various reports using the Data
Protector reporting functionality.
• Media management The information stored in the IDB enables Data Protector to
allocate media during backup, copy, and object consolidation sessions, track
media attributes, group media in different media pools, and track media locations
in tape libraries.
IDB size and growth consideration
The IDB can grow very big and have a significant impact on backup performance
and the Cell Manager system. Therefore, the Data Protector administrator must
understand the IDB and, according to needs, decide which information to keep in
the IDB and for how long. It is the administrator’s task to balance between restore
time and functionality on the one hand, and the size and growth of the IDB on the
other. Data Protector offers two key parameters to assist in balancing your needs:
logging level and catalog protection. See also “IDB growth and
performance” on page 197.
The IDB on the Windows Cell Manager
IDB location
The IDB on the Windows Cell Manager is located in the directory
Data_Protector_program_data\db40 (Windows Server 2008) or
Data_Protector_home\db40 (other Windows systems).
IDB format
The IDB on the Windows Cell Manager stores all text information in UNICODE,
double-byte format. Therefore, the IDB grows slightly faster than the IDB on the UNIX
Cell Manager, which stores information in the ASCII format.
The UNICODE format allows for full support of filenames and messages localized to
other languages.
188
The Data Protector internal database
The IDB on the UNIX Cell Manager
IDB location
The IDB on the UNIX Cell Manager is located in the /var/opt/omni/server/db40
directory.
IDB format
The IDB on the HP-UX and Solaris Cell Manager stores all text information in ASCII
single- and multi-byte formats.
The ASCII format limits the support of filenames and messages localized to other
languages. When backing up files with filenames in a double-byte format, such as
UNICODE, the filenames are converted to the ASCII format and may not appear
correctly in the Data Protector user interface. However, the files and filenames will
be restored correctly.
For more information, see “Internationalization” on page 340.
The IDB in the Manager-of-Managers environment
In the Manager-of-Managers (MoM) environment, you can use the Centralized Media
Management Database (CMMDB), which allows you to share devices and media
across several cells. For more information on the MoM functionality, see “Enterprise
environments” on page 45.
IDB architecture
The IDB consists of the following parts:
• MMDB (Media Management Database)
• CDB (Catalog Database), divided into two parts: filenames and other CDB records
• DCBF (Detail Catalog Binary Files)
• SMBF (Session Messages Binary Files)
• SIBF (Serverless Integrations Binary Files for the NDMP integration)
Each of the IDB parts stores certain specific Data Protector information (records),
influences IDB size and growth in different ways, and is located in a separate directory
on the Cell Manager. See Figure 56 on page 190.
Concepts guide
189
For robustness considerations and recommendations for optimizing robustness by
relocating some IDB directories, see the online Help index: “robustness of IDB“.
Underlying technology
The MMDB and CDB parts are implemented using an embedded database consisting
of tablespaces. This database is controlled by the RDS database server process. All
changes to the MMDB and CDB are updated using transaction logs. The transaction
logs are stored in the db40\logfiles\syslog directory. The CDB (objects and
positions) and the MMDB parts represent the core part of the IDB.
The DCBF, SMBF and SIBF parts of the IDB consist of binary files. Updates are direct
(no transactions).
Figure 56 IDB parts
Media Management Database (MMDB)
MMDB records
The Media Management Database stores information about the following:
• Configured devices, libraries, library drives, and slots
• Data Protector media
• Configured media pools and media magazines
190
The Data Protector internal database
MMDB size and growth
The MMDB does not grow very big in size. The largest portion of the MMDB is
typically occupied by information about the Data Protector media. Space consumption
is in the range of 30 MB. Ffor more details , see “IDB size estimation ” on page 203.
MMDB location
The MMDB is located in the following directory:
• On Windows Server 2008:
Data_Protector_program_data\db40\datafiles\mmdb
• On other Windows systems: Data_Protector_home\db40\datafiles\mmdb
• On UNIX systems: /var/opt/omni/server/db40/datafiles/mmdb
Catalog Database (CDB)
CDB records
The Catalog Database stores information about the following:
• Backup, restore, copy, object consolidation, and media management sessions.
This is a copy of the information sent to the Data Protector Monitor window.
• Backed up objects, their versions, and object copies.
• Positions of backed up objects on media. For each backed up object, Data
Protector stores information about the media and data segments used for the
backup. The same is done for object copies and object mirrors.
• Pathnames of backed up files (filenames) together with client system names.
Filenames are stored only once per client system. The filenames created between
backups are added to the CDB.
Filename size and growth
The biggest and fastest growing part of the CDB is the filenames part. It typically
occupies 20% of the entire database. The growth of the filenames part is proportional
to the growth and dynamics of the backup environment, and not to the number of
backups.
A file or directory on the HP-UX or Solaris Cell Manager occupies approximately
50-70 bytes, and a file or directory on the Windows Cell Manager occupies 70-100
bytes in the IDB.
Filenames are stored in the fnames.dat file and in some other files, depending on
the filename length. The maximum size of each of these files is 2 GB. You are notified
Concepts guide
191
when one of these files starts running out of space, so that you can add new files to
extend the size of the filenames part of the IDB.
Size and growth for CDB (objects and positions)
The CDB records other than filenames occupy a minor share of space in the IDB.
Space consumption is in the range of 100 MB for a medium size backup environment.
For more details, see “IDB size estimation ” on page 203.
CDB location
The CDB is located in the following directory:
• On Windows Server 2008:
Data_Protector_program_data\db40\datafiles\cdb
• On other Windows systems: Data_Protector_home\db40\datafiles\cdb
• On UNIX systems: /var/opt/omni/server/db40/datafiles/cdb
Detail Catalog Binary Files (DCBF)
DCBF information
The Detail Catalog Binary Files part stores file version information. This is information
about backed up files, such as file size, modification time, attributes/protection, and
so on.
One DC (Detail Catalog) binary file is created for each Data Protector medium used
for backup. When the medium is overwritten, the old binary file is removed and a
new one is created.
DCBF size and growth
In an environment where filesystem backups using the Log all option are typical, the
DCBF occupies the largest part (typically 80%) of the IDB. To calculate the size of
DCBF, use the following formula: dcbf_file_in_bytes is approximately
num_of_files_on_tape x 30_bytes . Logging level and catalog protection
can be used to specify what is actually stored in the IDB and for how long. See “IDB
growth and performance: key tunable parameters” on page 198.
By default, one DC directory, db40\dcbf, is configured. Its default maximum size
is 16 GB. You can create more DC directories and have them on different disks on
the Cell Manager, thus extending IDB size. The maximum number of supported
directories per cell is 50.
192
The Data Protector internal database
DCBF location
By default, the DCBF is located in the following directory:
• On Windows Server 2008: Data_Protector_program_data\db40\dcbf
• On other Windows systems: Data_Protector_home\db40\dcbf
• On UNIX systems: /var/opt/omni/server/db40/dcbf
Consider the disk space on the Cell Manager and relocate the DC directory, if
necessary. You can create more DC directories and locate them to different disks.
Create several DC directories only if the number of media/DC binary files grows
very large (several thousand) or if you have space problems. For more information,
see the online Help index: “DC directories“.
Session Messages Binary Files (SMBF)
SMBF records
The Session Messages Binary Files stores session messages generated during backup,
restore, copy, and media management sessions. One binary file is created per
session. The files are grouped by year and month.
SMBF size and growth
The SMBF size depends on the following:
• The number of sessions performed, since one binary file is created per session.
• The number of messages in a session. One session message occupies
approximately 200 bytes on Windows and 130 bytes on UNIX systems. You can
change the amount of messages displayed when backup, restore, and media
management operations are performed by specifying the Report level option.
This also influences the amount of messages stored in the IDB. For more details,
see the online Help.
SMBF location
The SMBF is located in the following directory:
• On Windows Server 2008: Data_Protector_program_data\db40\msg
• On other Windows systems: Data_Protector_home\db40\msg
• On UNIX systems: /var/opt/omni/server/db40/msg
You can relocate the directory by editing the SessionMessageDir global option.
For more information on the Data Protector global options file, see the HP Data
Protector troubleshooting guide.
Concepts guide
193
Serverless Integrations Binary Files (SIBF)
SIBF records
The Serverless Integrations Binary Files stores raw NDMP restore data. This data is
necessary for restore NDMP objects.
SIBF size and growth
The SIBF does not grow very big in size. For more details, see “IDB size estimation
” on page 203. For NDMP backups, the SMBF grows proportionally to the number
of objects backed up. Approximately 3 KB are used for each backed up object.
SIBF location
The SIBF is located in the following directory:
• On Windows Server 2008: Data_Protector_program_data\db40\meta
• On other Windows systems: Data_Protector_home\db40\meta
• On UNIX systems: /var/opt/omni/server/db40/meta
IDB operation
During backup
When a backup session is started, a session record is created in the IDB. Also, for
each object and each object mirror in the session, an object version record is created.
All these records are stored in the CDB and have several attributes. The Backup
Session Manager updates media during a backup. All media records are stored in
the MMDB and are allocated for a backup depending on policies.
When a data segment is written to the tape and then to a catalog segment, then for
each object version that was part of this data segment, a media position record is
stored in the CDB. In addition, the catalog is stored in the DC (Detail Catalog) binary
file. One DC binary file is maintained per Data Protector medium. A DC binary file
is named MediumID_TimeStamp.dat. If a medium is overwritten during a backup,
its old DC binary file is removed and a new one is created.
All session messages generated during backups are stored in session messages binary
files (the SMBF part).
194
The Data Protector internal database
If transaction logging is enabled, an IDB backup removes old transaction logs and
starts creating new ones, which are necessary for an IDB recovery.
During restore
When configuring a restore, Data Protector performs a set of queries in the CDB and
DCBF parts to enable users to browse virtual filesystems of backed up data. These
browse queries are done in two steps. The first step is to select a specific object
(filesystem or logical drive). If this object has many backup versions and/or copies
stored, this can take some time because Data Protector scans the DCBF to build a
lookup cache for later browsing. The second step is browsing the directories.
After specific versions of files are selected, Data Protector determines the required
media and locates media position records that are used by the selected files. These
media are then read by Media Agents and data is sent to the Disk Agents that restore
the selected files.
During object copying or object consolidation
During an object copy or object consolidation session, the same processes run as
during a backup and a restore session. Basically, data is read from source media
as if it was restored and written to target media as if it was backed up. An object
copy or object consolidation session has the same effect on the IDB operation as
backup and restore. For details, see “During backup” on page 194 and “During
restore” on page 195.
Exporting media
When a medium is exported, the following is removed:
• All the media position records from that medium are removed from the CDB.
• All objects and object copies that now have no positions on any other media are
removed from the CDB part.
• Obsolete sessions (whose media have either been overwritten or exported) older
than 30 days are removed (this can be modified using the KeepSession variable
from the global option file). Session messages of such sessions are also removed.
• The medium record is removed from the MMDB part, and the DC binary file for
that medium is removed from the DCBF.
Concepts guide
195
Removing the detail catalog
When the detail catalog is removed for a specific medium, its DC binary file is
removed. The same result is achieved by removing the catalog protection for all
object versions and object copies on that medium (the next daily maintenance of DC
binary files removes the binary file). All other records stay in the CDB and MMDB
and it is possible to run a restore from such media (however, browsing is not possible).
Filenames purge
DC binary files show whether a given file is backed up on a related medium or not,
but the filenames are actually stored in the CDB. A filename is considered “used” if
it is marked as backed up in at least one DC binary file. Over time, it can happen
that a large number of filenames are not used. To remove such filenames, Data
Protector scans all DC binary files and then removes unused filenames.
File versions purge
When the catalog protection of all object versions stored on a specific medium
expires, automatic daily maintenance of DC binary files removes the respective binary
file.
Overview of IDB management
IDB configuration
One of the most important steps in setting up your Data Protector backup environment
is to configure the IDB. The initial configuration enables you to set your internal
policies regarding IDB size, the location of IDB directories, the IDB backup necessary
in case of IDB corruption or a disaster, and the configuration of IDB reports and
notifications.
IMPORTANT:
It is highly recommended to schedule an IDB backup to be performed on a daily basis.
Creating a backup specification for the IDB backup is part of the IDB configuration.
196
The Data Protector internal database
IDB maintenance
Once you configure the IDB, its maintenance is reduced to a minimum, mainly acting
on notifications and reports.
IDB recovery
An IDB recovery is needed if some of the IDB files are missing or corrupted. The
recovery procedure depends on the level of corruption.
For detailed information, see the online Help index: “IDB, recovery“.
IDB growth and performance
For proper IDB configuration and maintenance it is necessary to understand the key
factors that influence the IDB growth and performance, as well as the key tunable
parameters that you can adapt to your needs, and thus handle the growth and
performance of the IDB as efficiently as possible.
Key IDB growth and performance factors
The key factors for IDB growth and performance are the following:
• Logging level settings Logging level defines the amount of detail written to the
IDB during backup. The more detailed logging level you use, the greater influence
it has on the IDB. For details, see “IDB growth and performance: key tunable
parameters” on page 198.
• Catalog protection settings Catalog protection determines how long the information
about backed up data is available in the IDB. The longer period of catalog
protection you set, the greater influence it has on the IDB. For details, see “IDB
growth and performance: key tunable parameters” on page 198.
• Number of backed up files Data Protector keeps track of each file and each
version of that file. Different backup types impact the IDB differently. For
information on backup types, see “Full and incremental backups” on page 71.
• Number of backups
The more often you perform a backup, the more information is stored in the IDB.
• Filesystem dynamics The number of files created and removed between backups
can have a significant impact on the growth of the filenames part of the IDB. The
Report on System Dynamics gives you information about the system
dynamics. You can avoid the IDB growth due to filesystem dynamics by using the
Log Directories logging level.
Concepts guide
197
• Growth of your backup environment The number of systems being backed up in
the cell influences the IDB growth. Plan for the growth of your backup environment.
• Character encoding used for your filenames (applicable for UNIX only) Depending
on the filename encoding, a character in the filename can take up from one to
three bytes in the IDB. Shift-JIS encoded filenames, for example, take up to three
bytes in the IDB, while pure ASCII filenames take up only one byte. The character
encoding is relevant for growth of filename part of IDB on UNIX (on Windows,
all characters take up two bytes in the IDB).
• Number of object copies and object mirrors The more object copies and object
mirrors you create, the more information is stored in the IDB. For object copies
and object mirrors, the IDB stores the same information as for backed up objects,
except for filenames.
IDB growth and performance: key tunable parameters
The logging level and catalog protection are the main factors of the IDB growth and
performance. Their impact on the IDB depends on the settings you use. For a graphic
representation of the impact of different logging level and catalog protection settings,
see Figure 57 on page 199.
198
The Data Protector internal database
Figure 57 The influence of logging level and catalog protection on IDB
growth
Logging level as an IDB key tunable parameter
What is logging level?
Logging level determines the amount of details about backed up files and directories
written to the IDB. You can always restore your data, regardless of the logging level
used during backup.
Data Protector provides four logging levels that control the amount of details about
files and directories written to the IDB:
Log All
Logs all detailed information about backed up files and directories
(names, versions, and attributes).
Log Files
Logs all detailed information about backed up files and directories
(names and versions). This represents approximately 30% of all
detailed information about backed up files and directories.
Concepts guide
199
Log Directories
Logs all detailed information about backed up directories (names,
versions, and attributes). This represents approximately 10% of all
detailed information about backed up files and directories.
No Log
No information about backed up files and directories is logged to
the IDB.
The different settings influence the IDB growth, the backup speed, and the convenience
of browsing for data to be restored.
Impact on performance
The logging level defines the amount of data written to the IDB during a backup. This
also influences the IDB speed, and therefore the backup process.
Logging level and browsing for restore
Changing the level of stored information affects your ability to browse files using the
Data Protector GUI during a restore. If the No Log option is set, browsing is not
possible; if the Log Directories option is set, browsing of directories is possible;
if the Log Files option is set, full browsing is possible but file attributes (size,
creation, and modification dates and so on) are not displayed.
Regardless of the logging level set, it is always possible to restore your data:
• Instead of browsing for your data, you can always manually select a file to restore
(if you know the name of the file).
• You can retrieve information about backed up data from the media.
Logging level and restore speed
The restore speed is approximately the same when the Log All, Log
Directories, or Log Files options are set.
If the No Log option is set, the restore speed can be slower when restoring single
files. This is because Data Protector has to read all data from the beginning of an
object before finding a file to be restored.
In case of a full system restore, the whole object should be read anyway, so the
logging level settings do not play an important role.
200
The Data Protector internal database
Catalog protection as an IDB key tunable parameter
What is catalog protection?
Catalog protection determines how long the information about backed up data is
available in the IDB. This is different from data protection, which determines how
long the backed up data is available on the medium itself. If there is no catalog
protection, you can still restore your data, but you cannot browse for it in the Data
Protector GUI.
Catalog protection is based on the fact that the data stored last is most important
and accessed most frequently. Old files are seldom searched for, and therefore it is
allowable for their search to take more time.
Expired catalog protection
Once the catalog protection expires, the information is not immediately removed
from the IDB. Data Protector removes it automatically once per day. Since the
information in the IDB is organized on a per-medium basis, it is removed completely
when catalog protection expires for all objects on the medium.
Impact on performance
Catalog protection settings do not have any impact on the backup performance.
Catalog protection and restore
When catalog protection expires, data is restored as if it were backed up using the
No Log option. See “Logging level as an IDB key tunable parameter” on page 199.
Recommended usage of logging level and catalog protection
Always use catalog protection
Always set a reasonable level of catalog protection. The only exception is if the Log
None option is set (in this case catalog protection does not apply anyway).
If you set the catalog protection to Permanent, the information in the IDB is removed
only when media are exported or deleted. In this case, the size of the IDB grows
linearly until the data protection period is reached, even if the number of files in the
cell does not change. For example, if the data protection period is one year and
media are recycled, then significant growth of the IDB stops after one year. The
addition of new catalogs is approximately equal to the removal of old ones. If catalog
Concepts guide
201
protection is set for 4 weeks, then significant growth of the IDB stops after 4 weeks.
Therefore, in this case, the IDB is 13 times larger if the catalog protection is set to
Permanent.
It is recommended that catalog protection includes at least the last full backup. For
example, you can set a catalog protection of 8 weeks for full backups and one week
for incremental backups.
Use different logging levels in the same cell
A cell often consists of mail (or similar) servers that generate a large number of files
on a daily basis, database servers that store all information in a handful of files, and
some user workstations. Since the dynamics of these systems are rather different, it
is very difficult to prescribe one setting that suits them all. Therefore, it is recommended
to create several backup specifications with the following logging level settings:
• For mail servers, use the Log Directories option.
• For database servers, no logging is necessary as they have their own restore
policies. Therefore, use the No Log option.
• For workstations, the Log All or Log Files options allow for searching and
restoring different versions of files. For backups with the Log Directories or
No Log options set, you can import catalogs from the media, which, in a
reasonably short time, allows the possibility to browse for the selected object. For
information on importing catalogs from media, see the online Help index:
“importing, catalogs from media“.
Different logging levels for object copies
Backed up objects and object copies or mirrors of these objects can have the same
or different logging levels. Depending on your backup policy, the selected logging
level of object copies can be more or less detailed than that of the source objects.
For example, you can specify the No Log option for object mirrors if you create
these mirrors just to ensure a successful completion of a backup session. Or, you can
specify the No Log option for a backup object to increase the backup performance,
and then specify the Log All option for this object in a subsequent object copy
session.
Specifics for small cells
If the number of files in a cell is small and will remain small (a million files or less)
and the systems in the cell perform usual business activities, you can always use the
Log All option, which is the Data Protector default. However, you need to take
care of IDB growth and set a reasonable level of catalog protection.
202
The Data Protector internal database
Specifics for large cells
If the number of files grows into the tens of millions, or there are tens of thousands
of files generated on a daily basis, and you use the Log All option, then backup
speed and IDB growth will become a problem in a relatively short period of time. In
this situation, you have the following options:
• Reduce the logging level to the smallest acceptable level. Setting the Log Files
option can reduce the IDB size to a third, and setting the Log Directories
option to almost a tenth. This, of course, depends on the nature of the file systems
in the cell.
• Reduce the catalog protection to a minimum.
• Split the cell in two. As a final solution, you can always introduce another IDB
and redirect half of the systems into it.
You can configure Report on System Dynamics, which informs you about
dynamics of the growth of filenames on a particular client.
IDB size estimation
If you mainly perform filesystem backups, the IDB can grow to a significant size
(larger than 16 GB) under certain conditions. If you perform disk image or online
database, it is very likely that your IDB will not grow beyond 2 GB.
To estimate the size of the IDB use the Internal Database Capacity Planning Tool
located at:
• On the UNIX Cell Managers:
/opt/omni/doc/C/IDB_capacity_planning.xls
• On the Windows Cell Manager:
Data_Protector_home\docs\IDB_capacity_planning.xls
You can also use this tool to estimate the size of the IDB in environments with online
databases (Oracle, SAP R/3).
Concepts guide
203
204
The Data Protector internal database
6 Service management
In this chapter
Service Management, reporting, and monitoring help administrators manage their
backup environments more effectively. This chapter describes the concepts behind
the service management features and benefits available in both a standalone Data
Protector installation and through its integration with HP service management products.
It is organized as follows:
“Overview” on page 205
“Native Data Protector functionality” on page 207
“Service management integrations” on page 215
Overview
Enterprise information technology (IT) departments are increasingly using service
management tools, techniques, and methods to set service level expectations, measure
service delivery against those expectations, and to justify future service expansion.
Because IT groups must manage the risk of data loss, data backup and recovery are
critical elements in IT service delivery and management. Threats ranging from user
error to viruses or other unauthorized data access and modification, or the occasional
failure of the storage device itself put data at risk constantly. Business-critical data
loss can cost the enterprise thousands, even millions of dollars per hour of downtime.
Users, however, may perceive data backup as something that can slow down or
deny access to services while the backup is being conducted. But without this key
activity, the continued availability and timeliness of services can be compromised
and placed at significant risk.
While all data is at risk, not all data justifies equal recovery ability. IT departments
must protect the business-critical data to a higher level of protection than the less
valuable data - and do so cost effectively.
Concepts guide
205
Service management measures and reports are a key tool IT managers can use to
demonstrate value delivered to the organization and also to maintain competitive
cost structures. Service providers use Service Level Agreements (SLAs), that typically
establish availability and performance objectives, to document provider-customer
contractual expectations.
Demonstrating SLA compliance requires constant monitoring and periodic reporting
to show whether SLA expectations have been met. Data Protector, out of the box has
monitoring, notification, and reporting tools to document backup and recovery
operations. Integration with other service management products consolidates service
views, service performance data, and other capabilities into one console, giving you
better information and insight into overall IT service delivery.
Data Protector provides IT service managers with key data to enable operative
monitoring and planning of backup and data recovery operations. This data can be
used in service availability and recovery planning activities that are key if service
agreements are to be adhered to. In addition, Data Protector information can be
used to implement cost management and chargeback models for true IT financial
management.
Data Protector and service management
Data Protector provides service management supports and can be integrated with
service management applications, such as Operations Manager Windows,
Performance Agent (formerly MeasureWare Agent), Reporter, and Service Information
Portal.
Data Protector service management falls into two categories: native (or out-of-the-box)
and application integrations. The items in each category are described in more detail
later in this chapter.
206
Service management
Figure 58 Service management information flow
Native Data Protector functionality
The functionality described in the following sections comes with Data Protector “out
of the box.”
Key functions
• Data Protector has been equipped to track the elapsed times of key operations
and to register this data as well as volume data using the Application Response
Measurement Version 2.0 API (ARM 2.0 API). Registration of this data can be
performed with HP Performance Agent (PA).
• Built-in monitoring of running sessions allows you to instantly react to occurrences
in your backup environment.
• The Data Protector built-in notification and reporting engine allows you to receive
concise reports as well as immediate alerts in many different formats (such as
ASCII, HTML, and spreadsheet compatible format) and delivered in various ways
(such as e-mail, SNMP, broadcast (available on Windows only), write to file, and
send to external command). As the Data Protector built-in notification engine can
Concepts guide
207
•
•
•
•
•
send alerts via SNMP, it is possible to integrate virtually any application that can
receive SNMP traps.
Data Protector backup session auditing stores information about all backup tasks
that were performed over extended periods for the whole Data Protector cell, and
provides this information on demand in an integral and printable fashion for
auditing and administrative purposes.
The integration of Data Protector with HP Operations Manager software allows
you to receive alerts from Data Protector on the OM console and have automatic
actions performed.
The Data Protector capability to send major and critical events into the Windows
Event Log opens up a variety of interesting integration possibilities.
The integration with HP Operations Manager Windows (OMW)automatically
forwards Data Protector major and critical events to the OMW console. Automatic
actions can be set up to react upon failures in the backup environment.
The Data Protector built-in Java-based online reporting allows you to do online
reporting from wherever you are in your network (even from a remote location)
without the need to have the Data Protector user interface installed on your local
system. This functionality requires a Web browser.
Application Response Measurement version 2.0 (ARM 2.0 API)
What Is ARM?
The ARM API is an emerging standard for measuring end-to-end response times of
transactions in distributed environments. Application programs that use the ARM API
act as sources of response time information (and also user supplied information that
may be relevant to a particular transaction) for ARM compliant system management
and monitoring tools such as HP Performance Agent (PA). PA will log ARM transaction
information in its repository for subsequent analysis and reporting. It can also raise
real time alerts (or “alarms”) when the elapsed time of a specific transaction, such
as a backup operation, exceeds a predefined threshold. When a real time alert is
raised, a number of actions are possible, including, but not limited to, informing a
208
Service management
central operations console, such as HP Operations Manager software, paging a
system operator, or taking automated remedial action to resolve the problem.
Table 13 ARM functionality
Transaction description
(ARM 1.0)
Additional data logged to
ARM (ARM 2.0)
Usage
Backup specification session
duration
Processed data [MB]
Availability and recovery
planning. Chargeback.
Object backup session
duration
Processed data [MB]
Availability and recovery
planning. Chargeback.
Restore session duration
Recovered data [MB]
Availability and recovery
planning
IDB check duration
IDB size [MB]
Data Protector architecture
management
IDB purge duration
IDB size after purge and
number of purged records
Data Protector architecture
management
As Data Protector is already ARM equipped, it is a fairly simple task to integrate
Data Protector with an application like PA that supports the ARM API. On Windows
platforms, this is completely automatic. If Data Protector is installed on a system where
PA is already present or vice versa, the transaction data will immediately show up
in PA and HP Performance Manager (PM). On HP-UX, the only required task is to
create a link from a PA library to a Data Protector directory. For more information,
see the online Help index: “ARM integration, installing“.
Another way to interface PA with Data Protector is the Data Source Integration (DSI).
This is important if the application that you use for transaction tracking is not ARM
2.0 compliant. ARM 1.0 permits you to log time-specific data only, such as the
duration of a backup session. With DSI you can add the capability to report any
data which can be retrieved from the command line to tools like PA. This enables
highly customized reporting.
Concepts guide
209
Integration with HP Operations Manager software
Functionality of the Data Protector OM integration
Data Protector integrates with HP Operations Manager software (OM). OM simplifies
management of large networks by allowing the operator to monitor and administer
the network and the applications from a single point. Once Data Protector is integrated
in the OM environment, the network administrator can immediately see if anything
is wrong during backup and react upon the information given. Data Protector
messages can be displayed in the OM message window.
Functionality of the Data Protector Operations Manager Windows
The Data Protector Operations Manager on Windows (OMW) provides the following
functionality:
• Data Protector writes all major and critical messages that occur during backup,
restore or any other operation to the Windows Event Log. Operations Manager
Windows (OMW) then uses these events and forwards them to the OMW console,
so that an operator can react to them.
• Service monitoring
OMW monitors all Data Protector services running on the Cell Manager as well
as any Data Protector client system. In case of failure of any of these services,
OMW immediately alerts the operator. OMW can also be configured in such a
way that it automatically attempts to restart the failed service.
SNMP traps
SNMP traps allow a Service Management application to receive and process an
SNMP trap message when a Data Protector event occurs or when an SNMP trap is
sent as a result of Data Protector’s checking and maintenance mechanism. For more
information on Data Protector on configuring SNMP traps, see the online Help index:
“SNMP, reports send methods“.
The monitor
The Data Protector monitor is a part of the Data Protector user interface and allows
you to supervise and to take corrective action on currently running backup, restore,
and media management sessions. Monitoring lets you view all sessions in a cell and
shows you detailed messages and the current status of these sessions. In a multi-cell
environment, you can view the sessions that run on computer systems in other cells.
210
Service management
From the monitor’s user interface, you can abort a backup, restore, or media
management session or respond to “mount” requests.
If you make use of the Manager-of-Managers, you can monitor sessions of multiple
cells simultaneously from one user interface.
Reporting and notification
Data Protector reporting represents a powerful, customizable, and flexible tool for
managing and planning your backup environment. Data Protector has always had
a rich set of built-in reports that system administrators have relied upon to manage
Cell Managers. IT Service Providers now can use these same reports to demonstrate
data protection SLA compliance. Built-in reports that are especially relevant to service
level management include:
• Inventory/Status Reports such as the host_not_conf report, which contains
information about unprotected systems, the dl_sched report, which lists all
scheduled backups, object copy, and object consolidation as well as the
media_list report, which is a media inventory report.
• Capacity Utilization Reports such as the licensing report, which is a Data
Protector license utilization report, and the dev_unused report, which lists devices
that are currently not used for backup, object copy, or object consolidation and
are consequently available.
• Problem Reports such as the session_statistics report, which consists of
information about failed backup, copy, and consolidation sessions. An
administrator can receive an hourly, daily, or weekly E-mail report on failed jobs
and the reasons for failure.
The notification and reporting capabilities that have always been part of the Cell
Manager (and that have been extended significantly from earlier versions) also allow
you to:
• Choose from numerous pre-configured reports (including, but not limited to, reports
such as sessions in a specific time frame, IDB reports, and device usage report)
• Specify your own parameters for those reports (such as time frames, backup,
copy, and consolidation specifications, and groups of backups)
• Select from various different output formats (such as ASCII, HTML, and spreadsheet
compatible formats)
• Schedule those reports with the Data Protector built-in scheduler
• Trigger report sending based on events (such as device failure, mount requests,
and end of sessions)
Concepts guide
211
• Select from many delivery methods used to deliver reports (such as e-mail, SNMP,
broadcast (available on Windows only), write to file, and send to external
command)
You can combine most of these different formats, delivery methods, schedules, and
triggers.
Some examples are shown below:
Reporting and notification examples
• Every morning at 7:00, a report about all backup, copy, and consolidation
sessions in the last 24 hours is created and sent by e-mail in the ASCII format to
the backup administrator's mailbox. Additionally, the same report is written to a
file on your Web server in the HTML format so that others can also access this
information.
• In event of a device failure or a mount request, a broadcast message is
immediately sent to the backup administrator's Windows workstation, and an
external command is triggered, which activates the backup administrator's pager.
• At the end of a backup session, every end user whose system has been backed
up receives an e-mail in ASCII format that contains a backup status report.
Event logging and notification
The Data Protector Event Log is a central repository of all Data Protector-related
notifications. The Data Protector built-in notification engine sends alerts or activates
the Data Protector reporting mechanism based on the log entries. The event log is
the information source for SLA-compliance reports in Data Protector or in HP software
management applications. In addition to reports, log entries feed HP software
management applications via the Data Protector SPI (SMART Plug-In) so that they can
trigger preventive or corrective actions (for details, see the example under 3.1).
Since the Data Protector built-in notification engine can send alerts via SNMP, virtually
any application that can receive SNMP traps can integrate with Data Protector.
Integration with Operations Manager and Reporter is an example of SNMP trap-based
implementation.
The Event Log is accessible only for Data Protector users in the Admin group and for
Data Protector users that are granted the Reporting and notifications user
rights. You can view or delete all events in the Event Log.
212
Service management
Data Protector log files
Some Service Management applications, such as HP Operations Manager software,
allow you to specify when and which log files should be monitored for a specific log
entry. If the specified entry is detected in the file, an action can be specified. In OM
this is called Log file encapsulation.
You can configure such a Service Management application to monitor Data Protector
log files for specific log entries (Data Protector events) and define an action that is
to be executed in case a particular Data Protector event is detected.
For more information on Data Protector log files, see the HP Data Protector
troubleshooting guide. Note that there is no log files formatting specification provided.
Windows application log
Some Service Management applications, such as Operations Manager Windows
(OMW), monitor the Windows Application Log.
To enable automatic forwarding of all Data Protector messages and messages about
the Data Protector services (if they are stopped) to Windows Application Log, set the
EventLogMessages variable in the Data Protector global options file to 1. For
more information on the Data Protector global options file, see the HP Data Protector
troubleshooting guide.
Java-based online reporting
Data Protector comes with a Java-based online reporting capability that lets you
configure, run, and print all Data Protector built-in reports, live and interactive. During
reporting operations, Data Protector Java reporting directly accesses the Cell Manager
to retrieve current data. You can make this Java applet available through a Web
server, copy it to the client machine for direct access, or use it locally. Using this
facility only requires a supported Web browser; there is no need to have the Data
Protector GUI installed on the system. Not only can you use the Java reporting facility
to get online access to your reports, but you can also configure your reporting structure
through it, such as adding new reports to a schedule or changing a report's
parameters.
Concepts guide
213
Data Protector checking and maintenance mechanism
Data Protector has a rich automated daily self-check and maintenance mechanism,
which improves its operational reliability and predictability. Data Protector’s self-check
and maintenance tasks include:
• “Not Enough Free Media” check
• “Data Protector License Expiration” check
For more information, see the online Help index: “checks performed by Data
Protector”.
Central management, distributed environment
The Data Protector MoM enables administrators to centrally manage an enterprise
environment consisting of several Data Protector Cell Managers. The MoM system
administrator performs configuration, media management, monitoring, and status
reporting tasks for the whole enterprise from a single console. With MoM, managing
many Data Protector Cell Managers is as convenient as managing just one. IT service
providers can administer larger clients’ environments without adding employees. For
more information on MoM, see the online Help index: “MoM environment“.
Using the data provided by Data Protector
What can I do with the data?
Here are some examples of what you can do with the data that Data Protector
provides:
• Real-time alerting of backup or restore sessions that exceed the specified time
window (PA).
• Creating graphs of backup duration of key systems in your environment to detect
trends in operation time (PM).
• Forecasting of the IDB growth to be able to spot points in time where certain limits
will be reached (PM Planner).
• Regular e-mail reports to back up operators, end users, and management (Data
Protector built-in reporting with the capability to send e-mails).
• Backup reports written to a Web server to make them available on an on-demand
basis (built-in Data Protector reporting with the capability to write HTML).
214
Service management
• Sending major and critical Data Protector events to your network management
solution, such as HP Network Node Manager (Data Protector built-in notification
engine sending SNMP traps).
Service management integrations
The following Data Protector integrations can be installed to help facilitate service
management and to give you single-point access to rich service management
functionality.
Key functions
•
•
•
•
•
Standard and custom report formats
Availability of a “trouble ticket” interface for Data Protector
Facilitates a specific, consistent, measurable level of service
Availability of Data Protector information through a web interface
Graphical representation of data
Figure 59 Example of an IT service provider environment with service
management access through the client portal
Data Protector OM-R integration
The integration of Data Protector with HP Operations Manager software (OM) is
extended by adding HP Reporter 3.7 or 3.8 (English version). With Reporter, service
providers can generate reports from the OM console as a central management point.
Integration with Reporter adds a variety of new reports in the following categories:
• Backup Session Reports
Concepts guide
215
• Administration Reports
• Media Pool Reports
• Performance
An IT Service Provider can use these reports to demonstrate its SLA compliance to a
customer. For example, the “Data Protector Transaction Performance” Report consists
of service performance metrics (one of the IT SLA parameters):
Figure 60 Data Protector Reporter example
In addition to SLA compliance reports, An IT Service Provider can generate monthly
operational reports for the Data Protector environment. For example, the “Data
Protector Operational Error Status” report aggregates the “problem” data and can
be used by an IT service provider for operational planning.
216
Service management
Figure 61 Operational error status report
Data Protector OM SIP
This integration also uses SIP to provide Data Protector information through a
web-based interface. It does not require OVO to be installed. The integration provides
information through tables and gauges.
Concepts guide
217
Figure 62 Direct SIP integration example
218
Service management
7 How Data Protector operates
In this chapter
This chapter describes the operation of Data Protector. It explains Data Protector
processes (on UNIX) and services (on Windows), backup and restore sessions, and
media management sessions.
It is organized as follows:
“Data Protector processes or services” on page 219
“Backup sessions” on page 220
“Restore sessions” on page 225
“Object copy sessions” on page 229
“Object consolidation sessions” on page 233
“Media management sessions” on page 235
Data Protector processes or services
Data Protector runs several background processes (on UNIX) and services (on
Windows) that enables it to run backup and restore sessions. It provides the necessary
communication paths, activates backup and restore sessions, starts Disk Agents and
Media Agents, stores information about what was backed up, manages media, and
performs similar functions.
Inet
The Data Protector Inet service runs on each Windows system
in the Data Protector cell. Inet is responsible for communication
between systems in the cell and starts other processes needed
for backups and restores. The Data Protector Inet service is
started when Data Protector is installed on a system. On UNIX
systems, the system inet daemon (INETD) starts the Data
Protector Inet process.
Concepts guide
219
CRS
The CRS (Cell Request Server) process (service) runs on the
Data Protector Cell Manager. It starts and controls backup
and restore sessions. The service is started when Data
Protector is installed on the Cell Manager system and is
restarted each time the system is restarted.
MMD
The MMD (Media Management Daemon) process (service)
runs on the Data Protector Cell Manager and controls media
management and device operations. The process is started
by the Cell Request Server process (service).
RDS
The RDS (Raima Database Server) process (service) runs on
the Data Protector Cell Manager and manages the IDB. The
process is started when Data Protector is installed on the Cell
Manager.
UIProxy
The Java GUI Server (UIProxy service) runs on the Data
Protector Cell Manager. It is responsible for communication
between the Java GUI Client and the Cell Manager, moreover,
it performs business logic operations and sends only important
information to the client. The service is started as soon as Data
Protector is installed on the Cell Manager.
For instructions on how to manually start or stop the Data Protector processes and
services, see online Help.
Backup sessions
This section describes how a backup session is started, what happens during a
backup session, and the processes and services involved.
What is a backup session?
When a backup specification is started it is called a backup session. The backup
session copies data from a source, typically a hard disk, to a destination, typically
tape media. The result of a backup session is a copy of data on the backup media,
the media set.
220
How Data Protector operates
Scheduled and interactive backup sessions
Scheduled backup session
A scheduled backup session is started by the Data Protector Scheduler at the time
you have specified. You can view the progress of the scheduled backup session in
the Data Protector monitor.
Interactive backup session
An interactive backup session is started from the Data Protector user interface directly.
The Data Protector monitor starts immediately and you can view the progress of the
backup session. Note that multiple users can monitor the same backup session. You
may want to stop monitoring by disconnecting the user interface from the session.
The session will then continue in the background.
Backup session data flow and processes
What happens in a backup session?
The information flow of a backup session is shown in Figure 63 on page 222. Note
that the data flow and processes described here are for a standard network backup.
For data flow and processes specific to other types of backup, such as direct backup,
see the related chapter.
When a backup session is started, the following happens:
1.
The Backup Session Manager (BSM) process is started on the Cell Manager
system and controls the backup session. This process reads the backup
specification for information on what to back up, and which options, media,
and devices to use for the backup.
2. The BSM opens the IDB and writes to the IDB information about the backup
session, such as generated messages, details about the backed up data, and
the devices and media that were used for the session.
3. The BSM starts Media Agents (MAs) on the systems with devices configured for
backup. A new Media Agent is started for each drive used in parallel. The
number of Media Agents that can be started in the cell is limited by the cell
configuration and the number of licenses you have purchased.
In a backup session with object mirroring, the BSM also starts Media Agents
that will be used for mirroring.
Concepts guide
221
4. The BSM starts Disk Agents (DAs) for each disk to be backed up in parallel. The
actual number of Disk Agents started depends on the concurrency of Disk Agents
configured in the backup specification. This is the number of Disk Agents that
can be started to send data in parallel to a Media Agent, thus allowing a device
to stream.
5. Disk Agents read data from disks and send it to the Media Agents that write
data to media.
In a backup session with object mirroring, Media Agents used for writing mirrored
objects are daisy-chained. Each Media Agent writes the received data to media
and forwards it to the next Media Agent in the chain.
6. The BSM monitors the progress of the session and starts new Disk Agents and
new Media Agents as necessary.
7.
When the backup session is completed, the BSM closes the session.
Figure 63 Backup session information flow (1)
How many sessions can run concurrently?
A number of backup sessions can run in the cell at the same time. This number is
limited by resources in the cell, such as the availability of devices and the configuration
of the Cell Manager, for instance, processor speed, main memory size, and similar.
222
How Data Protector operates
To prevent Data Protector processes from exceeding system capabilities, the maximum
number of concurrent backup sessions is limited. The limit is configurable.
Figure 64 on page 223 shows multiple sessions running concurrently.
Figure 64 Backup session information flow - multiple sessions
Pre-exec and post-exec commands
Data Protector pre-exec commands enable you to execute some actions before a
backup or a restore session. Data Protector post-exec commands enable you to
execute some actions after a backup or a restore session. A typical pre-exec action
would be to shut down a database to put data in a consistent state.
The pre-exec and post-exec commands can be set for a backup specification and,
as such, executed on the Cell Manager system, or they can be specified as a backup
object option and thus executed on the client system where the respective Disk Agent
is running.
Pre-exec and post-exec script commands can be written as executables or shell scripts.
These are not supplied by Data Protector and must be written separately by, for
example, the backup operator.
Concepts guide
223
Queuing of backup sessions
Timeout
When a backup session is started, Data Protector tries to allocate all needed resources,
such as devices. The session is queued until the required minimum resources are
available. If the resources are still unavailable after the timeout, the session is aborted.
The timeout period can be set using the SmWaitForDevice global option.
Optimizing the load
To optimize the load on the Cell Manager, Data Protector can, by default, start up
to five backup sessions at the same time. The default value can be modified in the
global options file. If more are scheduled at the same time, the extra sessions are
queued and started subsequently as others are finished.
Mount requests in backup sessions
What is a mount request?
A mount request in a backup session appears when Data Protector needs a new
medium for backup and the medium is not available.
Data Protector issues a mount request for one of the following reasons:
Issuing a mount request
• There is not enough space on the backup media and there are no new media
available.
• Data Protector media allocation policy for backup requires a medium that is not
available in the device.
• The order of media used for backup is defined in the pre allocation list and media
are not available in this order.
For more information, see “Adding data to media during backup sessions” on page 148
and “Selecting media for backups ” on page 147.
Responding to a mount request
Responding to a mount request includes providing the required media and telling
Data Protector to proceed with the backup.
Data Protector allows you to configure what happens when a mount request is issued:
224
How Data Protector operates
Sending notification to an operator
You can configure a Data Protector notification to send an e-mail to the operator with
information about the mount request. The operator can take the appropriate actions,
such as manually loading the needed media or aborting the session. For more
information, see “Reporting and notification” on page 211.
Automating a mount request
You can configure automated actions for the handling of mount requests. To do this,
write a script or a batch program that performs the desired action.
Backing up with disk discovery
What is disk discovery?
In backing up with disk discovery, Data Protector creates a detailed list of disks on
the target system when the backup session is started, and backs up all disks. Therefore,
all local disks on the system are backed up even though they were not present on
the system when the backup was configured. Backup with disk discovery is particularly
useful in dynamic environments, where configurations change rapidly. It enables you
to select or exclude specific directories in the backup.
How does it compare to a standard backup?
In a standard backup, you explicitly configure specific disks, directories or other
objects for backup by configuring them in the backup specification. Therefore, only
these objects are backed up. If you add new disks to the system or want to back up
some other objects, you must manually edit the backup specification and these new
objects. You can select, as you configure the backup, the method you want to use disk discovery or standard backup.
Restore sessions
This section describes how a restore session is started, what happens during a restore
session, and the processes and services involved.
What is a restore session?
In a restore session, data is copied from a backup copy, typically on a tape medium,
back to a disk.
Concepts guide
225
A restore session is started interactively. You tell Data Protector what to restore, let
Data Protector determine the needed media, select some options and start the restore.
You and other users can monitor the progress of the session.
Restore session data flow and processes
What happens in a restore session?
When a restore session, as shown in Figure 65 on page 227, is started, the following
happens:
1.
The Restore Session Manager (RSM) process is started on the Cell Manager
system. This process controls the restore session.
2. The RSM opens the IDB, reads the information about media needed for restore,
and writes the information about the restore session to the IDB, such as generated
messages.
3. The RSM starts Media Agents (MAs) on the systems with devices used for restore.
For each drive used in parallel, a new Media Agent is started.
4. The RSM starts Disk Agents (DAs) for each disk restored in parallel. The actual
number of Disk Agents started depends on the objects you selected for restore.
For more information, see “Parallel restores” on page 228.
5. Media Agents read data from media and send it to the Disk Agents that write
the data to disks. The RSM monitors the progress of the session and starts new
Disk Agents and new Media Agents as necessary.
6. When the restore session is completed, the RSM closes the session.
226
How Data Protector operates
Figure 65 Restore session information flow
How many restore sessions can run concurrently?
A number of restore sessions can run in the cell at the same time. This number is
limited by resources in the cell, such as the Cell Manager and systems with connected
devices.
Queuing of restore sessions
Timeout
When a restore session is started, Data Protector tries to allocate all needed resources,
such as backup devices. The session is queued for as long as the required minimum
resources are not yet available. Data Protector tries to allocate the resources for a
specific period of time, the timeout. Timeout is user configurable. If the resources are
still unavailable after the timeout, the session is aborted.
Concepts guide
227
Mount requests in a restore session
What is a mount request?
A mount request appears in a restore session when the media needed for restore are
not available in the device. Data Protector allows you to configure a desired action
that should happen when a mount request appears.
Responding to a mount request
Responding to a mount request includes providing the required media or any copy
of media and telling Data Protector to proceed with the restore.
Parallel restores
What is a parallel restore?
In a parallel restore, interleaved data from multiple objects is read concurrently from
media in a single path and restored. A parallel restore significantly improves restore
performance when restoring multiple objects from the same media. For more
information, see Figure 66 on page 229.
How does it compare to a standard restore?
Data from multiple Disk Agents is (most of the time) multiplexed and stored on the
media. See Figure 41 on page 149. In a standard restore, Data Protector reads
multiplexed data from the media and assembles only the parts needed for the selected
object. When the next object is restored, Data Protector must rewind the media and
read the parts for the other object, assuming both objects are on the same medium
and written using multiplexing.
228
How Data Protector operates
Figure 66 Parallel restore session flow
In a parallel restore, Data Protector reads multiplexed data for all selected objects
and assembles the parts needed for all the objects on the fly, sending the right data
to the right Disk Agents. This improves performance when reading from the media.
The performance is additionally improved if the selected objects are written to different
physical disks. In this case, data is copied to multiple disks at the same time.
Fast multiple single file restore
Data Protector uses discontinuous object restore to improve restore performance.
After restoring a specific file or tree, Data Protector repositions itself directly on the
next file or tree on the medium, if there's at least a single segment between the files
or trees, and continues the restore.
Within an individual restore object you can start multiple Disk Agents. This way the
restoring of multiple single files that are located all over the medium is much faster
than if Data Protector were to traverse the medium.
Object copy sessions
This section describes how an object copy session is started, what happens during
the session, and the processes and services involved.
Concepts guide
229
What is an object copy session?
An object copy session is a process that creates an additional copy of the backed
up data on a different media set. During an object copy session, the selected backed
up objects are copied from the source to the target media.
Automated and interactive object copy sessions
Automated object copy session
An automated object copy session can either be scheduled or started immediately
after a backup. A scheduled object copy session is started at the time you have
specified using the Data Protector Scheduler. A post-backup object copy session is
started after the specified session finishes. You can view the progress of the automated
object copy session in the Data Protector monitor.
Interactive object copy session
An interactive object copy session is started from the Data Protector user interface
directly. The Data Protector monitor starts immediately and you can view the progress
of the session. Multiple users can monitor the same object copy session. You may
want to stop monitoring by disconnecting the user interface from the session. The
session will then continue in the background.
Object copy session data flow and processes
What happens in an object copy session?
The information flow of an object copy session is shown in Figure 67 on page 232.
When an object copy session is started, the following happens:
1.
The Copy and Consolidation Session Manager (CSM) process is started on the
Cell Manager system. This process reads the object copy specification for
information on what to copy and which options, media, and devices to use. It
also controls the object copy session.
2. The CSM opens the IDB, reads the information about the media needed for
copying, and writes the information about the object copy session, such as
generated messages, to the IDB.
3. The CSM locks the devices. The session is queued until all read Media Agents
and the minimum required write Media Agents are locked, with the same timeout
as for backup. If the resources are still unavailable after the timeout, the session
is aborted.
230
How Data Protector operates
4. The CSM starts the Media Agents on the systems with devices configured for
copying. The Media Agents load the source and target media allocated
according to the backup policies.
5. Media Agents read the data from the source media and connect to the Media
Agents loaded with the target media.
If destination devices are not specified per object, Data Protector selects them
automatically from those you selected in the object copy specification according
to the following criteria in the order of priority:
• destination devices with the same block size as source devices are selected
before those with a different block size
• locally attached devices are selected before network attached devices
6. Media Agents loaded with the target media accept connections from the Media
Agents loaded with the source media and start writing object copies to the target
media.
If the block size of the source device is smaller than the block size of the
destination device, blocks are repackaged at this stage of the object copy session.
7.
When the object copy session is completed, the CSM closes the session.
How many sessions can run concurrently?
A number of object copy sessions can run in the cell at the same time. This number
is limited by the resources in the cell, such as the Cell Manager and the systems with
connected devices.
Concepts guide
231
Figure 67 Object copy session information flow
Queuing of object copy sessions
Timeout
When an object copy session is started, Data Protector tries to allocate all needed
resources. The session is queued until the required minimum resources are available.
If the resources are still unavailable after the timeout, the session is aborted. The
timeout period can be set using the SmWaitForDevice global option.
Mount requests in an object copy session
What is a mount request?
A mount request in an object copy session is issued when a source or a target medium
needed for the object copy operation is not available.
Responding to a mount request
Responding to a mount request includes providing the required medium and confirming
the mount request. If the required source medium has media copies, you can provide
a copy instead of the original medium.
232
How Data Protector operates
Object consolidation sessions
This section describes how an object consolidation session is started, what happens
during the session, and the processes and services involved.
What is an object consolidation session?
An object consolidation session is a process that merges a restore chain of a backup
object, consisting of a full backup and at least one incremental backup, into a new,
consolidated version of this object. During an object consolidation session, Data
Protector reads the backed up data from the source media, merges the data, and
writes the consolidated version to the target media.
For more information, see Chapter 11 on page 257.
Automated and interactive object consolidation sessions
Automated object consolidation session
An automated object consolidation session can either be scheduled or started
immediately after a backup. A scheduled object consolidation session is started at
the time you have specified using the Data Protector Scheduler. A post-backup object
consolidation session is started after the specified backup session finishes. You can
view the progress of an automated object consolidation session in the Data Protector
monitor.
Interactive object consolidation session
An interactive object consolidation session is started from the Data Protector user
interface directly. The Data Protector monitor starts immediately and you can view
the progress of the session. Multiple users can monitor the same object consolidation
session. You may want to stop monitoring by disconnecting the user interface from
the session. The session will then continue in the background.
Object consolidation session data flow and processes
When an object consolidation session is started, the following happens:
1.
The Copy and Consolidation Session Manager (CSM) process is started on the
Cell Manager system. This process reads the object consolidation specification
for information on what to consolidate and which options, media, and devices
to use. It controls the object consolidation session.
Concepts guide
233
2. The CSM opens the IDB, reads the information about the needed media, and
writes the information about the object consolidation session, such as generated
messages, to the IDB.
3. The CSM locks the devices. The session is queued until all read Media Agents
and the minimum required write Media Agents are locked, with the same timeout
as for backup. If the resources are still unavailable after the timeout, the session
is aborted.
4. The CSM starts the Media Agents on the systems with devices that will be used
in the session. The Media Agents load the source and target media allocated
according to the backup policies.
If destination devices are not specified per object, Data Protector selects them
automatically from those you selected in the object consolidation specification
according to the following criteria in the order of priority:
• destination devices with the same block size as source devices are selected
before those with a different one
• locally attached devices are selected before network attached devices
5. One Media Agent reads the full object version. It sends the data to another
Media Agent that reads incremental object versions. The latter Media Agent
does the actual consolidation and sends the data to the Media Agent that writes
the data to the target media.
If the full backup and the incremental backups reside in the same file library,
the same Media Agent reads all the backups and consolidates them.
If the block size of the source device is smaller than that of the destination device,
blocks are repackaged.
6. When the object consolidation session is completed, the CSM closes the session.
How many sessions can run concurrently?
A number of object consolidation sessions can run in the cell at the same time. Object
consolidations sessions are treated like backup sessions and their number is limited
by the same factors.
Queuing of object consolidation sessions
Timeout
When an object consolidation session is started, Data Protector tries to allocate all
needed resources. The session is queued until the required minimum resources are
234
How Data Protector operates
available. If the resources are still unavailable after the timeout, the session is aborted.
The timeout period can be set using the SmWaitForDevice global option.
Mount requests in an object consolidation session
What is a mount request?
A mount request in an object consolidation session is issued when a source or a
target medium needed for the object consolidation operation is not available.
Responding to a mount request
Responding to a mount request includes providing the required medium and confirming
the mount request. If the required source medium has media copies, you can provide
a copy instead of the original medium.
Media management sessions
What is a media management session?
A media management session is used to perform a certain action on the media, such
as initializing media, scanning the content, verifying data on the media, and copying
media.
Logging to the IDB
Information about a media management session, such as generated messages, is
stored in the IDB.
Data Protector monitor and media management session
A media management session can be viewed in the monitor window. If you close
the Data Protector GUI, the session will continue in the background.
Media management session data flow
What happens in a media management session?
When a media management session is started, the following happens:
1.
The Media Session Manager (MSM) process is started on the Cell Manager
system. This process controls the media session.
Concepts guide
235
2. The MSM starts the Media Agents (MAs) on the system that has devices used
for the media management session.
3. Media Agents perform the requested operation and send generated messages
to the Data Protector user interface, where you can track the progress. The session
is also stored in the IDB.
4. When the session is complete, the MSM closes the session.
How many sessions can run?
A number of media management sessions can run in the cell at the same time if they
do not use the same resources, such as devices or media.
236
How Data Protector operates
8 Integration with database
applications
In this chapter
This chapter gives a brief description of the integration of Data Protector with database
applications, such as Microsoft Exchange Server, Oracle Server, IBM DB2 UDB, and
Informix Server.
It is organized as follows:
“Overview of database operation” on page 237
“Filesystem backup of databases and applications” on page 239
“Online backup of databases and applications” on page 239
For a detailed list of supported integrations, see the HP Data Protector product
announcements, software notes, and references.
Overview of database operation
From the user’s perspective, a database is a set of data. Data in a database is stored
in tables. Relational tables are defined by their columns and are given a name. Data
is stored in rows in the table. Tables can be related to each other, and the database
can be used to enforce these relationships. Data can thus be stored in relational
format or as object-oriented structures such as abstract data types and methods.
Objects can be related to other objects, and objects can contain other objects. A
database is usually managed by the server (manager) process that maintains data
integrity and consistency.
Whether you use relational structures or object-oriented structures, databases store
data in files. Internally, these are database structures that provide a logical mapping
of data to files, allowing different types of data to be stored separately. These logical
divisions are called tablespaces in Oracle, dbspaces in Informix Server, and segments
in Sybase.
Concepts guide
237
Figure 68 Relational database
Figure 68 on page 238 shows a typical relational database with the structures
described below.
Data files are physical files that contain all of a database’s data. They change
randomly and can be very large. They are internally divided into pages.
Transaction logs record all database transactions before they are further processed.
Should a failure prevent modified data from being permanently written to data files,
the changes can be obtained from log files. Any kind of recovery is done in two
parts: roll forward, which applies transaction changes into the main database and
roll back, which removes uncommitted transactions.
Control files hold information about the physical structure of the database, such as,
database names, names and locations of a database’s data files and log files, and
the time stamp of the database’s creation. This control data is kept in control files.
These files are critical for the operation of the databaseThe cache of the database server process contains the most-often used pages of the
data files.
The following is the standard flow of transaction processing:
238
Integration with database applications
1.
A transaction is first recorded into the transaction log.
2. Changes required in the transaction are then applied to cached pages.
3. From time to time sets of modified pages are flushed to data files on disk.
Filesystem backup of databases and applications
Databases are constantly changing while they are online. Database servers consist
of multiple components that minimize response time for connected users and increase
performance. Some data is kept in the internal cache memory and some in temporary
log files, which are flushed at checkpoints.
Because data in a database can change during a backup, a filesystem backup of
database files makes no sense without putting the database server into a special
mode or even offline. Saved database files have to be in a consistent state, otherwise
the data is of no use.
The following steps are required to configure a filesystem backup of the database
or application:
• identify all data files
• prepare two programs that are able to shut down and start up the database,
respectively
• configure the filesystem backup specification with all the data files included and
specify the shut-down program as a pre-exec command and the start-up program
as a post-exec command
This method is relatively simple to understand and configure but has one key
disadvantage: the database is not accessible during the backup, which is
unacceptable for most business environments.
Online backup of databases and applications
To overcome the necessity to shut down the database during a backup, database
vendors have prepared interfaces that can be used to put databases temporarily into
special modes to save the data to tapes. Server applications are thus online and
available to users during the backup or restore process. These application-specific
interfaces allow backup products, like Data Protector, to back up or restore logical
units of the database application. The functionality of the backup APIs varies
depending on the database vendor. Data Protector integrations are available for
major databases and applications. For a detailed list of supported integrations, see
the HP Data Protector product announcements, software notes, and references.
Concepts guide
239
The essence of the backup interface is that it provides the backup application with
consistent data (even if it may not be consistent on the disk) while at the same time
keeping the database operational.
Figure 69 Data Protector integration with databases
Figure 69 on page 240 shows how a relational database is integrated with Data
Protector. Data Protector provides a Database Library that is linked in to the database
server. The database server sends data to Data Protector and requests data from it.
Database utilities are used to trigger backup and restore operations.
A typical procedure to configure the backup of a database through the Data Protector
integration is as follows:
1.
A database/application-specific agent is installed on the database system
2. The Data Protector integration is configured for each database. Data needed
for Data Protector to work with this database are stored on the database system
(into configuration files or registry entries). Typically, this includes pathnames
and user names/passwords.
3. The backup specification is prepared using the Data Protector user interface.
Besides the key advantage of the database being online all the time there are also
other benefits of using the Data Protector integrations with the databases:
240
Integration with database applications
• There is no need to specify the location of data files. These can be located on
different disks.
• The logical structure of the database can be browsed. It is possible to select only
a subset of the database.
• Applications are aware of backup operation and keep track of which parts are
backed up.
• Several modes of backup are possible. Besides full backups, users can select
(block level) incremental backups or only the backup of transaction logs.
• Several modes of restore are possible and after the restore of data files, the
database can automatically restore transaction logs and apply them as configured.
Concepts guide
241
242
Integration with database applications
9 Direct backup
In this chapter
This chapter introduces the direct backup concept and the technologies that enable
it. The chapter also discusses the direct backup configurations that are supported by
Data Protector.
It is organized as follows:
“Overview” on page 243
“Requirements and support” on page 249
“Supported configurations” on page 250
Overview
The storage industry’s demand for backup solutions that minimize application
downtime and system loads while maximizing backup speeds is growing. Data
volume is also growing; it has doubled every 1.5 years over the last 20 years and
continues to grow even faster.
Applications and services need to be online nearly all the time with maximum
performance. Backup windows are narrow and performance degradation due to
backup (or anything else) is no longer acceptable.
In addition, the requirement for solutions that do not demand substantial investment
in solution-specific equipment is also growing.
This multi-pronged requirement has led to the development and introduction of new
direct or “serverless” backup technologies.
For enterprises and service providers that manage mission critical Oracle
environments, Data Protector’s direct backup feature is a non-intrusive, serverless
backup extension to HP’s family of network backup solutions.
Concepts guide
243
Direct backup extends the benefits of HP’s ZDB solution by moving data directly from
disk to tape and making the load on the backup server negligible or even making
the backup server optional.
It minimizes the impact on database production servers by using hardware-based
mirroring technologies rather than intrusive software-based snapshots.
Additionally, the direct backup solution is fully integrated with the industry-standard
XCopy (ANSI T10 SCP-2 Extended Copy Standard) command embedded in HP
StorageWorks tape libraries (as well as external fiber channel SCSI bridges), removing
the need for a separate “data mover” appliance.
NOTE:
For an explanation of which applications, operating systems, and devices that direct
backup supports in HP Data Protector A.06.10, see “Supported
configurations” on page 250.
Direct backup
What does it mean to make a direct backup? This backup solution is “serverless,”
meaning that it does not use a dedicated backup server to move data and data is
not moved across the LAN. The data moves directly from the client system to be
backed up to a tape device without moving data through a backup server.
Direct backup supports backup of application data files and control files and of disk
images -- either raw disk or raw logical volume.
Direct backup uses existing split-mirror and Storage Area Network (SAN) technology
to:
• access application data with minimal impact to the application - there is minimal
utilization of the application server (providing little to no application down time)
• move data without encountering bottlenecks associated with network traffic and
LAN speed.
In order to support direct/serverless backups, Data Protector also incorporates new
technology to resolve target file systems and move data across the SAN. This new
technology, based on the XCopy standard, provides a method for moving data from
the target system to the tape device without moving data through a server. For a brief
explanation of XCopy, see “About XCopy” on page 247.
This direct disk to tape data path (over SAN) helps to lower the need for equipment
investment and boost the utilization of existing equipment.
244
Direct backup
Backup types
Direct backup supports backup of application data files and control files and of disk
images -- either raw disk or raw logical volume.
Direct backup benefits
Because the data mover is in the SAN bridge and the technology that interprets the
target system is built into the General Media Agent, direct backup users can use a
low-cost management server to drive the backup and can avoid having to invest in
multiple servers to perform block identification.
Additionally, direct backup is designed to let you leverage hardware capabilities to
increase uptime and instant recovery capabilities to reduce restore time.
Direct backup is not limited to proprietary file systems and logical volume managers
(LVMs).
Direct backup adds value to your backup solution both by addition and omission.
For example, direct backup:
• Takes advantage of the most advanced XCopy features to speed backups
• Enables maximum uptime by leveraging existing hardware mirroring and snapshot
capabilities
• Lets you take advantage of Data Protector’s industry-leading instant recovery
capability to speed recovery
• Requires very little CPU and memory resource from the XCopy host device
How direct backup works
As with any other Data Protector backup, you will make a backup specification to
control how and when the backup happens.
• The General Media Agent on the application server quiesces the application
• The split mirror agent on the application server and backup host splits the mirror
• The General Media Agent on the backup host:
• resolves the disk of the target system
• calculates the resolve information
• calls XCopy
• XCopy, in turn, retrieves the target data and moves it across the bridge to the
tape device.
Concepts guide
245
Figure 70 on page 246 shows a basic direct backup configuration. The Resolve Agent
is on a separate backup host in this configuration. Data, however, does not move
through this host.
Figure 70 Direct backup architecture
Environment
This section describes the direct backup environment in terms of the devices that need
to be connected and what they need to be connected to. It also discusses the required
agents and where they are installed.
For information about platform, tape drive, and library support, see “Supported
configurations” on page 250.
Direct backup requires the General Media Agent to be off the application server.
Also, the Resolve Media Agent must be on the application server or other host and
must have access to the XCopy engine. For discussion of Resolve Agent placement,
see “Supported configurations” on page 250.
Direct backup requires that:
• The disk array, XCopy engine, application server, and tape drive or library are
attached to the SAN.
• The Resolve host and application server are attached to the LAN.
246
Direct backup
• The HP StorageWorks Disk Array XP (XP) is configured as Business Copy (BC)
with mirrors configured with enough disk space.
• The SAN is properly configured in order to provide access to source (disks) and
target (tape) devices from both the XCopy engine and the host where the Data
Protector General Media Agent is running. This means that LUN masking and
SAN zoning must be configured so that:
• The General Media Agent host has access to the XCopy engine
• The General Media Agent host has access to the target tape drive or library
• The SSEA host has access to the source disk
• The XCopy engine has access to the source disk
• The XCopy engine has access to the tape drive or library
About resolve
The Resolve program is a proprietary component of Data Protector that understands
the native disk layout of many different file systems. Resolve allows Data Protector
direct backup to back up data written by many kinds of operating systems without
needing many servers running those operating systems.
Resolve examines the raw information on the disk and chooses the appropriate
method of interpreting the disk’s file system. Note that Resolve does not read the data
itself; it reads only information related to disk location. It then returns information
suitable for direct input to the XCopy engine.
About XCopy
XCopy is a National Committee for Information Technology (NCITS) standard that
allows two devices to communicate with one another without the help of another
computer/server in between.
XCopy specifies a set of SCSI commands that, when given to an XCopy engine,
allow the transfer of data from one device to another without the need to have a
computer/server in between. The data flows from the source device (either block or
streaming, i.e., disk or tape) to the destination device (either block or streaming)
through XCopy.
It assumes that the streaming (tape) device is set up and the device is ready to
write/read the data (that is, the drive is online, there is a tape in the drive, and the
tape is properly positioned at the starting point for the write/read.) This frees the
controlling server from reading the data from one device into its memory and then
writing the information to the destination device. With XCopy, all the server does is
send the XCopy commands to the XCopy engine and then wait for the results.
Concepts guide
247
XCopy + Resolve
Prior to Resolve, you needed a server with a matching file system to get this
information. This was because even with the right server, getting this information
could be difficult as the operating system might have converted the actual physical
sectors into its logical view before returning the information. Resolve removes the
need for multiple servers to support multiple file systems and the difficulties presented
by file system-specific information formats.
Direct backup process flow
The process flow for direct backup is listed below. These are the basic steps -- start
to finish -- of a direct backup.
•
•
•
•
•
•
•
•
Read backup specification
Determine what to back up
Quiesce application
Split mirror
Release application
Resolve blocks
Move data -- The XCopy engine
Reconnect and resynchronize mirror
Backup stages for data files
Original data files that are to be backed up go through several stages before they
end up as copies that are used later to perform restores. The direct backup process
(generally) follows these steps:
1.
Obtain data-file consistency (quiesce the application)
2. Read meta data (file attributes) and group files into objects
3. Obtain data-file stability (use Split-Mirror technology for point-in- time data
stability)
4. Map data files to list of disk blocks (using Resolve technology)
5. Move disk blocks to tape (using XCopy technology)
Typically each stage is managed by one Data Protector agent. Agents are spawned
by the Backup Session Manager (BSM). All errors that cannot be handled internally
by agents are reported through the BSM to the user and stored in the Internal
248
Direct backup
Database. The Backup Media Agent (BMA) writes catalog segments and delimiters
between data and catalog segments known as file-marks.
Restore
There are two restore options when using direct backup:
• If you are using the HP StorageWorks XP disk array and you have instant recovery
capability, you can use it to restore the data. For an explanation of using instant
recovery, see the HP Data Protector zero downtime backup administrator's guide.
• Restore of information backed up using direct backup can also be accomplished
using the standard Data Protector network restore.
Note that in either case it is important to verify that the application server can handle
the load of the restore. This is not an issue on the backup end, as the data does not
pass through the server during this process. On restore, however, the data does
impact the server.
Requirements and support
This section lists the requirements for using direct backup successfully and the file
systems and applications that direct backup supports.
•
•
•
•
•
Data Protector Cell Manager running on any supported operating system
Resolve Agent running on HP-UX 11.11
Support of application servers running HP-UX 11.11
Support of HP LVM on HP-UX 11.11
XCopy host, source disk, destination device, and XCopy engine must be in the
same SAN zone.
• File System support:
• Veritas’ VxFS 3.1, 3.3
• Application support:
• Oracle 9.i
•
•
•
•
•
Raw volume support
Support of ServiceGuard environments for the application server
Restore through the standard Data Protector restore interface
Support of instant recovery for the XP
XCopy engine in the bridge
Concepts guide
249
Supported configurations
Three hosts: CM, application, Resolve
This solution uses three hosts: one each for the Cell Manager, Resolve Agent, and
application. Although this configuration requires three machines, the Resolve host
can be a less expensive host and the resource load is split, helping you avoid
performance impact on the application.
Note that in this configuration, the Cell Manager host can be running any of the
operating systems supported by Data Protector. The application and Resolve Agent
hosts must be running HP-UX 11.11.
250
Direct backup
Figure 71 Basic three host configuration
Two Hosts: Cell Manager/Resolve Agent and application
This solution uses two hosts: one for the Cell Manager and Resolve Agent and one
for the application. Although this configuration does require two machines, the
resource load is split; this can help you avoid performance impact on the application.
Also, the machine that hosts the Cell Manager and the Resolve Agent can be of
minimal processing capacity.
Note that in this configuration, both hosts must be running HP-UX 11.11.
Basic configuration: single host
This solution uses a single host on which the Cell Manager, application, and Resolve
Agent are all installed. As all three components are running on the same physical
machine, they are sharing resources (I/O channels, CPUs, memory, etc.) for their
activities. This configuration minimizes the equipment required to facilitate direct
backup. However, because resources are shared, the Cell Manager and General
Concepts guide
251
Media Agent may impact the application database’s performance (XCopy’s processing
requirements are negligible).
Note that in this configuration, the host must be running HP-UX 11.11.
252
Direct backup
10 Disk backup
In this chapter
This chapter introduces the concepts associated with backing up data to disk and
the technologies that enable it. It also discusses the disk-to-disk backup configurations
that are supported by Data Protector.
It is organized as follows:
“Overview” on page 253
“Disk backup benefits” on page 254
“Data Protector disk-based devices” on page 255
Overview
Industry has requirements for increasingly faster methods of backing up and restoring
data. In addition, it has become more and more important that the time required for
data backup and restore be reduced to a minimum so as not to interrupt the day-to-day
running of company applications.
Many applications and databases frequently make small changes to existing files or
produce many new files containing business-critical data throughout the working
day. These files need to be backed up immediately to guarantee the data in them
will not be lost. This requirement means that a fast medium that can store large
amounts of data that works without interruption is necessary for storing data.
Disk-based storage media have become increasingly cheaper in recent years. At the
same time, the storage capacity of disks has risen. This has led to the availability of
low-cost, high-performance single disks and disk arrays for storing data.
Disk backup (also known as disk-to-disk backup) is becoming ever more important.
In the past, tape storage was the favored medium for backup and restore because
of its price and effectiveness in meeting disaster recovery requirements. Today, more
and more businesses are augmenting their tape storage backup solutions with faster
disk-based backup solutions. This ensures faster data backup and recovery.
Concepts guide
253
Disk backup benefits
There are many situations in which it is advantageous to use disk-based devices when
performing backups. Disk-based devices are, in fact, specific files in specified
directories, to which you can back up data instead of or in addition to backing it up
to tape. The following list indicates some situations in which disk-based devices are
particularly useful:
• Many applications and databases continuously generate or change a large
number of files, which contain business-critical data. Under these circumstances,
it is necessary to continuously back up the files concerned, in order to guarantee
the capability of restoring them without data loss.
In these environments, tape devices typically have to operate in stop/start mode,
because they do not receive a constant data stream. This may result in the tape
device limiting access to the files concerned. In addition, the lifetime of the backup
device may be greatly reduced.
Alternatively backups can be performed to any disk-based device, overcoming
the limitations described. As a short-term backup solution, this is adequate in
itself. If a longer term backup solution is required, the data in the disk-based
devices can be moved periodically to tape to free up the disk space. This process
is known as disk staging.
• In environments that have fast, high-capacity disk drives and slow tape drives,
you can shrink the backup window by performing backup to disk-based devices
first and moving the data to tape later.
• Using disk-based devices for backup enables you to take advantage of advanced
backup strategies such as synthetic backup.
• Disk-based devices are useful for providing fast restore capability for recently
backed up data. For example, backup data could be kept in a disk-based device
for 24 hours to enable fast, convenient restore.
• Mechanically, a disk-based device is quicker to use than a tape. When using a
disk-based device there is no need to mount and unmount a tape. When backing
up or restoring a small amount of data, a disk-based device is quicker because
it does not need the initialization time that a tape drive requires. With a disk-based
device there is no need to load or unload media, which consumes a significant
amount of time in a small backup or restore. The advantages of using a disk-based
device are even more evident when restoring from an incremental backup.
• The risk of media problems such as faulty tapes and tape mounting failures are
reduced to a minimum. The availability of RAID disk configurations provides
protection of data in cases where a disk fails.
• Overhead costs are reduced because there is no need for tape handling.
254
Disk backup
• Overall, disk-based storage space is becoming increasingly cheaper even if
compared to tape-based storage.
Data Protector disk-based devices
Data Protector has the following disk-based devices:
• Standalone file device
• File jukebox device
• File library device
Standalone file device
The standalone file device is the simplest disk-based backup device. It consists of a
single slot to which data can be backed up. Once configured, its properties cannot
be changed. The file device has a maximum capacity of 2 TB, if this file size is
supported by the operating system on which the device is running.
File jukebox device
The file jukebox device is a special version of the Data Protector jukebox device. The
jukebox device can be configured to back up either optical or file media. The jukebox
device used to back up file media is referred to as the file jukebox device. The type
of media to be backed up by the jukebox is specified during device configuration.
The file jukebox device consists of multiple slots to which you can back up data.
Configuration is a two phase process, firstly a file jukebox device is created and
then one or several drive(s) is configured for it. Once the device has been configured
it is possible to change its properties. Each slot in the file jukebox device has a
maximum capacity of 2 TB. The device’s maximum capacity is equal to:
Number of slots X 2 TB
File library device
The file library device is the most sophisticated disk-based backup device. It has
multiple slots called file depots to which you can back up data. The configuration of
the file library device is completed in a single stage. It is possible to change the
properties of the file library device at any time. The device’s maximum capacity is
the same as the maximum that can be saved on the filesystem on which the device
resides. Each file depot has a maximum capacity of up to 2 TB. File depots are
created automatically as required.
The file library device has intelligent disk space management. It anticipates potential
problems saving data to it. A warning message is written in the event log if the amount
Concepts guide
255
of free disk space approaches the configured minimum amount required for the
device to work. This enables you to free more disk space in good time for the device
to continue saving data. If all the space allocated to the file library device is ever
completely used, a warning message appears on the screen with instructions as to
how to solve the problem.
The file library device automatically creates more file depots if a particular backup
requires more space than is available in a single file depot.
Recommended disk-backup device
Hewlett-Packard recommends using the file library device as the preferred disk-based
backup device. The file library device is the most flexible and intelligent of the set of
disk-based backup devices. It can be re-configured at any time during use and is
capable of performing more sophisticated disk space handling than any other
disk-based backup devices. Furthermore, it enables the use of advanced backup
strategies such as synthetic backup.
For description of the file library device functionality, see the online Help index: “file
library devices“.
Data format
The data format of the disk-based devices is based on the tape data format. Data
Protector converts the data to be backed up into tape format before it writes the data
to the disk-based device.
With file libraries used for virtual full backup, distributed file media format must be
used. Select this format in the device’s properties.
Configuration
It is possible to set properties for the all disk devices both during the initial device
setup and after the devices are in operation. The degree of changes that can be
made to the properties of each device vary according to the device.
Backing up to a disk device
A backup can be made to a disk-based device by creating a normal Data Protector
backup specification.
256
Disk backup
11 Synthetic backup
In this chapter
This chapter introduces the concept of synthetic backup and explains the synthetic
backup solution provided by Data Protector.
It is organized as follows:
“Overview” on page 253
“Disk backup benefits” on page 254
“Data Protector disk-based devices” on page 255
“Restore and synthetic backup” on page 260
Overview
With the volume of data increasing and backup windows shrinking, performing a
full backup often presents a problem in terms of time and storage space. On the
other hand, having many incremental backups can be problematic because each
incremental increases the time needed to perform a restore.
As backup to disk is gaining popularity due to the high performance and capacity
as well as increasingly lower price of disks, new opportunities have arisen. The
industry's requirements are to minimize the backup window, minimize the load on
production servers and the network, and enable a quick restore. These requirements
are met by synthetic backup.
Synthetic backup is an advanced backup solution that produces a synthetic full
backup, an equivalent to a conventional full backup in terms of data, without putting
stress on the production servers or the network. A synthetic full backup is created
from a previous full backup and any number of incremental backups.
Performing synthetic backup eliminates the need to run regular full backups. Instead,
incremental backups are run, and subsequently merged with the full backup into a
Concepts guide
257
new, synthetic full backup. This can be repeated indefinitely, with no need to run a
full backup again.
In terms of restore speed, a synthetic full backup is equivalent to a conventional full
backup. The restore chain consists of only one element, so a restore is as quick and
simple as possible.
Synthetic backup benefits
Synthetic backup brings the following benefits:
• It eliminates the need for full backups. After the initial full backup, only incrementals
are performed, which significantly reduces the time needed for the backup.
• Consolidation of backed up objects is performed on the device server, putting
no stress on either the production servers or the network.
• A type of synthetic backup, called virtual full backup, is even more efficient. Virtual
full backup consolidates data using pointers, which eliminates unnecessary
duplication of data.
• A restore from a synthetic full backup is as fast as from a conventional full backup,
as there is no need to retrieve data from incremental backups. This eliminates the
reading of each incremental backup in the restore chain, and if tape devices are
used, also loading and unloading of several media and seeking for object versions.
How Data Protector synthetic backup works
Data Protector synthetic backup enables you to merge a full backup and any number
of incremental backups into a new, synthetic full backup.
To enable synthetic backup, the use of enhanced incremental backup is required.
Enhanced incremental backup must be turned on before the full backup and the
incremental backups are performed.
A synthetic full backup can be created from a full backup that is written to a disk or
tape device and incremental backups that are written to a disk-based device, a Data
Protector file library. The synthetic full backup can, again, be written to a disk or
tape device.
If all the backups, full and incremental, are written to the same file library that uses
distributed file media format, an even more efficient type of synthetic backup is
available, called virtual full backup. This solution uses pointers to consolidate data
rather than copy the data. As a result, the consolidation takes less time and avoids
unnecessary duplication of data.
258
Synthetic backup
The following figures explain the concept of synthetic backup and virtual full backup.
They show how a synthetic full backup or a virtual full backup is created from a full
backup and any number of incremental backups.
Figure 72 Synthetic backup
Figure 72 on page 259 shows how a synthetic full backup is created. The Restore
Media Agent (RMA) reads the full backup from the backup medium, which can be
a tape or a disk. The data is sent to another RMA, which reads the incremental
backups from the file library and consolidates the data. The consolidated data is
then sent to the Backup Media Agent (BMA), which writes the synthetic full backup
to the backup medium, which can, again, be a tape or a disk.
Later on, the synthetic full backup is typically merged with subsequent incremental
backups into a new synthetic backup. The procedure can be repeated indefinitely,
either after each incremental backup, or at a desired interval.
Concepts guide
259
Figure 73 Virtual full backup
Figure 73 on page 260 shows how a virtual full backup is created. With this type of
backup, all the backups reside in a single file library that uses distributed file media
format. The Restore Media Agent (RMA) reads the information about the full backup
and the incremental backups, and generates the data for the virtual full backup. The
generated data is sent to the Backup Media Agent (BMA), which creates the virtual
full backup in the file library.
Synthetic backup and media space consumption
If synthetic backups are performed frequently, and the sources are kept, this typically
means significant space consumption on the backup media. However, if virtual full
backup is performed, the backup media space consumption is minimized.
With virtual full backup, the space consumption largely depends on the size of the
backed up files. If the files are significantly larger than the block size used, virtual
full backup achieves maximum savings of the space compared to normal synthetic
backup. On the other hand, if the files are smaller than the block size, the savings
are rather small.
Restore and synthetic backup
Restore from a synthetic full backup is equivalent to restore from a conventional full
backup. The following figures present different situations, supposing you need to
restore your data to the latest possible state. In all examples, a full backup and four
incremental backups of the backup object exist. The difference is in the use of synthetic
backup.
260
Synthetic backup
Figure 74 Full and incremental backups
In Figure 74 on page 261, conventional backups were performed. To restore to the
latest possible state, you need the full backup and all four incremental backups. The
restore chain consists of five elements, which often reside on different media.
Such a restore can take a considerable amount of time, as each incremental backup
must be read. If tape devices are used, time is spent for loading and unloading of
several media and seeking for object versions to restore.
Figure 75 Synthetic backup
In Figure 75 on page 261, a synthetic full backup exists, which is used for restore by
default. The restore chain consists of only two elements, namely the synthetic full
backup and the subsequent incremental backup. The restore is significantly simpler
and quicker than that without the synthetic full backup. In the figure, both possible
restore chains are shown.
Concepts guide
261
Figure 76 Regular synthetic backup
Figure 76 on page 262 shows a situation where a synthetic backup was performed
after each incremental backup. This strategy enables the simplest and quickest restore
to the latest possible state, or to any earlier point in time that was backed up. Only
one element is required for restore, namely the synthetic full backup of the desired
point in time.
Figure 77 Synthetic backup and object copy
In Figure 77 on page 262, a synthetic backup was performed and then copied. This
provides additional safety. The restore to the latest possible state can use any of the
three different restore chains shown. By default, Data Protector selects the optimum
restore chain, which normally includes the synthetic full backup or its copy. In case
of missing media, a media error, or similar, an alternative restore chain is used.
How data protection periods affect restore from synthetic backup
Data protection of a conventional full backup and all incremental backups that precede
synthetic full backup does not compromise a successful restore.
262
Synthetic backup
By default, the last synthetic full backup in the backup chain is used for restore,
irrespective of whether the preceding backups are still valid or their protection has
already expired and the objects are removed from the IDB.
For additional safety, set data protection to permanent so that data on the media is
not overwritten unintentionally.
Concepts guide
263
264
Synthetic backup
12 Split mirror concepts
In this chapter
This chapter introduces the split mirror backup concept and discusses the
configurations that are supported by HP.
It is organized as follows:
“Overview” on page 243
“Supported configurations” on page 269
Overview
Modern high availability (HA) storage configurations introduce new demands on
backup concepts. The configuration consists of one of numerous variations of single
or multiple mirror structures.
The usual approach is to use one replica (mirror copy) for the backup task, while the
source volumes still serve the application. See Figure 78 on page 266.
Concepts guide
265
Figure 78 Split mirror backup concept
The target volumes in replica are typically connected to a separate client, which also
has tape devices connected to allow for local backup. Usually, hardware mirror
technologies such as HP StorageWorks Disk Array XP or EMC Symmetrix are used
to create a replica such as:
• HP StorageWorks ContinuousAccess XP or
• HP StorageWorks BusinessCopy XP
The availability of the application remains almost permanent, with the exception of
a short period of time (lasting several seconds to a few minutes). This time is needed
to make the data on the disk consistent and perform the actual split of the mirrors.
The data must be consistent so that the application can make use of the data after a
restore. Normally, the replica is not created at the time of backup, but is already
available and synchronized to provide high availability to the application. The backup
and the resyncing of the replica does not affect the application performance, since
this occurs in parallel on separate hardware.
As the application client and backup client are different (in most cases), it is very
important that all cached information (database cache, filesystem cache) on the client
is flushed to the disk before the backup mirror is split off. One of the following options
can achieve this:
266
Split mirror concepts
• Databases can be put into backup mode
• Databases can be taken offline
• A mount point can be unmounted
Only when this is carried out before a replica is consistent. However, if a database
is running on a filesystem or a rawdisk, there is no need to unmount the filesystem
or rawdisk as the database ensures that data is really written to the disk and not to
the filesystem cache.
For an online database backup, a raplica alone cannot be restored. The archive log
files from the application client are also needed. An archive log backup can be
started right after a split, when the database is taken out of backup mode.
The use of one replica in combination with the HP StorageWorks ContinuousAccess
XP technology to perform the backup does take away high availability of storage for
the duration of the backup. Additional mirrors retain full high availability of storage
and allow for the same backup approach.
The backup client can be a centralized backup client for multiple application clients
running different applications. In such cases, the backup client must run on the same
operating system as the application client, so as to access mirrored resources in a
native way.
The backup client should be capable of performing backups in a reasonable amount
of time. Though, theoretically, almost 24 hours may be required to perform a backup,
the restore time must be considered as well. It is thus recommended to have a backup
client that can perform the backup in 2 to 4 hours. It is recommended to perform the
restore through the application client.
In this approach the bulk of the data transfer happens via the backup client and its
access to the replica. The LAN connection between the backup client and application
client is only used to coordinate processes that are involved in the backup. There are
processes running on each client to allow the automation of the split.
Instant recovery
Data Protector instant recovery takes advantage of the split mirror technology to
provide instant data restore. The solution is based on zero downtime backup (ZDB)
solutions like the HP StorageWorks Disk Array XP integration, which uses split mirror
technology.
During a split mirror backup session, a replica is used for the purpose of moving the
data to a backup medium (tape). After a backup is completed, the replica can be
discarded and disk pair prepared for the next backup session by resynchronization,
or the replica can be left unchanged for the purpose of instant recovery. Several
replicas can exist at the same time. For example, HP StorageWorks Disk Array XP
Concepts guide
267
allows up to three replicas, and each can have an additional two copies if cascading
is used.
During the instant recovery, the data on the specified replica (left unchanged for the
purpose of instant recovery) is synchronized to the application client source volumes
without restoring from a backup medium.
Data Protector will only use the first three replicas because secondary mirrors cannot
perform fast-resynchronization, which is critical for ensuring minimal restore time.
Instant recovery is only possible using the HP StorageWorks BusinessCopy XP
configuration (local mirror - dual host and local mirror - single host configurations).
ZDB to tape and ZDB to disk+tape
During ZDB-to-tape and ZDB-to-disk+tape sessions, a replica of the application data
is streamed to a tape device, which is connected to a separate backup system, using
Data Protector Disk Agent and General Media Agent, with minimal impact on the
application system. After the backup is completed, the replica is either:
• discarded - ZDB to tape
• retained and can be used for instant recovery - ZDB to disk+tape
ZDB to disk
During a ZDB-to-disk session, the original data is not moved to a backup medium
(tape) from the replica. The replicas (up to three) can be used for various purposes,
such as offline data processing or instant recovery; the latter is possible only if HP
StorageWorks BusinessCopy XP configuration was used. It is only possible to restore
objects from a ZDB-to-disk session by using the instant recovery functionality.
Replica set rotation
Several replicas can exist at the same time. HP StorageWorks Disk Array XP allows
up to three replicas, and each can have an additional two copies if cascading is
used. Data Protector can use only disks from the first three replicas (first level mirrors
or MUs) for backup and instant recovery purposes. The additional six copies
(cascading mirrors) are not supported.When configuring a ZDB backup specification
for a source volume (LDEV) with first level mirrors configured or when restoring to
such a source volume, it is, using Data Protector, possible to define a replica set from
which this integration selects one replica for the current session.
Backup clients and clusters
The backup client should not be used as a failover server for the application client.
It is recommended to have application and backup services on separate clusters.
268
Split mirror concepts
Supported configurations
Local mirror - dual host
This solution uses a local mirroring functionality such as Business Copy XP. Both disks
are in the same disk array, which means the I/O infrastructure of the RAID system is
actually shared between the application client (or host) and the backup client.
As the application client and the backup client are two physically different systems,
they can use their own resources (I/O channels, CPUs, memory, etc.) for their
dedicated activities, such as backup, without interfering with each other. In this way,
the backup performance does not impact the database performance.
Figure 79 Local mirror - dual host (full performance, Zero Downtime
Backup)
The Data Protector split mirror backup integration allows automatic handling of mirror
status as well as tight integration with applications such as SAP R/3 and Oracle (to
ensure data consistency and application/database-aware backups). Only if the
application/database is aware of a backup can a secure operation be guaranteed
and native application tools be used for restore. The impact of a backup on the
application is reduced to the time needed to perform a split of the mirror and put the
Concepts guide
269
database into a consistent mode that permits the split, as well as to take it out of this
mode again.
This configuration enables an offline backup of a very large database in a short time,
as well as an online backup that creates very few archive log files, since the backup
mode time of the database is kept to a minimum.
A small number of archive logs reduces the space needed for the archive logs in
total, as well as speeds up the recovery process of the database. After a restore of
an online database, a recovery is needed to return the database to a consistent state.
All archive logs that have been created during the backup must be applied. In a split
mirror backup, only the archive log files created during the split are applied.
Local mirror - single host
In cases where no dedicated backup server is available, both functions (application
and backup) are performed on the same client (or host). Offline backups of mail
applications, for instance, could reduce the downtime of the application to minutes
instead of hours in this way.
In this type of configuration, only disk image (raw disk) and filesystem backups are
supported. Database and application backups, like Oracle and SAP R/3, cannot be
supported, since the database has to be mounted on the backup server, which would
not be possible on the same server that has the database already mounted.
Remote mirror
Remote mirror technology, such as Continuous Access XP, enhances the configurations
shown earlier due to the fact that the backup and application processes utilize different
disk array resources at different locations.
270
Split mirror concepts
Figure 80 Split mirror - remote mirror (LAN-free remote backup - data
HA)
The remote mirror transfers data to a physically separate site where it can be backed
up to locally available tapes. This allows the separation of production data from
backup data, eliminating the risk of a fire or other disaster damaging both the
production and the backup environment at the same time.
No network resources are required to sync the mirrors during a backup. Although
data is not transferred through the network, Data Protector still needs the
communication between the Cell Manager and its clients.
This solution allows you to centralize a backup service by mirroring the application
data from several production sites (A and C in this case) to a central location or
central disk array. In this way, your investment in a backup service (server and tape
library) can be consolidated and combined with the high availability of a remote
mirror configuration.
The remote site cannot be used as an automatic disaster recovery site during the time
of the backup, as the link between the two sites is split for the duration of the backup
(and both disks are out of sync). This means that in case of a site A failure, site B
cannot take over automatically (as it normally would) for x hours (x being the time
the data takes to stream to the tape). This problem applies to local mirroring as well.
However, it is particularly important for the remote solution, as the concept of a
Concepts guide
271
remote disaster recovery site using hardware mirror concepts is widely accepted in
the industry.
Local/remote mirror combination
If the customer has a need for a permanently available recovery site (provided, for
example, by a MetroCluster) in addition to a zero downtime backup solution, the
combination of a remote mirror and a local mirror can be used.
This solution allows for full split mirror advantage together with a full recovery solution
at the remote site. In this example, the remote mirror is constantly maintained with
only the local link split for backup purposes. This gives the cluster the continuous
ability to fail over to the remote site (site B).
Figure 81 Local/remote mirror combination (disaster recovery
integrated backup [Service HA - HP-UX only])
In order to have the failover functionality independent of the backup operation, the
backup client must be a separate additional client outside the cluster. If a MetroCluster
solution is implemented, the cluster arbitration client could be the backup client.
272
Split mirror concepts
Other configurations
There are many other possible split mirror configurations that provide some particular
advantage or fulfill a specific user need. However, each configuration has its specific
behavioral pattern that imposes specific requirements on the control functions in order
to guarantee backup and recovery. It is important to control and specify which
configurations are supported.
All the configurations shown above are supported by HP. For an updated list of
supported configurations,see the following URL: http://www.hp.com/support/
manuals.
In the event that you want to back up data in a configuration not listed, this does not
mean that it cannot be supported. Please contact your local HP representative or HP
Consulting to investigate the supportability of additional configurations.
Concepts guide
273
274
Split mirror concepts
13 Snapshot concepts
In this chapter
This chapter introduces the snapshot backup concepts and discusses the configurations
that are supported by HP.
It is organized as follows:
“Overview” on page 275F
“Supported configurations” on page 281
Overview
The rapidly expanding requirement for high availability storage configurations has
led to the introduction of new zero downtime backup (ZDB) technologies. The
advances in storage virtualization technology have provided the opportunity for an
alternative to conventional split mirror technology.
Within the Data Protector ZDB solution, different disk array technologies are combined
with the latest developments in the snapshot technology, to create snapshots of
application or database data stored on a disk array. These snapshots can
subsequently be kept on a disk array as point-in-time copies of the original data for
instant recovery purposes or can be used to produce ZDB-to-tape sessions on a
backup system. The processes concerned have minimal impact on the application
server, providing an effective ZDB solution.
Storage virtualization
The term “storage virtualization” is used to describe the technology that separates
the logical representation of storage from the actual physical storage components.
This means the creation of logical volumes out of a pool of physical disks residing
in a disk array. A logical volume is limited by the boundaries of the pool, but may
span over any number of physical disks within the disk array. Logical volumes can
be presented to one or multiple host systems. You cannot have control over the exact
Concepts guide
275
allocation of logical volumes on physical disks, but you can influence it with a choice
of protection characteristics.
RAID
Redundant Array of Inexpensive Disks (RAID) technology is used to control the way
in which the data is distributed across the physical disks within a disk array. Various
levels of RAID are available, providing different levels of data redundancy and data
security, speed and access time. For example, RAID0 provides no duplication of
data, RAID1 provides duplication of all data, RAID5 provides protection of data by
parity.
The snapshot integrations for Data Protector are designed to work with disk arrays
that use the snapshot technology, such as HP StorageWorks Virtual Array and HP
StorageWorks Enterprise Virtual Array.
Snapshot concepts
In a typical basic setup using the snapshot technology, a single disk array might be
connected to separate application and backup systems. The disk array can be used
as a storage device by both the application system and the backup system and logical
volumes can be mounted on either. Using this arrangement, the application system
uses logical volumes within the disk array to store its data during its normal operation.
The logical volumes storing the application system data are for the needs of Data
Protector snapshot integrations also referred to as source volumes. When a snapshot
backup is performed, the application data residing on the source volumes is replicated
and written to other logical volumes of the same disk array, also referred to as target
volumes. This replicated data is also referred to as snapshot data and presents an
almost instantaneous point-in-time copy of a given filesystem or volume. The set of
thus created target volumes is referred to as replica. Once the replica for snapshot
data is created, the primary data can continue being modified without affecting the
backup operation.
276
Snapshot concepts
Figure 82 Snapshot backup
The backup client is set up as a Data Protector client with tape devices connected,
to allow a local backup to be performed.
When a backup session begins, the application client enters the backup mode of
operation while the backup client is being prepared for the backup process; a
snapshot of the application data is produced.
Once the backup client is ready and the replica for the snapshot data is created, the
application is returned to normal operation.
During the time that the application client is in backup mode (or the application may
be stopped for a brief period, depending on the application), the impact on
application availability is minimal.
If a ZDB to tape is specified, the snapshot data is then streamed to tape media on
the backup client. During the tape media streaming operation, the application client
can run undisturbed.
Since the application client and backup client are different (in most cases), it is very
important that all cached information (database cache, filesystem cache) on the
application client is flushed to the array before the snapshot is made. One of the
following options can achieve this:
Concepts guide
277
• Databases can be put into backup mode
• Databases can be taken offline
• A mount point can be unmounted
For an online database backup, snapshot data alone does not suffice for a restore.
The archive log files from the application client are also needed. An archive log files
backup utilizing the standard Data Protector backup procedure can be started
immediately after creating snapshots, when the database is taken out of backup
mode.
Snapshot data of the application data is produced using the virtual disk array
technologies, such as:
• HP StorageWorks Business Copy Virtual Array
• HP StorageWorks Enterprise Virtual Array
Snapshot backup types
Within the Data Protector snapshot integrations, the following types of snapshot
backups are available:
• ZDB to tape
• ZDB to disk
• ZDB to disk+tape
ZDB to tape and ZDB to disk+tape
During ZDB-to-tape and ZDB-to-disk+tape sessions, a point-in-time snapshot data of
the application data is streamed to a tape device, which is connected to a separate
backup system, using Data Protector Disk Agent and General Media Agent, with
minimal impact on the application system. After the backup is completed, the snapshot
data is either:
• discarded - ZDB to tape
• retained and can be used for instant recovery - ZDB to disk+tape
ZDB to disk
During a ZDB-to-disk session, the same standard snapshot technology is used as in
ZDB to tape and ZDB to disk+tape, however, the snapshot data is not streamed to
a backup medium (tape device) from the snapshot copy and is retained on a disk
array. It can be used for instant recovery. The session effectively ends after the
snapshot data is created.
278
Snapshot concepts
Instant recovery
During snapshot backup sessions, several snapshot copies of data can be produced
and can be retained on a disk array, each point-in-time copy in its own replica. The
retained snapshot copies of data can then be used for various purposes, such as
offline data processing or instant recovery. Only the point-in-time copies produced
during ZDB-to-disk and ZDB-to-disk+tape sessions can be restored using the instant
recovery functionality.
Using the instant recovery functionality, the point-in-time copy from a selected replica
is restored within a disk array and returned to its state at the point in time that the
snapshot data was produced. This process does not involve any restore of data from
tape media, dramatically reducing the overall restore time.
Application archive log files are not included in snapshot backup, therefore to restore
and apply them, they need to be restored from tape media.
Replica set and replica set rotation
The maximum number of replicas that can be kept concurrently on a disk array is
dependant on the disk array used. The replicas kept on the disk array for the same
backup specification form the replica set for that backup specification. The replica
set is defined by the maximum number of replicas that are to be kept on a disk array
for a particular backup specification. When during a snapshot backup session, this
number is reached, the snapshot data in the oldest replica in the replica set is
overwritten; if the number is not reached yet, a new replica is created - these two
actions are referred to as replica set rotation.
Types of snapshots
Depending on a disk array used, different types of snapshots can be created during
a Data Protector snapshot backup session. The Data Protector snapshot integrations
utilize the following types of snapshots:
• copy-on-write snapshots with the preallocation of disk space
• copy-on-write snapshots without the preallocation of disk space
• snapclones
Snapshots with the preallocation of disk space
The creation of copy-on-write snapshots with the preallocation of disk space requires
the same amount of disk capacity to be allocated as for the source volume. Data is
Concepts guide
279
not written to that reserved space until necessary. As the data changes on the source
volume, the snapshot data on the target volume is updated with the original data.
Since this snapshot technique caches only the difference between the ever-changing
original data content against the point-in-time state, copy-on-write snapshots with the
preallocation of disk space are depended on their source volumes; if the data on
source volumes is lost, the associated snapshots are useless.
Snapshots without the preallocation of disk space
Copy-on-write snapshots without the preallocation of disk space also represents a
point-in-time copy of the original data but it does not require preallocation of disk
capacity. The disk capacity is allocated dynamically on as-needed basis. As the data
on source volume changes, free space in a disk array is used for the creation of the
snapshot. Copy-on-write snapshots without the preallocation of disk space are intended
to be short-lived snapshots. Note that their size grows dynamically and may eventually
run out of storage capacity if they are not deleted regularly.
The main benefit of copy-on-write snapshots without the preallocation of disk space
over copy-on-write snapshots with the preallocation of disk space is in significant
reduction of costs. Considerably less additional storage capacity for replication space
is needed, if the snapshots are deleted regularly, than with a standard snapshot
technology.
Since this snapshot technique caches only the difference between the ever-changing
original data content against the point-in-time state, copy-on-write snapshots without
the preallocation of disk space are depended on their source volumes; if the data
on source volumes is lost, the associated snapshots are useless.
Snapclones
The first part of the snapclone creation is similar to the creation of a copy-on-write
snapshot with the preallocation of disk space, which is followed by the cloning
process. During this process, all data from the source volume is copied to the target
volume. A snapclone enables immediate access to the replicated data while the
cloning process runs in the background using the disk array idle time. When the
cloning process is finished, the snapclone becomes a full data copy that represents
a point-in-time state of the source volume; if the data on source volume is lost, you
can always revert to the snapclone.
280
Snapshot concepts
Supported configurations
Basic configuration: single disk array - dual host
Both hosts are connected to the same disk array, so that the I/O infrastructure of the
RAID system is actually shared between the application client and the backup client.
As the application client and the backup client are two physically different systems,
they can use their own resources (I/O channels, CPUs, memory, etc.) for their
dedicated activities, such as backup, without interfering with each other. In this way,
the impact of the backup on the database performance is minimal.
Figure 83 Single disk array - dual host (full performance, Zero
Downtime Backup)
The Data Protector snapshot integrations allows automatic handling of disk array
status as well as tight integration with applications such as SAP R/3, Oracle or
Microsoft SQL or Exchange Server (to ensure data consistency and
application/database-aware backups). Only if the application/database is aware
of a backup can a secure operation be guaranteed and native application tools be
used for restore. The impact of a backup on the application is reduced to the time
to perform the following steps:
Concepts guide
281
1.
Put the database into a consistent mode that permits a snapshot to be taken.
2. Perform a snapshot of the application data.
3. Return the database to normal operating mode.
This configuration enables an offline backup of a very large database in a short time,
as well as an online backup that creates very few archive log files, since the backup
mode time of the database is kept to a minimum.
The small number of archive logs reduces the space needed for the archive logs in
total, as well as speeding up the recovery process of the database. After a restore
of an online database, a recovery is needed to return the database to a consistent
state. All archive logs that have been created during the backup must be applied.
In a snapshot backup, only the archive log files created during the snapshot are
applied.
Other supported configurations
Figure 84 Multiple disk arrays - dual host
With this solution, both hosts are connected to multiple disk arrays. The I/O
infrastructure of the RAID systems is shared between the application client and the
backup client.
282
Snapshot concepts
Figure 85 Multiple application hosts - single backup host
With this solution, multiple application hosts may be connected to a single or multiple
disk arrays, which are, in turn connected to a single dedicated backup host. The I/O
infrastructure of the RAID systems is shared between the application clients and the
backup client.
Concepts guide
283
Figure 86 Disk array(s) - single host
In cases where no dedicated backup server is available, both functions (application
and backup) can be performed on the same client (or host). Offline backups of mail
applications, for instance, could reduce the downtime of the application to minutes
instead of hours in this way.
284
Snapshot concepts
Figure 87 LVM mirroring - HP StorageWorks Virtual Array only
In the supported configurations described previously, only Business Copy functionality
is possible with the HP StorageWorks Virtual Array integration. However, by using
LVM mirroring, it is possible to create snapshot copies of data between different
virtual arrays, writing to both at the same time. This allows the emulation of the
Continuous Access plus Business Copy functionality that is available with HP
StorageWorks Disk Array XP.
Concepts guide
285
Figure 88 Campus Cluster with LVM Mirroring - HP StorageWorks
Virtual Array only
With this configuration, it is possible to emulate Continuous Access plus Business
Copy functionality, with standard cluster failover functionality. This is often required
for mission critical applications.
Backup clients and clusters
The backup client should not be used as a failover server for the application client.
It is recommended to have application and backup services on separate clusters.
Other configurations
There are many other possible disk array configurations that provide some particular
advantage or fulfill a specific user needs. However, each configuration has its specific
behavioral pattern that imposes specific requirements on the control functions in order
to guarantee backup and recovery. It is important to control and specify which
configurations are supported.
Only the configurations shown are supported by HP. For an updated list of supported
configurations, see the following URL: http://www.hp.com/support/manuals.
286
Snapshot concepts
In the event that you want to back up data in a configuration not listed, this does not
mean that it cannot be supported. Please contact your local HP representative or HP
Consulting to investigate the supportability of additional configurations.
Concepts guide
287
288
Snapshot concepts
14 Microsoft Volume Shadow
Copy service
In this chapter
This chapter introduces the Microsoft Volume Shadow Copy service (VSS) concept
and its role in the backup and restore process. It also outlines the backup and restore
flow when using this feature.
The chapter is organized as follows:
“Overview” on page 289
“Data Protector Volume Shadow Copy integration” on page 293
“VSS filesystem backup and restore” on page 295
For detailed information on the integration, see the HP Data Protector integration
guide. For detailed information on the filesystem backup and restore, see the Data
Protector online Help.
Overview
A traditional backup process is based on the direct communication between the
backup application (application, which initiates and performs backup) and an
application to be backed up. This backup method requires from the backup
application an individual interface for each application it backs up.
The number of applications on the market is constantly increasing. The necessity of
handling application specific features can cause difficulties in backup, restore, and
storage activities. An effective solution to this problem is introducing a coordinator
among the actors of the backup and restore process.
Concepts guide
289
VSS
Volume Shadow Copy service (VSS) is a software service introduced by Microsoft on
Windows operating systems. This service collaborates with the backup application,
applications to be backed up, shadow copy providers, and the operating system
kernel to implement the management of volume shadow copies and shadow copy
sets.
The idea of the Volume Shadow Copy service is to provide a unified communication
interface that can coordinate backup and restore of any application regardless of
their specific features. With this approach, a backup application does not need to
handle each application to be backed up specifically. However, this approach is
applicable to a backup application only in case it conforms to the VSS specification.
What is a shadow copy?
A shadow copy refers to a volume that represents a duplicate of the original volume
at a particular moment in time. The volume shadow copy technology provides a copy
of the original volume at a certain point in time. The data is then backed up from the
shadow copy, not from the original volume. The original volume continues to change
as the backup process continues, but the shadow copy of the volume remains constant.
Shadow copy is basically a snapshot backup, which allows applications and users
to continue writing to data volumes, even if they are in the middle of a backup
process, while the backup is getting data from a shadow copy of the original volume.
A shadow copy set is a collection of shadow copies created in the same point in
time.
What is a writer?
A writer refers to any process that initiates change of data on the original volume.
Writers are typically applications (for example, MSDE Writer for MS SQL Server) or
system services (for example, System Writer and Registry Writer) that write persistent
information on a volume. Writers participate in the shadow copy synchronization
process by assuring data consistency.
What is a shadow copy provider?
A shadow copy provider refers to some entity that performs the work involved in
creating and representing the volume shadow copies. Shadow copy providers own
the shadow copy data and expose the shadow copies. Shadow copy providers can
be software (including a system provider, MS Software Shadow Copy Provider) or
hardware (local disks, disk arrays).
290
Microsoft Volume Shadow Copy service
The example of the hardware provider is disk array, which has its hardware
mechanism of providing point-in-time state of a disk. A software provider operates
on physical disks and uses software mechanism for providing point-in-time state on
a disk. The system provider, MS Software Shadow Copy Provider, is a software
mechanism, which is a part of the Windows Server 2003 operating system.
The VSS mechanism guarantees that all hardware providers will be offered for creating
shadow copy before all software providers. If none of them is able to create a shadow
copy, VSS will use the MS Software Shadow Copy Provider for the shadow copy
creation, which is always available.
Data Protector and VSS
The Volume Shadow Copy service enables coordination among the backup
application, writers, and shadow copy providers during the backup and restore
process.
Figure 89 on page 292 and Figure 90 on page 292 show differences between the
traditional backup model and the model with the VSS coordinator.
Concepts guide
291
Figure 89 Actors of the traditional backup model
Figure 90 Actors of the VSS backup model
In the traditional model, the backup application had to communicate with each
application it backed up individually. In the VSS model, the backup application
communicates with the VSS only, and the VSS coordinates the whole backup process.
292
Microsoft Volume Shadow Copy service
VSS benefits
The advantages of using Volume Shadow Copy service are as follows
• A unified backup interface for all writers.
• A unified backup interface for all shadow copy providers.
• Writers provide data integrity at application level. Intervention from the backup
application is unnecessary.
Data Protector supports the Microsoft Volume Shadow Copy service at two levels:
• Within the Microsoft Volume Shadow Copy service integration, Data Protector
provides a shadow copy backup and restore of VSS-aware writers, including
ZDB and instant recovery functionality.
• Within the Disk Agent functionality, Data Protector provides VSS filesystem backup.
The Data Protector VSS integration supports a consistent shadow copy backup only
for VSS-aware writers. Consistency in this case is provided by the writer. Whenever
applications are not VSS-aware, a shadow copy is created. The consistency of the
shadow copy data is not guaranteed at application level, however, it is improved in
comparison to a non-VSS filesystem backup.
The table below outlines the differences between using Data Protector VSS integration
backup, VSS filesystem backup, and non-VSS filesystem backup:
Table 14 Benefits of using VSS
Data Protector VSS
integration backup
VSS filesystem backup
Non-VSS filesystem
backup
Open files
No open files.
No open files.
If files are open, backup
may fail.
Locked files
No locked files.
No locked files.
If files are locked,
backup skips them.
Data
integrity
Provided by the writer.
Crash consistency state
(in the event of a power
failure, for example).
None (inherent).
Data Protector Volume Shadow Copy integration
The Data Protector integration with Microsoft Volume Shadow Copy service provides
full support for VSS-aware writers. This includes automatic detection of VSS-aware
Concepts guide
293
writers, and backup and restore functionality. For detailed information on the
integration, see the HP Data Protector integration guide.
VSS backup
In case of VSS-aware writers’ backup, the consistency of data is provided at writer
level and does not depend on the backup application. Data Protector follows the
requirements provided by the writers when selecting what to back up.
During the backup of VSS-aware writers, Data Protector does not communicate with
each writer individually, but through the VSS interface. It uses the VSS integration
agent to connect the Volume Shadow Copy service, which coordinates the backup
process. VSS provides Data Protector with the writer-related metadata necessary for
performing a consistent backup and restore. Data Protector examines this data and
identifies the volumes to be backed up. Data Protector then requests VSS to create
a shadow copy of the specified volumes.
NOTE:
A Writer Metadata Document (WMD) is metadata provided by each writer. Writers
identify themselves by the metadata and instruct the backup application what to back
up and how to restore the data. Data Protector therefore follows the requirements provided
by the writer when selecting the volumes to be backed up and the restore method.
Volume Shadow Copy service synchronizes the writers and providers. After a backup
shadow copy is created, VSS communicates this information to Data Protector. Data
Protector performs a backup from the shadow copy volume to the media and then
notifies VSS that the shadow copy can be released.
VSS restore
VSS integration restore refers to the restore of data which was backed up using the
Volume Shadow Copy service and a writer. During the restore procedure, Volume
Shadow Copy service coordinates communication between Data Protector and the
writers.
When restoring VSS-aware writers, Data Protector first restores all the relevant
metadata to identify the backup components and to determine the restore method.
It then connects to the Volume Shadow Copy service and declares that the restore is
about to begin. VSS coordinates the writers’ activities during the restore. After Data
Protector has successfully restored the data, VSS informs the writers that the restore
has been completed and the writers can access the restored data and start their
internal processing.
294
Microsoft Volume Shadow Copy service
VSS filesystem backup and restore
Some applications are not aware of the Volume Shadow Copy service. Such
applications cannot guarantee consistency of data during the creation of a shadow
copy. The VSS mechanism cannot coordinate the activities of these applications in
order to perform a consistent backup.
However, you can still benefit from the VSS functionality. The cooperation between
the backup application and a shadow copy provider can be still used to assure a
higher level of data consistency. Microsoft calls this state of data consistency “crash
consistent data”. This means that the VSS mechanism commits all pending I/O
operations and holds incoming writing requests during the preparation of a shadow
copy volume. In this way, all files on the filesystem are closed and unlocked when
the shadow copy is being created.
Microsoft Volume Shadow Copy functionality allows the creation of a volume shadow
copy without the participation of the applications being backed up. In this case, the
shadow copy volume is created and then backed up by Data Protector. This approach
can be used with applications that are not aware of the VSS mechanism.
IMPORTANT:
When applications that are not aware of the VSS mechanism are being backed up, data
consistency from the applications’ point of view cannot be guaranteed. Data consistency
is the same as in the event of a power failure. Data Protector cannot guarantee any data
consistency when applications are not actively participating in the creation of a shadow
copy.
The consistency of data in a VSS filesystem backup is improved in comparison to a
non-VSS filesystem backup. VSS allows you to create shadow copy backups of
volumes and exact point-in-time copies of files, including all open files. For example,
databases that are held open exclusively and files that are open due to operator or
system activity are backed up during a VSS filesystem backup. In this way, files that
have changed during the backup procedure are copied correctly.
The advantages of VSS filesystem backup are as follows:
• A computer can be backed up while applications and services are running.
Therefore, applications can continue to write data to the volume during a backup.
• Files that are open are no longer skipped during the backup process because
they appear closed on the shadow copy volume at the time of the creation of the
shadow copy.
Concepts guide
295
• Backups can be performed at any time without locking out users.
• There is little or no impact on the performance of the application system during
the backup process.
Backup and restore
VSS backup is implemented as an additional Windows filesystem backup on Windows
Server 2003. The level of data integrity is slightly improved in comparison to a
traditional backup of active volume. For detailed information on Windows filesystem
backup and restore, see the online Help.
During a VSS filesystem backup, applications cannot effectively contribute to data
consistency because they are not aware of the VSS mechanism. However, Data
Protector and a provider can still cooperate in creating volume shadow copies. VSS
filesystem backup offers the option of backing up data as it appears at a certain
point-in-time, regardless of system I/O activity during the backup.
When Data Protector requests a backup of the volumes specified in the backup
specification, the VSS mechanism commits all pending I/O operations, holds incoming
writing requests, and prepares a shadow copy volume.
When the shadow copy is created, Data Protector starts its normal backup procedure,
except that the source volume is replaced by the newly created shadow copy. If
shadow copy creation fails, Data Protector will proceed with a normal filesystem
backup if such behavior was specified in the backup specification.
A computer is backed up while files are open and services are running. Files are not
skipped during such a backup. VSS allows services and applications to continue
running uninterrupted on the actual volumes while a shadow copy is being made.
After the backup is completed, the shadow copy is deleted.
The restore of data backed up using the VSS filesystem backup does not differ from
the standard restore procedure.
296
Microsoft Volume Shadow Copy service
A Backup scenarios
In this appendix
This Appendix describes two scenarios: one for company XYZ and one for company
ABC. Both companies plan to enhance their data storage systems. Their current
backup solutions are described along with the inherent problems. Solutions are then
proposed to alleviate the problems and to meet the future data storage needs of both
companies.
Considerations
In both cases, the following considerations must be taken into account when
formulating a company’s backup strategy:
• How critical system availability (and backup) is to the company
• The need to keep the backed up data at a remote location in case of disaster.
• The level of business continuance. This includes the recovery and restore plan
for all critical systems.
• The security of backed up data
The need to guard premises to prevent unauthorized people from entering.
This also includes safeguarding all relevant data against unauthorized access
with physical access prevention and electronic password protection.
• The type of data that needs to be backed up
The company’s data can be divided into categories like company business data,
company resource data, project data, and personal data, each with its own
specific requirements.
• Performance aspects for backups and restores
• Network and system topology
Determine which systems can use what network links and what transfer rates
are possible.
• Time window
Concepts guide
297
Define the periods of time during which backups of specific systems can be
done.
• Local versus network backups
Determine which systems, that the backup devices are connected to, are
backed up locally and which are backed up over the network.
• Backup policy implementation
• How backups are done, and which backup options are used
This defines the frequency of full and incremental backups. It also defines the
backup options that are used, and whether the backups are permanently
protected with the backup media stored at a remote site.
• How the systems are grouped into backup specifications
Consider how best to group backup specifications. This can be on the basis
of departments, data types, or backup frequency.
• How the backups are scheduled
Consider using the staggered approach, whereby full backups are scheduled
for different clients (backup specifications) on different days to avoid network
load, device load, and time window issues.
• Retaining data on media and information about backups
Consider protecting data from being overwritten by newer backups for a
specified amount of time.
Define the period of time that the Data Protector Catalog Database should
store information about backups.
• Device configuration
Determine the devices to be used for backups and the systems they are connected
to. Connect the backup devices to systems with the greatest amount of data so
that as much data as possible is backed up locally and not through the network.
This increases backup speed.
If you have large amounts of data to back up, consider using a library device
• Media management
Determine the type of media to be used, how to group the media into media
pools, and how to position objects on the media.
• Vaulting
Decide whether to store media to a safe place, where it is kept for a specific
period of time.
• Backup administrators and operators
298
Backup scenarios
Determine the administration and operations rights for the backup systems users.
Company XYZ
XYZ is a translation agency providing the following services:
•
•
•
•
•
Translation, localization, language editing, and proof-reading
Certification of translated documents
Simultaneous and consecutive interpretation
Desktop publishing and graphic design
Rental of conference interpreting equipment
XYZ is currently growing at 20-25 percent per year. Their current backup solution is
not able to keep pace with this growth. The backup process is very labor intensive
because of the manual process in handling backup tapes.
Environment
This section describes the present-day hardware and software environment of XYZ
and how the data storage policy is implemented.
XYZ is divided into three departments, which are connected to a Corporate Network
backbone:
• English Department
• Other Languages Department
• Admin Department
The hardware and software environment of XYZ is depicted in Table 15 on page 299
and the current backup topology in Figure 91 on page 300.
Table 15 Hardware and software environment of XYZ
Depart.
#Servers
#Clients
Current
data
Projected
data (in 5
Years)
Current devices
English
1 Windows
2000
15
Windows
35 GB
107 GB
3 HP
StorageWorks
DAT24
autoloaders
Concepts guide
299
Depart.
#Servers
#Clients
Current
data
Projected
data (in 5
Years)
Current devices
Other
Languages
1 AIX
11 UX
22 GB
67 GB
2 HP
StorageWorks
DAT24
autoloaders
Admin
1 HP-UX
5 UX
10 GB
31 GB
1 HP
StorageWorks
DAT24 autoloader
Figure 91 on page 300 shows how the XYZ backup environment is organized.
Figure 91 Current XYZ backup topology
XYZ currently has three servers with an estimated total data volume of 67 GB. In the
English Department, data is copied manually by each of the employees to their
respective servers at the end of each day. One of the Windows 2000 clients in this
department accounts for approximately a third of the data (12 GB).
The backup of clients in the Other Languages Department is done through a Network
File System, while the backup of clients in the Admin Department is done through
300
Backup scenarios
network shares. Employees in the Other Languages Department also work on
Saturdays.
Problems with the current solution
The current backup solution is not able to keep pace with the growth rate of XYZ.
The actual backup process is very labor intensive. The current backup process makes
it impossible to consolidate backup management or create an enterprise-wide backup
architecture. Each of the backup servers is managed individually. There is no
capability for a central backup management. The problems of the current backup
solution include the following:
• The backup solution is not automated.
• People must copy their work regularly, which creates a high potential for
errors.
• The backup utilities that are used are not the same, resulting in higher training
costs.
• The solutions used in the Other Languages and the Admin Departments are less
primitive but do have their problems. Network usage has a high impact on backup
performance. Moreover, not all data gets backed up. Only Network File System
shared files and network shared files are backed up in the Other Languages and
Admin Departments, respectively.
• Because there are three independent backup servers for the three departments,
there is no central control or administration of the following key areas:
• Device configuration
• Media management
• Backup configuration
• Scheduling
• Monitoring
• Restore operations
• Because each of the backup servers is managed individually, there is no central
reporting.
• The current solution does not offer disaster recovery capabilities. This is an
increasingly important setback. A disaster may result in the company losing a
significant part of its business.
Concepts guide
301
Backup strategy requirements
Requirements
After addressing the items under “Considerations” on page 297, the following
requirements have been identified for the backup solution of company XYZ:
• Backup Policy
• Full, weekly backups will occur and be completed within 12 hours.
• Daily incremental backups will occur at the end of each workday and will be
completed within 8 hours.
• A permanent data protection period will be included.
• Backup media will be stored at a remote site.
• Backup
All backup operations must require less manual intervention than currently.
• Restore
• Convenient and fast restore of recent data must be provided. Data to be
restored must be browsable for the first 3 weeks after backup.
• Restores of backups of data in the vault must be possible within two days.
• Network Connectivity
The backup servers and the departments will be connected to a 100TX Ethernet
LAN.
• Planned Growth
Growth in the current data capacities is projected at 20 to 25% per year in the
next five years.
• Software
The backup servers need to be running on one of the supported operating systems.
For information on supported operating systems for the Cell Manager, see the
HP Data Protector product announcements, software notes, and references.
• Protection Against Disaster
Upon completion of backups, the media will be stored on-site, where they will be
retrieved upon request for file restoration. After 20 days, they will be moved to
an off-site storage facility for protection in case of a disaster at the company site,
and to make space for new backups.
302
Backup scenarios
Proposed solution
Because of the limitations of the current backup solution for both performance and
enterprise-wide management, there is a need to redesign XYZ’s backup architecture
and strategy to meet its business objectives. An overview of the proposed solution is
given, followed by a detailed account of the solution. Note that this is a proposal
and not the only possible solution to XYZ’s storage management problems.
Solution overview
All clients and servers should be configured into a single Data Protector cell with the
Windows 2000 Server of the English Department as both the Cell Manager and
Installation Server for Windows systems. Use the HP-UX backup server of the Admin
Department as the Installation Server for UNIX systems. The backup devices consist
of an HP StorageWorks DLT 4115w Library, as well as two of the HP StorageWorks
DAT24 autoloaders that had been used to date. This suffices for the next five years
at the present data growth rate of 20 to 25% per year. The use of devices that have
been used to date provides an added advantage in case of disaster recovery. The
Windows 2000 client, which accounts for approximately a third of the data in the
English Department (12 GB), should be backed up locally to an HP StorageWorks
DAT24 autoloader. The proposed backup solution addresses the following key items:
•
•
•
•
•
Achieving high performing backups
Media management with minimum human effort
Simple and effective disaster recovery
Centralized backup reporting
Automation of most backup operations
All this is achieved with a single solution in combination with the proposed hardware:
Table 16 Proposed environment
Department
Current Data
Projected Data
(In 5 Years)
Devices
English*
35 GB
107 GB
HP DLT 4115
library
Other Languages
22 GB
67 GB
Admin
10 GB
31 GB
2 HP
StorageWorks
DAT24
autoloaders
Concepts guide
303
Department
Current Data
Projected Data
(In 5 Years)
Devices
* One HP StorageWorks DAT24 autoloader is currently used to locally back up the 12 GB
of data. The other HP StorageWorks DAT24 autoloader is used to back up the IDB and
configuration files. The rest of the data in this department is backed up remotely to the HP
StorageWorks DLT 4115 library.
The remaining 4 HP StorageWorks DAT24 autoloaders are used in a separate R&D
system, which is not of our configuration.
The software components proposed for the Enterprise Backup solution include HP
Data Protector A.06.10.
Proposed solution in detail
The following is a detailed account of the proposed solution:
• Cell Configuration
All clients and servers should be configured in a single Data Protector cell. The
Data Protector Cell Manager can run on the Windows 2000 Server of the English
Department.
All systems in the cell should be on the same LAN for maximum performance.
The Cell Manager should also be the Installation Server for Windows. Use the
HP-UX backup server of the Admin Department as the Installation Server for UNIX.
The HP StorageWorks DLT 4115w Library should be connected to the Cell
Manager as well as one HP StorageWorks DAT24 autoloader for backing up
the IDB and configuration files. The Windows 2000 client, which accounts for
approximately a third of the data in the English Department (12 GB) should be
backed up locally to an HP StorageWorks DAT24 autoloader.
The proposed backup environment is as depicted in Figure 92 on page 305:
304
Backup scenarios
Figure 92 Proposed XYZ backup topology
• The Cell Manager maintains the Catalog Database (CDB). This provides a
minimum of 20 days of file and directory detail on the current database.
Estimating the size of the IDB
The Internal Database Capacity Planning Tool was used to estimate the size of the
IDB in a year. The tool is located in the same directory as the rest of the Data Protector
online manuals. Input parameters shown in Figure 93 on page 306 include the number
of files in the environment (2 million), the growth factor (1.2), data protection (52
weeks), catalog protection (3 weeks), the number of full backups per week (1), and
the number of incremental backups per week (5).
Concepts guide
305
Figure 93 Input parameters
The results are shown in Figure 94 on page 306. In one year, the database is expected
to grow to approximately 419.75 MB.
Figure 94 Results
• Hardware
• Network
All systems should be on the same 100TX network for maximum performance.
This network has a sustained data transfer rate of 10 MB/s, or 36 GB/h, of
data.
• Backup Devices
306
Backup scenarios
The backup devices consist of an HP StorageWorks DLT 4115w Library as
well as two HP StorageWorks DAT24 autoloaders.
Why use the HP StorageWorks DLT 4115w Library?
The HP StorageWorks DLT 4115w Library has a single DLT4000 drive with
15 slots. It has a total compressed storage capacity of 600 GB and a maximum
sustained data transfer rate of 3 MB/s, or 10.5 GB/h, with data compression.
This is the transfer rate assumed for the remainder of this section. Currently,
the total amount of data to be backed up to the HP StorageWorks DLT 4115w
Library as a full backup, whether this is a single full backup, or the staggering
approach is used, is about 55 GB. Assuming that the size of an incremental
backup is approximately 5% of that of a full backup, a backup generation,
representing a full backup and all incremental backups based on this full
backup, requires (55+55*5%*5) GB, or 68.75 GB, of library space. In five
years time, this figure is projected to increase to about 210 GB. XYZ’s backup
policy requires that two backup generations of data be kept. Therefore, 210*2
GB, or 420 GB, of library space will be required for storage. The HP
StorageWorks DLT 4115w Library’s 600 GB storage capacity therefore
suffices.
Why use the HP StorageWorks DAT24 Autoloader?
The HP StorageWorks DAT24 autoloader has 6 24-GB data cartridges. It has
a total compressed storage capacity of 144 GB and a maximum sustained
data transfer rate of 2 MB/s, or 7 GB/h, with data compression. This is the
transfer rate assumed for the remainder of this section. Currently, the total
amount of data to be backed up to the HP StorageWorks DAT24 autoloader
connected to the aforementioned Windows 2000 client in the English
Department in a single full backup is 12 GB. Assuming that the size of an
incremental backup is approximately 5% of that of a full backup, a backup
generation, representing a full backup and all incremental backups based on
this full backup, requires (12+12*5%*5) GB, or 15 GB, of space. In five years
time, this figure is projected to increase to about 45 GB. XYZ’s backup policy
requires that two backup generations of data be kept. Therefore, 45*2 GB,
or 90 GB, of library space will be required for storage. The HP StorageWorks
DAT24 autoloader’s 144 GB storage capacity therefore suffices.
How long does a full backup last?
The Windows 2000 client in the English Department, which accounts for 12
GB of data is backed up locally to an HP StorageWorks DAT24 autoloader.
This device has a sustained data transfer rate of 2 MB/s, or approximately
7 GB/h. Therefore, a full backup of this Windows 2000 client takes about 2
hours. As the amount of data is growing at 20 to 25% per year, this client is
projected to hold about 36 GB of data in five years time. This data would
then be backed up in 6 hours.
Concepts guide
307
The Data Protector Catalog Database is approximately 0.4 GB in size. It is
backed up locally to an HP StorageWorks DAT24 autoloader, which has a
sustained data transfer rate of 2 MB/s or 7 GB/h. Data Protector by default
checks the integrity of the database before the database is backed up. It takes
less than half an hour to check the integrity of a 0.4 GB database and only
a few minutes to back up the database. Therefore, to check the integrity of,
and then back up the IDB and configuration files requires less than 1 hour.
The projected size of the database in five years time is 1.2 GB. It takes less
than an hour to check the integrity of a 1.2 GB database and less than half
an hour to back it up. Therefore, to check the integrity of, and then back up
the IDB and configuration files requires less than 2 hours.
All the other available data in the system, which is currently about 55 GB) is
backed up remotely to the HP StorageWorks DLT 4115w Library, which has
a sustained data transfer rate of 3 MB/s, or 10.5 GB/h. Most of this data is
via the 100TX network, which has a sustained data transfer rate of 10 MB/s,
or 36 GB/h, of data. This does not present a bottleneck. The backup of all
these data would therefore take about 5 to 7 hours to complete. This is well
within the allowed 12 hours. The problem would then be that in five years
time, when the data is projected to be about 170 GB, the backup would take
15 to 21 hours!
To solve this problem, use the staggering approach. Schedule the full backup
of data in the English Department for Fridays at 20:00, and that in the Other
Languages Department for Saturdays at 20:00 and that in the Admin
Department for Sundays at 20:00.
Table 17 The staggering approach
308
Mon
Tue
Wed
Thu
Fri
Sat
English
Incr1
Incr1
Incr1
Incr1
Full
Incr1
Other
Languages
Incr1
Incr1
Incr1
Incr1
Incr1
Full
Admin
Incr1
Incr1
Incr1
Incr1
Incr1
Backup scenarios
Sun
Full
Table 18 on page 309 shows the size and time requirements for these full
backups as of today, as well as the five year projection.
Table 18 Remote full backups to the HP DLT 4115 library
Department
Current Data/Backup Time
Projected Data/Backup
Time
English
23 GB / 3 h
70 GB / 7 h
Other Languages
22 GB / 3 h
67 GB / 7 h
Admin
10 GB / 1 h
31 GB / 3 h
Based on the assumption that the estimated size of an incremental backup is
5% of that of a full backup, a full backup of all data that is remotely backed
up in the largest department, the English Department, as well as incremental
backups of the other two departments is projected in five years to take
7+5%(7+3) hours, which is less than 8 hours. This is well within the allowed
12 hours.
• Media Pools
Media are grouped into media pools to provide better media tracking and control.
Group each of the two media types (DLT and DDS) in its own pool.
• Default DDS
This pool should be used for all DDS media.
• Default DLT
This pool should be used for all DLT media.
• DB_Pool
This pool should be used for the IDB and configuration files. The database
should be backed up to two media for security reasons.
• Backup Specifications
Configure five backup specifications, one for each department, and one for the
IDB and configuration files:
• ENG1_BS
Backup specification for the Windows 2000 client to be backed up locally
in the English Department. Schedule the backup specification such that Data
Protector will run a full backup every Friday and a level 1 incremental backup
every day, except Friday and Sunday at 20:00.
Concepts guide
309
Why use level 1 incremental backups?
To restore the latest data, only two media sets need to be accessed, one for
the latest full backup and one for the latest level 1 incremental backup prior
to the restore point-in-time. This simplifies and speeds up restore considerably.
• ENG2_BS
Backup specification for data in the English Department to be backed up
remotely to the HP StorageWorks DLT 4115w Library. Schedule the backup
specification such that Data Protector will run a full backup every Friday and
level 1 incremental backups every day, except Sunday at 20:00.
• OTH_BS
Backup specification for data in the Other Languages Department to be backed
up remotely to the HP StorageWorks DLT 4115w Library. Schedule the backup
specification such that Data Protector will run a full backup every Saturday at
20:00 and level 1 incremental backups every day, except Sunday at 20:00.
• ADM_BS
Backup specification for data in the Admin Department to be backed up
remotely to the HP StorageWorks DLT 4115w Library. Schedule the backup
specification such that Data Protector will run a full backup every Sunday at
20:00 and level 1 incremental backups every day, except Saturday at 20:00.
• DB_BS
Backup specification for the IDB and configuration files. Schedule the backup
specification such that Data Protector will run a full backup every day at 4:00.
At this time, other full and incremental backups would be completed and there
would be no CPU resource sharing problem between the Cell Manager and
other client systems. Two copies of the database should be made.
Backup options
• Use default Data Protector backup options. Set the following options as follows:
• Catalog Protection
Catalog protection sets the amount of time that the Data Protector Catalog
Database stores information about backed up versions, information about the
number of backed up files and directories, and messages stored in the
database. Once catalog protection expires the browsing of files and directories
using the Data Protector GUI is no longer possible. Set catalog protection to
20 days.
• Data Protection
310
Backup scenarios
Data protection determines the amount of time until each medium can be
reused. Set data protection to permanent so that data on the media is not
overwritten unintentionally.
• Concurrency
Set to 5 to allow up to five Disk Agents to concurrently write data to the HP
StorageWorks DLT 4115w Library. This will increase backup performance.
• Media Pool
For the IDB, select the DB_Pool with the appropriate media to be used for the
backup. Other objects use default media pools.
Restore options
• Use default Data Protector restore options. Set the following options as follows:
• List Restored Files
Set to ON to list the pathnames of files and directories that are restored. This
option can slow down the restore, if there are too many files to be restored.
• Display Statistical Information
Set to ON to display detailed statistical information about a specific restore
session, which includes the number of restored files and directories as well as
the amount of restored data.
• Reporting and Notifications
Email notifications will be set up for backup administrators for mount requests,
low database space, device errors, and end of session events for all backup
specifications. Optionally, email or broadcast notifications will be set up for those
end users interested in being notified about the success of backups of their systems.
To enable all users to easily determine the status of backup, set up client backup
information on the company intranet as follows:
1. Configure a report group with a Client Backup Report for each client. The
report should be logged to the file in HTML format.
2. Schedule the report group.
3. Link the logged files to the company intranet page.
• Vaulting
Vaulting is a process of storing media to a safe location for a specified period
of time.
Media will be moved to the vault once a week and replaced by new media in
the HP StorageWorks DLT 4115w Library and HP StorageWorks DAT24
Concepts guide
311
autoloaders. All actions excluding the actual moving of media to the vault are
done by the software solution, including queries done internally in the database
to prevent the administrator from having to find media that require ejection.
The second migration of media is done to move media from the vault to a security
company. This is done once a month. Data Protector provides a report on what
media need to be moved to a security company.
Track the location of media that are moved to a vault. This is important when you
want to restore from backups on media that were moved to a security company.
Data Protector allows you to perform the following vaulting tasks:
• Generate reports showing media stored at a specific location with data
protection expiring in a specified time
• Generate reports showing media used for backup within a specified time
frame
• Display a list of backup specifications that have used specified media during
the backup.
• Display a list of media needed for restore and the physical locations where
the media are stored.
• Filter media from the media view based on specific criteria, such as media
with expired protection.
• Restore
• Restore by Query
Requests for restores by query will be sent to the administrator. If the files were
last backed up less than 20 days before the request was placed, then the
administrator can use the Restore by Query restore task to select the files and
directories to be restored using a specified criteria. The administrator then
selects the Overwrite option to replace files and directories on the disk with
the versions on the media.
• Complete Filesystem Restore
Requests for the restore of whole filesystems will be sent to the administrator.
If the files were last backed up less than 20 days before the request is placed,
then the administrator can select the objects for restore and use the Restore
Into option.
With the Restore Into option selected, the object is restored with the exact
directory structure to a selected directory. Use a Windows or UNIX utility to
compare the restored object with the backed up object.
• Restore from a Vault
To restore data from a vault, which is, for example, 3 years old, send a request
to the administrator who then:
312
Backup scenarios
1.
Identifies the media needed for restore.
2. Brings the media from a vault, enters the media in the HP StorageWorks
DLT 4115w Library or other device and then scans the media.
3. Selects the specific object to be restored using the List From Media option,
if the media are not in the IDB.
4. Performs the restore.
Company ABC
ABC is a high growth software engineering company with headquarters in Cape
Town, South Africa. As a software engineering outsourcer for multinational partners,
ABC transparently sets up multi-site project teams and the accompanying infrastructure
to seamlessly execute a wide array of software engineering projects. ABC has been
growing at a rate of 30-40% per year. The growth rate is expected to slow down to
15 to 20% in the next five years.
Environment
This section describes the present-day hardware and software environment of ABC
and how the data storage policy is implemented.
ABC has offices at three locations. The main hardware data at the three locations is
given in Table 19 on page 313.
Table 19 Size of backup environment
Location
#Win
servers
#Win
clients
#UX
servers
#UX
clients
Current
data
Data (in
5 Years)
Current
devices
ABC Cape
Town
7
55
11
40
100
250
5 DAT24*
ABC
Pretoria
5
39
5
32
22
55
1 DAT24*
ABC
Durban
3
21
6
59
16
40
1 DAT24*
* HP StorageWorks DAT24 autoloader
Concepts guide
313
Three departments at ABC Cape Town use the Microsoft SQL database to store their
data and the company uses Microsoft Exchange Server for mailing services. These
databases, currently containing 11 GB and 15 GB of data, respectively, are backed
up to 2 HP StorageWorks DAT24 autoloaders.
The system architecture of ABC Cape Town includes the SAP R/3 system using Oracle
databases. Three HP T600 servers are used as SAP database servers. ABC Cape
Town uses K260 SAP application servers that are configured into application groups,
i.e. Sales and Distribution, Finance, and Production. The application servers are not
highly available. The current backup environment of ABC Cape Town is depicted in
Figure 95 on page 314.
Figure 95 Current ABC Cape Town backup topology
Currently, backups of the SAP database servers at ABC Cape Town are performed
using the SAP BRBACKUP and BRARCHIVE utilities to 3 HP StorageWorks DAT24
autoloaders. Data is copied manually by employees to their respective servers on a
daily basis. The Microsoft Exchange Server and Microsoft SQL database are backed
up separately each to an HP StorageWorks DAT24 autoloader by the backup
administrator.
The same system is used at both ABC Durban and ABC Pretoria, with the difference
that no SAP system is in place at these sites. Employees copy their data to their
respective servers. Data is backed up to an HP StorageWorks DAT24 autoloader on
a daily basis.
Two of the servers at ABC Pretoria have more than 500 000 files each.
314
Backup scenarios
Backup media are denoted by the name of the department, the name of the server
and first and last dates on which backups were performed on the media. At the end
of each quarter, media are sent for storage to a central offsite location.
Problems with current solution
The current backup solution has the following deficiencies:
•
•
•
•
•
•
•
•
There is no online backup solution of the SAP database server.
The backup solution is not centralized.
Backup operations are not fully automated.
Media management requires considerable human effort.
Disaster recovery is complex.
Backup operations last longer than the allowed time window.
The backup solution cannot keep pace with the high growth rate of ABC.
No reporting and notifications of important events pertaining to the backup.
Backup strategy requirements
Before addressing ABC’s backup strategy requirements, consider the items under
“Considerations” on page 297.
Requirements
The following section gives a description of ABC backup strategy requirements.
• Organizational policies regarding backups and restores
The company policy on archiving and storing data defines that weekly backups
be completed within 12 hours and that daily incremental backups be completed
within 8 hours.
• Maximum downtime for recovery
The allowed downtime has a significant impact on the investments into the network
infrastructure and the equipment needed for the backup. The following table lists,
for each type of data, the maximum acceptable downtime for recovery, that is,
how long specific data can be unavailable before recovered from the backup.
Table 20 Maximum acceptable downtime for recovery
Type of data
Maximum downtime
Company business data
6 hours
Concepts guide
315
Type of data
Maximum downtime
Company resource data
6 hours
Project data
1 day
Personal data
2 days
This recovery time mainly consists of the time needed to access the media and
the time required to actually restore data to a disk.
• How long specific types of data should be kept
Table 21 on page 316 shows how long data should be kept. This has implications
on the amount of backup media required.
Table 21 How long data should be kept
Type of data
Max data storage time
Company business data
5 years
Company resource data
5 years
Project data
5 years
Personal data
3 months
• How media with backed up data should be stored and maintained
Media should be kept in the tape library in the computer room. All data included
in the company backup system should be archived in full every week and
incrementally every day. The data should be stored at a security company.
• Amount of data that needs to be backed up
The amount of data that currently needs to be backed up is shown in Table
22 on page 316:
Table 22 Amount of data to be backed up
316
Location
Data (in GB)
ABC Cape Town
100
Backup scenarios
Location
Data (in GB)
ABC Pretoria
22
ABC Durban
16
Plans for future growth of the amount of data
ABC plans to grow at 15 to 20% per year. The amount of data to be backed up
is expected to grow accordingly. This has implications not only on the amount of
time it takes to run backups and backup devices needed for backup, but also on
the size of the IDB.
Table 23 Amount of data to be backed up in five years
Location
Data (in GB)
ABC Cape Town
250
ABC Pretoria
55
ABC Durban
40
• How often data needs to be backed up
Full backups of each type of data are carried out once a week on Fridays,
Saturdays, or Sundays. Level one incremental backups are carried out daily on
week days. However, if a full backup is carried out on Friday, then the
corresponding level one incremental backups are carried out on weekdays and
then on Saturday, skipping Friday.
Proposed solution
Because of the issues presented by the current backup solution, described in Problems
with current solution on page 315, ABC is undertaking a project to redesign its data
storage system.
Solution overview
Each of the three departments at ABC Cape Town must be configured into a
Manager-of-Managers (MoM) cell. Additionally, both ABC Durban and ABC Pretoria
should be configured into MoM cells, each with two Data Protector cells.
Concepts guide
317
Configure cell A as the MoM cell for the ABC Cape Town environment, cell D as the
MoM cell for the ABC Pretoria environment, and cell F as the MoM cell for the ABC
Durban environment. This configuration is depicted in Figure 96 on page 318.
Figure 96 ABC enterprise environment
The Cell Managers and Manager-of-Managers in all the 7 cells should be Windows
systems. Use a Centralized Media Management Database (CMMDB) in one of the
cells in each MoM environment and Catalog Databases in each of the 7 cells. The
Centralized Media Management Database allows you to share libraries between
cells within each MoM environment.
Each of the three locations should have its own library. Use the HP StorageWorks
DLT 4228w Library for the ABC Cape Town environment. Use HP StorageWorks DLT
4115w Libraries for ABC Pretoria and ABC Durban.
The three cells at the ABC Cape Town MoM environment should each have one SAP
database server. The SAP database servers share the HP StorageWorks DLT 4228w
Library. The Microsoft SQL and Microsoft Exchange databases are backed up locally
to HP StorageWorks DAT24 autoloaders.
The two cells at the ABC Pretoria MoM environment should also share a Centralized
Media Management Database. This should be configured on the MoM of cell D to
enable the sharing of the HP StorageWorks DLT 4115w Library between the cells.
318
Backup scenarios
The two cells at the ABC Durban MoM environment should also share a Centralized
Media Management Database. This should be configured on the MoM of cell F to
enable the sharing of the HP StorageWorks DLT 4115w Library between the cells.
The following is a detailed account of the proposed solution:
Proposed solution in detail
• Cell Configuration
Configure the departments into 7 cells, of which three are at ABC Cape Town,
and two each at ABC Pretoria and ABC Durban.
Why configure into seven cells?
• Because ABC’s departments are geographically dispersed, it would be difficult
to manage them from a single cell. Moreover, there may be networking
problems between the systems. The configuration also coincides with number
of departments, which is an important aspect in terms of security. Each of the
cells is also of the recommended size of 30 to 50 client systems. Note, however
that this number depends among other things on the number of files and
directories in individual client systems.
Then configure each of the three locations as a Manager-of-Managers environment.
The MoM allows you to efficiently, transparently and centrally manage your cells
from a single point. This then enables you to configure the Centralized Media
Management Database (CMMDB) in each MoM environment.
Why use the CMMDB?
• The Centralized Media Management Database (CMMDB) enables all cells in
a MoM environment to share devices and media. Each of the three MoM
environments at ABC can then use a single library, shared by client systems
in all cells in the environment. Using only one very large library for all ABC’s
data would not make much sense, because it would require that huge amounts
of data be transferred over WAN for backup purposes.
Use a Catalog Database in each of the 7 cells. The systems in the cells would be
as depicted in Table 24 on page 319:
Table 24 ABC cell configuration
MoM
environment
Cell
#Windows
servers
#Windows
clients
#UNIX
servers
#UNIX
clients
#SAP
ABC Cape
Town
A*
3
24
2
7
1
Concepts guide
319
MoM
environment
ABC Pretoria
ABC Durban
Cell
#Windows
servers
#Windows
clients
#UNIX
servers
#UNIX
clients
#SAP
B
2
11
5
21
1
C
2
20
4
12
1
D*
4
33
E
1
6
5
32
F*
2
10
4
30
p
1
11
2
29
#SAP is the number of SAP database servers
* represents a MoM cell
The Cell Managers and Manager-of-Managers in all the 7 cells should be
Windows systems.
Why choose the Windows system?
• Windows systems provide the native Unicode support and therefore require
less configuration to properly handle international character in file names.
Configure cell A as the Manager-of-Managers cell of the ABC Cape Town
environment and import the rest of the cells into the MoM environment. Configure
a Centralized Media Management Database in MoM cell A to allow you to share
the same library with cells B and C. Share the HP StorageWorks DLT 4228w
Library for the ABC Cape Town environment. With a capacity of 1,1 TB in
compressed format, this library should suffice for the company’s projected needs
in the next five years.
The three cells at ABC Cape Town should each have one SAP database server.
The SAP database servers share the HP StorageWorks DLT 4228w Library. The
Microsoft SQL and Microsoft Exchange databases are backed up locally to
existing HP StorageWorks DAT24 autoloaders. Each of the cells in the environment
should have its own Catalog Database. The configuration of the Cape Town
environment is depicted in Figure 97 on page 321.
320
Backup scenarios
Figure 97 ABC Cape Town enterprise backup environment
The two cells at the ABC Pretoria MoM environment should share a Centralized
Media Management Database. This should be configured on the MoM of cell D.
The purpose of using the CMMDB is to enable the sharing of the HP StorageWorks
DLT 4115w Library between the cells. Each of the cells in the environment should
have its own Catalog Database.
The two cells at the ABC Durban MoM environment should, likewise, share a
Centralized Media Management Database. This should be configured on the
MoM of cell F. Each of the cells in the environment should also have its own
Catalog Database.
Use an HP StorageWorks DLT 4115w Library for the ABC Pretoria environment
and for the ABC Durban environment. With a capacity of 600 GB in compressed
format, this library should suffice for the company’s projected needs in the next
five years in each of these environments.
Estimating the size of the IDB
The Internal Database Capacity Planning Tool was used to estimate the size of
the IDB in cell F in a year. This tool is located at:
• On the HP-UX and Solaris Cell Managers:
/opt/omni/doc/C/IDB_capacity_planning.xls
• On the Windows Cell Manager:
Data_Protector_home\docs\IDB_capacity_planning.xls
Input parameters shown in Figure 98 on page 322 include the number of files in
the environment (2 million), the growth factor (1.2), data protection (260 weeks),
Concepts guide
321
catalog protection (3 weeks), number of full backups per week (1), and number
of incremental backups per week (5).
Figure 98 Input parameters
The results are shown in Figure 99 on page 322. In one year, the database is
expected to grow to approximately 667.47 MB.
Figure 99 Results
You can also use the Internal Database Capacity Planning Tool to estimate the
size of the IDB in environments with online databases (Oracle, SAP R/3).
• Hardware
• Network
322
Backup scenarios
All systems in the same location should be on the same LAN for maximum
performance. Use the 100TX network to connect all the systems in each of the
locations and the WAN to connect the cells in the three locations. The 100TX
network has a sustained data transfer rate of 10 MB/s, or 36 GB/h, of data.
• Backup Devices
The backup devices consist of an HP StorageWorks DLT 4228w Library for
ABC Cape Town and two HP StorageWorks DLT 4115w Libraries for ABC
Pretoria and ABC Durban as well as 7 HP StorageWorks DAT24 autoloaders
for backing up the IDB and configuration files in all the cells and 2 HP
StorageWorks DAT24 autoloaders for backing up the Microsoft SQL database
and the Microsoft Exchange database at ABC Cape Town. The Microsoft
Exchange Server and the Microsoft SQL Server currently consist of 15 GB and
11 GB of data, respectively, while the rest of the data (100 GB -15 GB - 11
GB = 74 GB) is backed up using the three SAP database servers.
Why use the HP StorageWorks DLT 4228w Library?
• The HP StorageWorks DLT 4228w Library has two DLT4000 drive with
28 slots. It has a total compressed storage capacity of 1.1 TB and a
maximum sustained data transfer rate of 6 MB/s (2 x 3 MB/s), or 21
GB/h, with data compression. This is the transfer rate assumed for the
remainder of this section. Currently, the total amount of data to be backed
up to the HP StorageWorks DLT 4228w Library as a full backup, whether
this is a single full backup, or the staggering approach is used, is about
74 GB. Assuming that the size of an incremental backup is approximately
5% of that of a full backup, a backup generation, representing a full backup
and all incremental backups based on this full backup, requires
(74+74*5%*5) GB, or 92.5 GB, of library space. In five years time, this
figure is projected to increase to about 230 GB. ABC’s backup policy
requires that three backup generations of data be kept. Therefore, 230*3
GB, or 690 GB, of library space will be required for storage. The HP
StorageWorks DLT 4228w Library’s 1.1 TB GB storage capacity therefore
suffices.
The library at ABC Cape Town is shared among the three cells at the location.
The library at the ABC Pretoria environment is shared between cells D and E,
while that at ABC Durban is shared between cells F and G. Such a
configuration requires the use of the Data Protector Centralized Media
Management Database in each of the three MoM environments. These
databases are configured on the Manager-of-Managers of cells A, D, and F.
Concepts guide
323
Why use the HP StorageWorks DLT 4115w Library?
• The HP StorageWorks DLT 4115w Library has a single DLT4000 drive
with 15 slots. It has a total compressed storage capacity of 600 GB and
a maximum sustained data transfer rate of 3 MB/s, or 10.5 GB/h, with
data compression. This is the transfer rate assumed for the remainder of
this section. Currently, the total amount of data to be backed up at ABC
Pretoria to the HP StorageWorks DLT 4115w Library as a full backup,
whether this is a single full backup, or the staggering approach is used,
is about 22 GB. Assuming that the size of an incremental backup is
approximately 5% of that of a full backup, a backup generation,
representing a full backup and all incremental backups based on this full
backup, requires (22+22*5%*5) GB, or 27.5 GB, of library space. In five
years time, this figure is projected to increase to about 68.75 GB. ABC’s
backup policy requires that three backup generations of data be kept.
Therefore, 68.75*3 GB, or 206.25 GB, of library space will be required
for storage. The HP StorageWorks DLT 4115w Library’s 600 GB storage
capacity therefore suffices.
HP StorageWorks DAT24 autoloaders are used to back up the Microsoft
Exchange Server and Microsoft SQL Server at ABC Cape Town as well as
each of the 7 Cell Managers in the 3 MoM environments.
Why use the HP StorageWorks DAT24 Autoloader?
• The HP StorageWorks DAT24 autoloader has 6 24-GB data cartridges.
It has a total compressed storage capacity of 144 GB and a maximum
sustained data transfer rate of 2 MB/s, or 7 GB/h, with data compression.
This is the transfer rate assumed for the remainder of this section. Currently,
the total amount of data to be backed up to the HP StorageWorks DAT24
autoloader connected to the aforementioned Microsoft Exchange Server
at ABC Cape Town is 15 GB. Assuming that the size of an incremental
backup is approximately 5% of that of a full backup, a backup generation,
representing a full backup and all incremental backups based on this full
backup, requires (15+15*5%*5) GB, or 18.75 GB, of space. In five years
time, this figure is projected to increase to about 47 GB. ABC’s backup
policy requires that two backup generations of data be kept. Therefore,
47*2 GB, or 94 GB, of library space will be required for storage. The HP
StorageWorks DAT24 autoloader’s 144 GB storage capacity therefore
suffices.
How long does a full backup last?
The SAP database servers in the three cells at ABC Cape Town contain about 74
GB of data to be backed up to an HP StorageWorks DLT 4228w Library. This
324
Backup scenarios
library has two drives and a sustained data transfer rate of 6 MB/s (2 x 3 MB/s),
or 21 GB/h. Therefore, data is backed up to this library in up to 5 hours. The
projected amount of data in five years, 185 GB, would be backed up in 9 to 10
hours, which would still be within the acceptable 12 hours.
Cells D and E at ABC Pretoria share an HP StorageWorks DLT 4115w Library.
This library has a single drive and a sustained data transfer rate of 3 MB/s, or
10.5 GB/h. The total amount of data to be backed up in these cells is
approximately 22 GB. This would be backed up in 2 to 3 hours. The projected
amount of data in five years, 55 GB, would be backed up in 5 to 7 hours, which
would be within the acceptable 12 hours.
Similarly, the 16 GB in cells F and G at ABC Durban would be backed up in up
to 2 hours. The projected amount of data in five years, 40 GB, would be backed
up in about 4 hours, which would be within the acceptable 12 hours.
The largest, 1.3 GB, Data Protector Catalog Database at ABC Pretoria should
be backed up in a few minutes, when no database integrity checking is performed
beforehand. Data Protector by default checks the integrity of the database before
the database is backed up. The check operation takes less than an hour for a
1.3 GB database. Therefore, the IDB and configuration files at ABC Pretoria
should then be backed up in under 2 hours.
• Media Pools
Media are grouped into media pools to provide better media tracking and control.
Media pools facilitate the management of large numbers of media, reducing the
management effort of backup administrators to a minimum. Use the organizational
structure and the systems categories criteria to define the following media pools:
Table 25 ABC’s Media Pool Usage
Media pool name
Location
Description
CT_SAP_Pool
Cape Town
SAP database server
CT_SQL_Pool
Cape Town
Microsoft SQL Server
CT_Exchange_Pool
Cape Town
Microsoft Exchange Server
CT_DB_Pool
Cape Town
IDB
P_DLT_Pool
Pretoria
HP StorageWorks DLT 4115w
Library
P_DAT_Pool
Pretoria
HP StorageWorks DAT24
autoloaders
Concepts guide
325
Media pool name
Location
Description
P_DB_Pool
Pretoria
IDB
D_DLT_Pool
Durban
HP StorageWorks DLT 4115w
Library
D_DAT_Pool
Durban
HP StorageWorks DAT24
autoloaders
D_DB_Pool
Durban
IDB
• Backup Specifications
Configure backup specifications as follows:
• DB_A...G
Backup specifications for each of the 7 IDBs and configuration files. Schedule
the backup specification such that Data Protector will run a weekly full backup
and a level one incremental every day, except Sundays at 03.00.
Why use differential (incr1) backups?
• To restore the latest data only two media sets need to be accessed, one
for the latest full backup and one for the latest level 1 incremental backup
prior to the restore point-in-time. This considerably simplifies and speeds
up the restore. Where simple incremental backups are used, the number
of media sets may increase considerably, making the restore process more
complex and slower.
Two copies of the IDB and configuration files should be made, for security
reasons.
• SAP_A...C
Backup specification for the SAP database servers in cells A, B and C,
respectively. Use the staggering approach to avoid network load, device load,
and time window issues as depicted in Table 26 on page 326:
Table 26 The Staggering Approach for ABC Cape Town
326
Mon
Tue
Wed
Thu
Fri
Sat
Cell A
Incr1
Incr1
Incr1
Incr1
Full
Incr1
Cell B
Incr1
Incr1
Incr1
Incr1
Incr1
Full
Backup scenarios
Sun
Cell C
Incr1
Incr1
Incr1
Incr1
Incr1
Full
• SERVERS_A...G
Backup specifications for the company’s servers to prepare for disaster
recovery. Each time a new server is installed, or an existing server is upgraded,
this backup specification is updated. Schedule the backup specifications such
that Data Protector will run full backups as shown in Table 27 on page 327
and level 1 incremental backups every work day.
• USERS_D...G
Backup specifications for user data. This is the main production backup at
ABC Pretoria and ABC Durban. Schedule the backup specification such that
Data Protector will run a weekly full backup as shown in Table 27 on page 327
every Friday and level 1 incremental backups every work day. However, if a
full backup is carried out on Friday, then the corresponding level one
incremental backups are carried out on weekdays and then on Saturday,
skipping Friday.
Table 27 on page 327 shows the backup specification configuration in greater
detail.
Table 27 ABC’s backup specification configuration
Name
Cell
Description
Backup day
Time
DB_A
A
IDB
Saturday
03:00
DB_B
B
IDB
Saturday
03:00
DB_C
C
IDB
Saturday
03:00
SQL_A
A
Microsoft SQL database
Friday
20:00
EXCHANGE_A
A
Microsoft Exchange
database
Friday
20:00
SAP_A
A
SAP database server
Friday
20:00
SAP_B
B
SAP database server
Saturday
20:00
SAP_C
C
SAP database server
Sunday
20:00
SERVERS_A
A
Servers
Friday
23:00
SERVERS_B
B
Servers
Saturday
23:00
Concepts guide
327
Name
Cell
Description
Backup day
Time
SERVERS_C
C
Servers
Sunday
23:00
DB_D
D
IDB
Saturday
03:00
DB_E
E
IDB
Saturday
03:00
SERVERS_D
D
Servers
Friday
23:00
SERVERS_E
E
Servers
Saturday
23:00
USERS_D
D
User data
Saturday
0:00
USERS_E
E
User data
Sunday
0:00
DB_F
F
IDB
Saturday
03:00
DB_G
G
IDB
Saturday
03:00
SERVERS_F
F
IDB
Friday
23:00
SERVERS_G
G
Servers
Saturday
23:00
USERS_F
F
User data
Saturday
0:00
USERS_G
G
User data
Sunday
0:00
Backup options
Use default Data Protector backup options. Set the following options as follows:
• Log Directories
This filesystem backup option ensures that details only on directories are stored
in the Catalog Database. This disables the search feature during restore and
allows you to browse only directories. Use this option for backing up the two
servers with more than 500 000 files each in cell D. Not using this option
would result in a large increase in the size of the Data Protector Catalog
Database.
• Protection
Data should be easily accessible for a period of three weeks. Since we will
have one weekly full backup, we set catalog protection to 27 days (3 weeks*7
days+6 days=27 days).
328
Backup scenarios
Set data protection to 5 years for all backup specifications except for
Exchange_A, which is sued to back up personal mail. Set data protection for
this backup specification to 3 months.
• Concurrency
Set to 5 to allow up to five Disk Agents to concurrently write data to the library.
This will increase backup performance.
• Media Pool
Select appropriate media pools and media to be used for backup.
• Reporting and Notifications
Email notifications will be set up for backup administrators for mount requests,
low database space, device errors, and on end of session events for all the backup
specifications. Optionally, email or broadcast notifications will be set up for those
end users interested in being notified about the success of backups of their systems.
To enable all users to easily determine the status of backup, set up client backup
information on the company home page as follows:
1. Configure a report group with a Client Backup Report for each client. The
report should be logged to the file in HTML format.
2. Schedule the report group.
3. Link the logged files to the company home page.
• Vaulting
Vaulting is a process of storing media to a safe location for a specified period
of time.
Media will be moved to the vault once a week and replaced by new media in
the HP StorageWorks DLT 4228w Library, the HP StorageWorks DLT 4115w
Library, and HP StorageWorks DAT24 autoloaders. All actions excluding the
actual moving of media to the vault are done by the software solution including
queries done internally in the database to prevent the administrator from having
to find media that require ejection.
Track the location of media that are moved to a vault. This is important when you
want to restore from backups on media that were moved to the vault. Data
Protector allows you to perform the following vaulting tasks:
• Generate reports showing media stored at a specific location with data
protection expiring in a specified time
• Generate reports showing media used for backup within a specified time
frame
• Display a list of backup specifications that have used specified media during
the backup.
Concepts guide
329
• Display a list of media needed for restore and the physical locations where
the media are stored.
• Filter media from the media view based on specific criteria, such as media
with expired protection.
• Restore
• Restore by Query
Requests for restores by query will be sent to the administrator. If the files were
last backed up less than 3 weeks before the request was placed, then the
administrator can use the Restore by Query restore task to select the files and
directories to be restored using a specified criteria. The administrator then
selects the Overwrite option to replace files and directories on the disk with
the versions on the media.
• Complete Filesystem Restore
Requests for the restore of whole filesystems will be sent to the administrator.
If the files were last backed up less than 3 weeks before the request is placed,
then the administrator can select the objects for restore and use the Restore
Into option.
With the Restore Into option selected, the object is restored with the exact
directory structure to a selected directory. Use a Windows or UNIX utility to
compare the restored object with the backed up object.
• Restore from a Vault
To restore data from a vault, which is for instance 3 years old, send a request
to the administrator who then:
1. Identifies the media needed for restore.
2. Brings the media from a vault, enters the media in the HP StorageWorks
DLT 4228w Library, the or HP StorageWorks DLT 4115w Library or other
device and then scans the media.
3. Selects the specific object to be restored using the List From Media option,
if the media are not in the Data Protector Catalog Database.
4. Performs the restore.
330
Backup scenarios
B Further information
In this appendix
This appendix provides additional information about some of the aspects of Data
Protector concepts, including backup generations, examples of automated media
copying, and internationalization.
Backup generations
Data Protector provides a time/date related protection model. It is easy to map a
generation-based backup model to the time-based model, assuming regular backups
are done.
What is a backup generation?
A backup generation, shown in Figure 100 on page 332, represents a full backup
and all incremental backups based on this full backup. When the next full backup is
done, a new backup generation is created.
Backup generations help you to know how many full versions of backed up data you
have. For a successful point-in-time restore, you need at least one backup generation
(a full backup and all incrementals to that point-in-time). Keep more than one backup
generation, three for example, depending on your company policies for data
protection.
Concepts guide
331
Figure 100 Backup generations
You configure Data Protector to automatically maintain the desired number of backup
generations by selecting the appropriate data and catalog protection durations, and
scheduling for unattended backups, both full and incremental.
For example, to keep three backup generations while you have weekly full backups
and daily leveled incremental backups, specify data protection to 7*3+6=27 days.
A backup generation represents a full backup and all incremental backups until the
next full backup: therefore, the six in the formula represents incremental backups
before the next, fourth, backup generation belonging to the third backup generation.
You can set automatic media rotation (for the media with expired protection time)
through an appropriate pool usage concept. For more information, see “Implementing
a media rotation policy” on page 143.
Examples of automated media copying
After a backup finishes, you can use the automated media copy functionality to copy
the media, and then move either the originals or the copies to an off-site vault. You
can use either post-backup or scheduled media copying, depending on the availability
of devices.
The considerations that must be taken into account are the following:
• It is recommended to perform all backups first and then copy the media.
• During media copying, the media that are being copied are unavailable for
restore.
• You can only copy the entire medium, and not specific objects.
• After the copying, the source media that are copied and the copies are marked
as non-appendable, which means that you cannot append new backups to these
media.
• With scheduled media copying, the necessary devices and media must be
available at the scheduled time, otherwise the copy operation will be aborted.
332
Further information
Example 1: automated media copying of filesystem backups
Your company has a MoM environment with two cells, each containing 150 computer
systems (servers and workstations). On average, each system has 10 GB of data,
which means that you have 3000 GB of data that you want to back up.
You want to have daily Incr1 backups of the data, weekly full backups, and monthly
full backups for archiving purposes. The backups must be performed outside the
company's working hours, which means that they can start after 5 PM and must finish
before 8 AM on the next day; they can also run during weekends.
You decide to make copies of the backup media, which will remain on site for restore
purposes, and to move the originals to an off-site vault for safety reasons. The media
should be copied after the backups finish. To do this, you will use automated media
copying.
You use an HP StorageWorks 6/60 Tape Library with 6 LTO drives, and LTO Ultrium
1 media. Based on previous experience, you assume that the data transfer rate is
about 80 GB per hour, and the average capacity of a medium is 153 GB.
After the media copy operation, the source and the target media become
non-appendable. Considering this, you may want to minimize the number of media
required for the backup. It is recommended to start with empty media and use their
maximum capacity. You can achieve this by creating backup specifications with only
one device assigned. This ensures that a new medium will be used only after the
current medium is full. However, this will increase backup time compared to writing
to several media in parallel.
You decide to create 4 backup specifications. To save media space, the data is
divided between the backup specifications in such a way that the minimum number
of media possible is used. Only one device is used for each backup.
Automated media copying is performed after the backup is completed. You can use
all the available devices for the operation. This means that 3 devices will be used
for source media, and 3 devices for target media.
It is assumed that the media copying will take approximately the same amount of
time as the backup.
Incr1 backup
Configuring backups
You schedule Incr1 backups each day from Monday to Thursday at 6 PM. The data
protection is set to 4 weeks. Supposing that 30% of the data changes daily, you
Concepts guide
333
have 900 GB of data to back up. The data is divided among backup specifications
in the following way:
•
•
•
•
BackupSpec1
BackupSpec2
BackupSpec3
BackupSpec4
(Drive
(Drive
(Drive
(Drive
1) - 300 GB
2) - 300 GB
3) - 150 GB
4) - 150 GB
BackupSpec1 and BackupSpec2 require 2 media each and the backup takes
approximately 4 hours. BackupSpec3 and BackupSpec4 require 1 medium each
and the backup takes approximately 2 hours.
Configuring automated media copying
Automated media copying of each backup starts after the backup is completed. You
have 6 media to copy, and you can use all the drives in the library for the operation,
as soon as the drives are available.
You can use post-backup media copying to copy the media used with BackupSpec1
and BackupSpec2, since two drives (Drive 5 and Drive 6) are free and therefore you
do not need to worry about availability of the devices.
You configure post-backup media copying for BackupSpec1 and select Drive 1 as
the source device and Drive 6 as the target device. You set the same data protection
as original and specify the location of the media (for example, Shelf 1).
You also configure post-backup media copying for BackupSpec2 and select Drive 2
as the source device and Drive 5 as the target device. You set the same data
protection as original and specify the location of the media.
You use scheduled media copying to copy media used in BackupSpec3 and
BackupSpec4, because you will be using Drive 3 and Drive 4 for the copy operation,
and you have to wait until both backups finish. Note that if the devices are not
available at the time the media copying is scheduled, the operation will fail. For this
reason, it is recommended to add some time to the estimated backup time when
scheduling an automated media copy operation that will use the same devices.
You schedule the media copy operation an hour after the backup is estimated to
finish, select both BackupSpec3 and BackupSpec4 to be copied, and select Drive 3
as the source device and Drive 4 as the target device. You set the same data
protection as original and specify the location of the media.
For a graphic representation of the Incr1 backup and automated media copying,
see Figure 101 on page 335.
334
Further information
Figure 101 Incr1 backup and automated media copying
Full backup
Configuring backups
You schedule your weekly full backup on Friday at 6 PM. The data protection is set
to 8 weeks. You have 3000 GB of data to back up. The data is divided among
backup specifications in the following way:
•
•
•
•
BackupSpec1
BackupSpec2
BackupSpec3
BackupSpec4
(Drive
(Drive
(Drive
(Drive
1) - 1000 GB
2) - 1000 GB
3) - 500 GB
4) - 500 GB
BackupSpec1 and BackupSpec2 require 7 media each, BackupSpec3 and
BackupSpec4 require 4 media each. The backup is completed in approximately 14
hours.
Configuring automated media copying
Automated media copying of each backup starts after the backup is completed. You
have 22 media to copy, and all the devices are used as soon as they are available.
Concepts guide
335
Again, you use post-backup media copying to copy the media used with BackupSpec1
and BackupSpec2, and scheduled media copying to copy media used in
BackupSpec3 and BackupSpec4.
The devices and the data protection settings are the same as those used for the
copying of the Incr1 backup. The scheduled media copying starts an hour after the
backup is estimated to finish.
For a graphic representation of the full backup and automated media copying, see
Figure 102 on page 337.
336
Further information
Figure 102 Full backup and automated media copying
You schedule your monthly full backup on Sunday at 6 AM. This backup is intended
for archiving purposes, so it is normally not copied.
Figure 103 on page 338 presents an overview of the time when the devices are busy.
Note that this is a rough overview, so the graph ignores the partial overlap of some
of the backup and copy sessions.
Concepts guide
337
Figure 103 Overview of backup and automated media copy sessions
Example 2: automated media copying of Oracle database backups
Your company has an Oracle database of the size of 500 GB. You want to perform
a full backup of the database daily. The backup must be performed outside the
company's working hours, which means that it can start after 5 PM and must finish
before 8 AM on the next day; it can also run during weekends.
You use automated media copying to make copies of the backup media, which will
remain on site for restore purposes. The originals will be moved to an off-site vault
for safety reasons. The media should be copied after the backup finishes. To do this,
you will use post-backup media copying.
You use an HP StorageWorks 10/700 Tape Library with 10 LTO drives, and LTO
Ultrium 1 media. Based on previous experience, you assume that the data transfer
rate is about 80 GB per hour, and the average capacity of a medium is 153 GB.
The media used for backup and media copying become non-appendable after the
media copy operation, so you may want to use as much tape space as possible. On
the other hand, you want the backup to finish as soon as possible. You use 4 devices
for the backup. It is recommended to start with empty media and use their maximum
capacity.
Automated media copying starts after the backup is completed. You have 4 media
to copy, so you use 8 devices for the operation. This means that 4 devices will be
used for source media, and 4 devices for target media.
It is assumed that the media copying will take approximately the same amount of
time as the backup.
338
Further information
Full backup
Configuring backups
You schedule your daily full backup each day from Monday to Friday at 6 PM. The
data protection is set to 4 weeks. You have 500 GB of data to back up. You use
Drive 1, Drive 2, Drive 3, and Drive 4. The backup uses 4 media and is completed
in approximately 2 hours.
Configuring automated media copying
You use post-backup media copying because you have enough devices available.
You specify Drive 1, Drive 2, Drive 3, and Drive 4 as the source devices, and Drive
5, Drive 6, Drive 7, and Drive 8 as the target devices. You set the same data
protection as original and specify the location of the media.
For a graphic representation of the full database backup and automated media
copying, see Figure 104 on page 339.
Figure 104 Full database backup and automated media copying
You schedule your monthly full backup on Saturday at 12 PM. This backup is intended
for archiving purposes, so it is normally not copied.
Figure 105 on page 340 presents an overview of the time when the devices are busy.
Concepts guide
339
Figure 105 Overview of backup and automated media copy sessions
Internationalization
Internationalization is a way to design and implement a software product so that the
product interacts with the user's native language and according to the user's locale
settings (currency, time, date, number, and other formats). It enables the user to enter
their local language text data and correctly display it. Internationalization, as a
software development methodology, enables one to implement a single-source,
single-binary software that can be localized to several languages by translating the
actual texts, which are kept separate from the binaries. Internationalization is thus a
localization-enabling process. Data Protector is an internationalized product that
provides several native languages for the user interface.
Localization
Localization is the process of adapting a product or service to a particular language
and culture. It relates to the ability to provide localized screens, online Help, error
messages, manuals, etc.
Instead of sending actual message strings, Data Protector sends string IDs from agents
to the Cell Manager. The Cell Manager then forwards the strings to the GUI, which
then displays the messages in the correct language format. Note that file names and
directory names are not indexed. They are transmitted as text strings and presented
in the GUI as such. The implications of this approach are discussed in the section,
“File name handling” on page 341.
Data Protector is localized to various languages. For more information on available
languages, see the HP Data Protector product announcements, software notes, and
references, your supplier, or the local HP sales office.
340
Further information
File name handling
Handling file names in a heterogeneous environment (different operating systems
with different local settings, all in one cell) is a significant challenge. Data Protector
handles file names under various local settings (such as language, territory, and
character sets) that were in effect on the system when the file names have been
created. File names that have been backed up using some locale settings and then
viewed or restored using different locale settings, require a specific setup to be
displayed correctly.
Background
Different platform vendors have chosen to support different sets of languages using
a variety of character set representations or character encoding standards, such as
ISO 8859-1, Shift-JIS, EUC, Code Page 932, and Unicode. These encodings conflict
with one another - two encodings can use the same value for two different characters,
or use different values for the same character. After the creation of a file name, there
is no indication which code set was used. File names passed between systems using
different encodings may not display properly in the GUI.
Passing data between different platforms is not problematic if all platforms use the
same character set or if they use an implementation of Unicode (UTF-16 on Windows
and UTF-xx on other platforms), which accommodates all characters.
Unfortunately, the UTF-xx implementation of Unicode is not yet a standard on UNIX
systems. The components of the application can be distributed on several systems
and several platforms, like Windows XP Professional, Windows 2000, HP-UX, Solaris,
and AIX. Data on all these platforms has to be backed up and restored. Data Protector
cannot compensate for the lack of a common industry-wide representation of
languages and character sets, but minimizes the impact to the user.
Example
Under certain configurations in heterogeneous environments, the file names can
appear corrupted in the GUI. For example, when using Data Protector, it is possible
to back up files on HP-UX where the Disk Agent is running and to view those files
using the Data Protector GUI running on Windows. Unless identical code sets are
used on both platforms, file names may not display properly. This is because the
same character value can have a different meaning and appearance under a different
coded character sets.
Concepts guide
341
UNIX incompatibility example
Three users working on a Solaris system without Data Protector installed, each using
a different character set, create files on the same filesystem outside the ASCII character
range. If the users then use the ls command to display the files they created as well
as those created by the other users, the following happens:
• each user views their own file names correctly
• each user views the file names of the other users as corrupted. The corrupted file
names may even look different on the different systems.
The corrupted file names were created using a different code set than the one used
to perform the ls command. They do not have a "tag" indicating the code set which
was used for their creation. This happens on systems using native filesystem viewers,
for example,ls in the terminal window.
File name handling during backup
Data Protector reads file names using the Disk Agent (running on the respective client
to be backed up) and saves an original copy to a medium. The file names are also
converted to an “internal” code set and logged to the IDB, if the log filename
option is selected for the backup.
Browsing file names
The Data Protector GUI can be used to select the files for restore. This is done by
viewing the file names in the IDB on the system where the GUI is running. Data
Protector offers multiple encodings to view all file names that appear in its GUI. When
a specific character encoding is selected, Data Protector uses it to display characters
in filenames.
To correctly display filenames, select the same character encoding that was in effect
on the system, on which the files were created. Otherwise, file names appear
corrupted in the Data Protector GUI.
The correct file names can be restored to the same platform that backup was made
on.
For a list of configurations indicating the file name browsing restrictions, see the
online Help index: “internationalization”.
File name handling during restore
Files are typically restored to the same platform as was used for backup. The process
is as follows:
342
Further information
• the files to be restored are selected in the GUI
• Data Protector searches the tape for the specified data and restores it
• the original file names (original copies from the tape) are restored
Concepts guide
343
344
Further information
Glossary
access rights
See user rights.
ACSLS
(StorageTek specific term) The Automated Cartridge System
Library Server (ACSLS) software that manages the Automated
Cartridge System (ACS).
Active Directory
(Windows specific term) The directory service in a Windows
network. It contains information about resources on the network
and makes them accessible to users and applications. The
directory services provide a consistent way to name, describe,
locate, access and manage resources regardless of the physical
system they reside on.
AES 256–bit
encryption
Data Protector software encryption, based on the AES-CTR
(Advanced Encryption Standard in Counter Mode) encryption
algorithm that uses random keys of 256-bit length. The same
key is used for both encryption and decryption. With AES
256-bit encryption, data is encrypted before it is transferred
over a network and before it is written to media.
AML
(EMASS/GRAU specific term) Automated Mixed-Media library.
application agent
A component needed on a client to back up or restore online
database integrations.
See also Disk Agent.
application system
(ZDB specific term) A system the application or database runs
on. The application or database data is located on source
volumes.
See also backup system and source volume.
archived redo log
(Oracle specific term) Also called offline redo log. If the Oracle
database operates in the ARCHIVELOG mode, as each online
redo log is filled, it is copied to an archived log destination.
This copy is the archived redo log. The presence or absence of
Concepts guide
345
an archived redo log is determined by the mode the database
is using:
• ARCHIVELOG - The filled online redo log files are archived
before they are reused. The database can be recovered if
an instance or a disk fails. A “hot” backup can be performed
only when the database is running in this mode.
• NOARCHIVELOG - The filled online redo log files are not
archived.
See also online redo log.
346
archive logging
(Lotus Domino Server specific term) Lotus Domino Server
database mode where transaction log files are overwritten only
after they have been backed up.
ASR Set
A collection of files stored on several diskettes, required for
proper reconfiguration of the replacement disk (disk partitioning
and logical volume configuration) and automatic recovery of
the original system configuration and user data that was backed
up during the full client backup. These files are stored as an
ASR archive file on the Cell Manager (in
Data_Protector_home\Config\Server\dr\asr on a
Windows Cell Manager or in
/etc/opt/omni/server/dr/asr/ on a UNIX Cell Manager)
as well as on the backup medium. The ASR archive file is
extracted to three diskettes for 32-bit Windows systems or four
diskettes for 64-bit Windows systems after a disaster occurs.
You need these diskettes to perform ASR.
Audit Logs
Data files to which auditing information is stored.
Audit Report
User-readable output of auditing information created from data
stored in audit log files.
Auditing
Information
Data about every backup session that was performed over an
extended, user-defined period for the whole Data Protector cell.
autochanger
See library.
autoloader
See library.
Automatic Storage
Management
(Oracle specific term) Automatic Storage Management is an
Oracle 10g/11g integrated filesystem and volume manager
that manages Oracle database files. It eliminates complexity
Glossary
associated with managing data and disk and provides striping
and mirroring capabilities to optimize performance.
automigration
(VLS specific term) The functionality that allows data backups
to be first made to the VLS' virtual tapes and then migrated to
physical tapes (one virtual tape emulating one physical tape)
without using an intermediate backup application.
See also Virtual Library System (VLS) and virtual tape.
BACKINT
(SAP R/3 specific term) SAP R/3 backup programs can call the
Data Protector backint interface program via an open interface,
which enables them to communicate with Data Protector
software. For backup and restore, SAP R/3 programs issue
orders for the Data Protector backint interface.
backup API
The Oracle interface between the Oracle backup/restore utility
and the backup/restore media management layer. The interface
defines a set of routines to allow the reading and writing of data
to the backup media, the creation, searching and removing the
backup files.
backup chain
See restore chain.
backup device
A device configured for use with Data Protector, which can write
data to and read data from storage media. This can be, for
example, a standalone DDS/DAT drive or a library.
backup generation
One backup generation includes one full backup and all
incremental backups until the next full backup.
backup ID
An identifier of an integration object that equals the session ID
of the backup of this object. The backup ID is preserved when
an object is copied, exported, or imported.
backup object
A backup unit that contains all items backed up from one disk
volume (logical disk or mount point). The backed up items can
be any number of files, directories, or the entire disk or mount
point. Additionally, a backup object can be a
database/application entity or a disk image (rawdisk).
A backup object is defined by:
• Client name: Hostname of the Data Protector client where
the backup object resides.
• Mount point: For filesystem objects — the access point in a
directory structure on the client where the backup object is
Concepts guide
347
located (drive on Windows and mount point on UNIX). For
integration objects — backup stream identification, indicating
the backed up database/application items.
• Description: For filesystem objects — uniquely defines objects
with identical client name and mount point. For integration
objects — displays the integration type (for example, SAP
or Lotus).
• Type: Backup object type. For filesystem objects — filesystem
type (for example, WinFS). For integration objects — “Bar”.
348
backup owner
Each backup object in the IDB has an owner. The default owner
of a backup is the user who starts the backup session.
backup session
A process that creates a copy of data on storage media. The
activities are specified in a backup specification or an interactive
session. All clients configured in one backup specification are
backed up together in one backup session using the same
backup type (full or incremental). The result of a backup session
is a set of media, which was written to, also called the backup
or media set.
See also backup specification, incremental backup, and full
backup.
backup set
A complete set of integration objects associated with a backup.
backup set
(Oracle specific term) A logical grouping of backed up files that
are created using the RMAN backup command. A backup set
is a complete set of files associated with a backup. The files can
be multiplexed to increase performance. A backup set contains
either datafiles or archived logs, but not both together.
backup
specification
A list of objects to be backed up, together with a set of devices
or drives to be used, backup options for all objects in the
specification, and days and time that you want backups to be
performed. The objects are entire disks/volumes or parts of them
such as files, directories, or even the Windows Registry for
example. File selection lists such as include-lists and exclude-lists
can be specified.
backup system
(ZDB specific term) A system connected to target volumes of one
or multiple application systems. The backup system is typically
connected to a backup device to perform the backup of the data
in a replica.
Glossary
See also application system, target volume, and replica.
backup types
See incremental backup, differential backup, transaction backup,
full backup, and delta backup.
backup to IAP
A Data Protector based backup to the HP Integrated Archiving
Platform (IAP) appliance. It takes advantage of the IAP capability
to eliminate redundancies in the stored data at a block (or chunk)
level, by creating a unique content address for each data chunk.
Only changed chunks are transmitted over the network and
added to the store.
backup view
Data Protector provides different views for backup specifications:
By Type - according to the type of data available for
backups/templates. Default view.
By Group - according to the group to which backup
specifications/templates belong.
By Name - according to the name of backup
specifications/templates.
By Manager - if you are running MoM, you can also set the
Backup view according to the Cell Manager to which backup
specifications/templates belong.
BC
(EMC Symmetrix specific term) Business Continuance are
processes that allow customers to access and manage instant
copies of EMC Symmetrix standard devices.
See also BCV.
BC
(HP StorageWorks Disk Array XP specific term) The Business
Copy XP allows to maintain internal copies of HP StorageWorks
Disk Array XP LDEVs for purposes such as data backup or data
duplication. The copies (secondary volumes or S-VOLs) can be
separated from the primary volumes (P-VOLs) and connected to
a different system for various purposes, such as backup and
development. For backup purposes, P-VOLs should be connected
to the application system, and one of the S-VOL mirror sets
should be connected to the backup system.
See also HP StorageWorks Disk Array XP LDEV, CA, Main
Control Unit, application system, and backup system.
BC EVA
(HP StorageWorks EVA specific term) Business Copy EVA is a
local replication software solution enabling you to create
point-in-time copies (replicas) of the source volumes using the
snapshot and clone capabilities of the EVA firmware.
Concepts guide
349
See also replica, source volume, snapshot, and CA+BC EVA.
350
BC Process
(EMC Symmetrix specific term) A protected storage environment
solution that has defined specially configured EMC Symmetrix
devices as mirrors or Business Continuance Volumes to protect
data on EMC Symmetrix standard devices.
See also BCV.
BC VA
(HP StorageWorks Virtual Array specific term) Business Copy
VA allows you to maintain internal copies of HP StorageWorks
Virtual Array LUNs for data backup or data duplication within
the same virtual array. The copies (child or Business Copy LUNs)
can be used for various purposes, such as backup, data analysis
or development. When used for backup purposes, the original
(parent) LUNs are connected to the application system and the
Business Copy (child) LUNs are connected to the backup system.
See also HP StorageWorks Virtual Array LUN, application
system, and backup system.
BCV
(EMC Symmetrix specific term) Business Continuance Volumes,
or BCV devices, are dedicated SLDs that are pre-configured in
the ICDA on which the business continuation operation runs.
BCV devices are assigned separate SCSI addresses, differing
from the addresses used by the SLDs they mirror. The BCV
devices are used as splittable mirrors of the primary EMC
Symmetrix SLDs that need to be protected.
See also BC and BC Process.
Boolean operators
The Boolean operators for the full text search functionality of the
online Help system are AND, OR, NOT, and NEAR. Used when
searching, they enable you to define your query precisely by
creating a relationship between search terms. If no operator is
specified in a multi-word search, AND is used by default. For
example, the query manual disaster recovery is equivalent to
manual AND disaster AND recovery.
boot volume/disk/
partition
A volume/disk/partition with files required for the initial step
of the boot process. Microsoft terminology defines the boot
volume/disk/partition as a volume/disk/partition containing
the operating system files.
BRARCHIVE
(SAP R/3 specific term) An SAP R/3 backup tool that allows
you to archive redo log files. BRARCHIVE also saves all the logs
and profiles of the archiving process.
Glossary
See also BRBACKUP, and BRRESTORE.
BRBACKUP
(SAP R/3 specific term) An SAP R/3 backup tool that allows an
online or offline backup of the control file, of individual data
files, or of all tablespaces and, if necessary, of the online redo
log files.
See also BRARCHIVE, and BRRESTORE.
BRRESTORE
(SAP R/3 specific term) An SAP R/3 tool that can be used to
restore files of the following type:
• Database data files, control files, and online redo log files
saved with BRBACKUP
• Redo log files archived with BRARCHIVE
• Non-database files saved with BRBACKUP
You can specify files, tablespaces, complete backups, log
sequence numbers of redo log files, or the session ID of the
backup.
See also BRBACKUP, and BRARCHIVE.
BSM
The Data Protector Backup Session Manager controls the backup
session. This process always runs on the Cell Manager system.
CA
(HP StorageWorks Disk Array XP specific term) Continuous
Access XP allows you to create and maintain remote copies of
HP StorageWorks Disk Array XP LDEVs for purposes such as
data duplication, backup, and disaster recovery. CA operations
involve the main (primary) disk arrays and the remote
(secondary) disk arrays. The main disk arrays contain the CA
primary volumes (P-VOLs), which contain the original data and
are connected to the application system. The remote disk arrays
contain the CA secondary volumes (S-VOLs) connected to the
backup system.
See also BC (HP StorageWorks Disk Array XP specific term),
Main Control Unit and HP StorageWorks Disk Array XP LDEV.
CA+BC EVA
(HP StorageWorks EVA specific term) The combination of
Continuous Access (CA) EVA and Business Copy (BC) EVA
enables you to create and maintain copies (replicas) of the
source volumes on a remote EVA, and then use these copies as
the source for local replication on this remote array.
See also BC EVA, replica, and source volume.
Concepts guide
351
352
CAP
(StorageTek specific term) Cartridge Access Port is a port built
into the door panel of a library. The purpose is to enter or eject
media.
catalog protection
Defines how long information about backed up data (such as
file names and file versions) is kept in the IDB.
See also data protection.
CDB
The Catalog Database is a part of the IDB that contains
information about backups, object copies, restores, media
management sessions,, and backed up data. Depending on the
selected logging level, it also contains file names and file
versions. This part of the database is always local to the cell.
See also MMDB.
CDF file
(UNIX specific term) A Context Dependent File is a file consisting
of several files grouped under the same pathname. The system
ordinarily selects one of the files using the context of the process.
This mechanism allows machine dependent executables, system
data, and device files to work correctly from all hosts in a cluster
while using the same pathname.
cell
A set of systems that are under the control of a Cell Manager.
The cell typically represents the systems on a site or an
organizational entity, which are connected to the same LAN.
Central control is available to administer the backup and restore
policies and tasks.
Cell Manager
The main system in the cell where the essential Data Protector
software is installed and from which all backup and restore
activities are managed. The GUI used for management tasks
can be located on a different system. Each cell has one Cell
Manager system.
centralized
licensing
Data Protector allows you to configure centralized licensing for
the whole enterprise environment consisting of several cells. All
Data Protector licenses are installed and kept on the Enterprise
Cell Manager system. You can then allocate licenses to specific
cells to suit your needs.
See also MoM.
Centralized Media
Management
See CMMDB.
Glossary
Database
(CMMDB)
Change Journal
(Windows specific term) A Windows filesystem feature that logs
a record of each change as it occurs to the files and directories
on a local NTFS volume.
Change Log
Provider
(Windows specific term) A module that can be queried to
determine which objects on a filesystem have been created,
modified, or deleted.
channel
(Oracle specific term) An Oracle Recovery Manager resource
allocation. Every allocated channel starts a new Oracle process,
which performs backup, restore, and recovery actions. The type
of channel allocated determines the type of media used:
• type 'disk'
• type ‘sbt_tape’
If the specified channel is of type ‘sbt_tape’ and Oracle is
integrated with Data Protector, the server process will attempt
to read backups from or write data files to Data Protector.
chunking
(IAP specific term) The process of dividing data into blocks
(chunks), where each chunk gets a unique content address. This
address is then used to determine whether a particular chunk is
already backed up to the IAP appliance. If the duplicate data
is identified (two addresses are identical, that is the address is
the same as for another data chunk already stored into IAP), it
is not backed up. This way, the data redundancy is eliminated
and the optimal data storage is achieved.
See also backup to IAP.
circular logging
(Microsoft Exchange Server and Lotus Domino Server specific
term) Circular logging is a Microsoft Exchange Server database
and Lotus Domino Server database mode, in which the
transaction log file contents are periodically overwritten after
the corresponding data is committed to the database. Circular
logging reduces disk storage space requirements.
client backup
A backup of all filesystems mounted on a client. Filesystems
mounted on the client after the backup specification was created
are not automatically detected.
Concepts guide
353
354
client backup with
disk discovery
A backup of all filesystems mounted on a client. When the
backup starts, Data Protector discovers the disks on the clients.
Client backup with disk discovery simplifies backup configuration
and improves backup coverage of systems that often mount or
dismount disks.
client or client
system
Any system configured with any Data Protector functionality and
configured in a cell.
cluster-aware
application
It is an application that supports the cluster Application
Programming Interface. Each cluster-aware application declares
its own critical resources (disk volumes (on Microsoft Cluster
Server), volume groups (on MC/ServiceGuard), application
services, IP names and addresses, and so on).
cluster continuous
replication
(Microsoft Exchange Server specific term) Cluster continuous
replication (CCR) is a high availability solution that uses cluster
management and failover options to create and maintain an
exact copy (CCR copy) of a storage group. A storage group is
replicated to a separate server. CCR removes any single point
of failure in your Exchange back-end servers. You can perform
backups using VSS on your passive Exchange Server node
where a CCR copy is located and thus reducing the load on the
active node.
A CCR copy is used for disaster recovery since you can switch
to the CCR copy in a few seconds. A replicated storage group
is represented as a new instance of Exchange writer called
Exchange Replication Service and can be backed up (using
VSS) like an ordinary storage group.
See also Exchange Replication Service and local continuous
replication.
CMD Script for
Informix Server
(Informix Server specific term) A Windows CMD script that is
created in INFORMIXDIR when an Informix Server database is
configured. The CMD script is a set of system commands that
export environment variables for Informix Server.
CMMDB
The Data Protector Centralized Media Management Database
(CMMDB) is the result of merging MMDBs from several cells in
the MoM environment. It allows you to share high-end devices
and media across multiple cells in a MoM environment. One
cell can control the robotics, serving the devices that are
connected to systems in other cells. The CMMDB must reside on
the Manager-of-Managers. A reliable network connection
Glossary
between the MoM cell and the other Data Protector cells is highly
recommended
See also MoM.
COM+ Registration
Database
(Windows specific term) The COM+ Registration Database and
the Windows Registry store COM+ application attributes, class
attributes, and computer-level attributes. This guarantees
consistency among these attributes and provides common
operation on top of these attributes.
command-line
interface (CLI)
A set of DOS and UNIX like commands that you can use in shell
scripts to perform Data Protector configuration, backup, restore,
and management tasks.
Command View
(CV) EVA
(HP StorageWorks EVA specific term) The user interface that
enables you to configure, manage, and monitor your HP
StorageWorks EVA storage system. It is used to perform various
storage management tasks, for example, creating virtual disk
families, managing storage system hardware, and creating
snapclones and snapshots of virtual disks. The Command View
EVA software runs on the HP Storage Management Appliance,
and is accessed by a Web browser.
See also HP StorageWorks EVA SMI-S Agent and HP
StorageWorks SMI-S EVA provider.
Command View
VLS
(VLS specific term) A web browser-based GUI that is used to
configure, manage, and monitor the VLS through a LAN.
See also Virtual Library System (VLS).
concurrency
See Disk Agent concurrency.
control file
(Oracle and SAP R/3 specific term) An Oracle data file that
contains entries specifying the physical structure of the database.
It provides database consistency information used for recovery.
copy set
(HP StorageWorks EVA specific term) A pair that consists of the
source volumes on a local EVA and their replica on a remote
EVA.
See also source volume, replica, and CA+BC EVA
CRS
The Cell Request Server process (service), which runs on the
Data Protector Cell Manager, and starts and controls the backup
and restore sessions. The service is started as soon as Data
Protector is installed on the Cell Manager. On Windows systems,
Concepts guide
355
the CRS runs under the account of the user specified at
installation time. On UNIX systems, it runs under the account
root.
356
CSM
The Data Protector Copy and Consolidation Session Manager
process controls the object copy and object consolidation
sessions and runs on the Cell Manager system.
data file
(Oracle and SAP R/3 specific term) A physical file created by
Oracle that contains data structures such as tables and indexes.
A data file can only belong to one Oracle database.
data protection
Defines how long the backed up data on media remains
protected, that is, Data Protector will not overwrite it. When the
protection expires, Data Protector will be able to reuse the media
in one of the next backup sessions.
See also catalog protection.
data stream
Sequence of data transferred over the communication channel.
Data_Protector_
home
On Windows Vista and Windows Server 2008, the directory
containing Data Protector program files. On other Windows
operating systems, the directory containing Data Protector
program files and data files. Its default path is
%ProgramFiles%\OmniBack, but the path can be changed
in the Data Protector Setup Wizard at installation time.
See also Data_Protector_program_data.
Data_Protector_
program_data
On Windows Vista and Windows Server 2008, the directory
containing Data Protector data files. Its default path is
%ProgramData%\OmniBack, but the path can be changed
in the Data Protector Setup Wizard at installation time.
See also Data_Protector_home.
database library
A Data Protector set of routines that enables data transfer
between Data Protector and a server of an online database
integration, for example, Oracle Server.
database
parallelism
More than one database is backed up at a time if the number
of available devices allows you to perform backups in parallel.
Data Replication
(DR) group
(HP StorageWorks EVA specific term) A logical grouping of
EVA virtual disks. It can contain up to eight copy sets provided
Glossary
they have common characteristics and share a common CA
EVA log.
See also copy set.
database server
A computer with a large database stored on it, such as the SAP
R/3 or Microsoft SQL database. A server has a database that
can be accessed by clients.
Dbobject
(Informix Server specific term) An Informix Server physical
database object. It can be a blobspace, dbspace, or logical
log file.
DC directory
The Detail Catalog (DC) directory contains DC binary files,
which store information about file versions. It represents the
DCBF part of the IDB, which occupies approximately 80% of
the IDB. The default DC directory is called the dcbf directory
and is located on the Cell Manager in the directory
Data_Protector_program_data\db40 (Windows Server
2008), Data_Protector_home\db40 (other Windows
systems), or /var/opt/omni/server/db40 (UNIX systems).
You can create more DC directories and use a custom location.
Up to 50 DC directories are supported per cell. The default
maximum size of a DC directory is 16 GB.
DCBF
The Detail Catalog Binary Files (DCBF) part of the IDB stores
information about file versions and attributes. It occupies
approximately 80% of the IDB. One DC binary file is created
for each Data Protector medium used for backup. Its maximum
size is limited by the file system settings.
delta backup
A delta backup is a backup containing all the changes made
to the database from the last backup of any type.
See also backup types.
device
A physical unit which contains either just a drive or a more
complex unit such as a library.
device chain
A device chain consists of several standalone devices configured
for sequential use. When a medium in one device gets full, the
backup automatically continues on a medium in the next device
in the device chain.
device group
(EMC Symmetrix specific term) A logical unit representing several
EMC Symmetrix devices. A device cannot belong to more than
Concepts guide
357
a single device group. All devices in a device group must be
on the same EMC Symmetrix unit. You can use a device group
to identify and work with a subset of the available EMC
Symmetrix devices.
358
device streaming
A device is streaming if it can feed enough data to the medium
to keep it moving forward continuously. Otherwise, the tape
has to be stopped, the device waits for more data, reverses the
tape a little and resumes to write to the tape, and so on. In other
words, if the data rate written to the tape is less or equal the
data rate which can be delivered to the device by the computer
system, then the device is streaming. Streaming significantly
improves the performance of the device and use of space.
DHCP server
A system running the Dynamic Host Configuration Protocol
(DHCP) providing dynamic IP address assignment and network
configuration for DHCP clients.
differential backup
An incremental backup that backs up changes made since the
last full backup. To perform this type of backup, specify the Incr1
backup type.
See also incremental backup.
differential backup
(Microsoft SQL Server specific term) A database backup that
records only the data changes made to the database after the
last full database backup.
See also backup types.
differential
database backup
A differential database backup records only those data changes
made to the database after the last full database backup.
direct backup
A SAN-based backup solution in which data movement directly
from disk to tape (or to other secondary storage) is facilitated
by the SCSI Extended Copy (Xcopy) command. Direct backup
lessens the backup I/O load on systems in a SAN environment.
The data movement is facilitated directly from disk to tape (or
to other secondary storage) by the SCSI Extended Copy (XCopy)
command. The command is provided by any element of the
infrastructure including bridges, switches, tape libraries, and
disk subsystems.
See also XCopy engine.
Glossary
directory junction
(Windows specific term) Directory junctions use the reparse
point concept of Windows. An NTFS 5 directory junction allows
you to redirect a directory/file request to another location.
disaster recovery
A process to restore a client’s main system disk to a state close
to the time when a (full) backup was performed.
Disk Agent
A component needed on a client to back it up and restore it.
The Disk Agent controls reading from and writing to a disk.
During a backup session, the Disk Agent reads data from a disk
and sends it to the Media Agent, which then moves it to the
device. During a restore session the Disk Agent receives data
from the Media Agent and writes it to the disk.
Disk Agent
concurrency
The number of Disk Agents that are allowed to send data to one
Media Agent concurrently.
disk discovery
The detection of disks during client backup with disk discovery.
During this backup, Data Protector discovers (detects) the disks
that are present on the client — even though they might not have
been present on the system when the backup was configured
— and backs them up. This is particularly useful in dynamic
environments, where configurations change rapidly. After the
disks are expanded, each inherits all options from its master
client object. Even if pre- and post-exec commands are specified
once, they are started many times, once per each object.
disk group
(Veritas Volume Manager specific term) The basic unit of data
storage in VxVM system. A disk group can consist of one or
more physical volumes. There can be more than one disk group
on the system.
disk image
(rawdisk) backup
A high-speed backup where Data Protector backs up files as
bitmap images. A disk image (rawdisk) backup does not track
the files and directory structure stored on the disk, but stores a
disk image structure on byte level. You can perform a disk image
backup of either specific disk sections or a complete disk.
disk quota
A concept to manage disk space consumption for all or a subset
of users on a computer system. This concept is used by several
operating system platforms.
disk staging
The process of backing up data in several phases to improve
the performance of backups and restores, reduce costs of storing
Concepts guide
359
the backed up data, and increase the data availability and
accessibility for restore. The backup stages consist of backing
up data to one media type first (for example disk) and later
copying it to a different media type (for example tape).
360
distributed file
media format
A media format, available with the file library, which supports
a space efficient type of synthetic backup called virtual full
backup. Using this format is a prerequisite for virtual full backup.
See also virtual full backup.
Distributed File
System (DFS)
A service that connects file shares into a single namespace. The
file shares can reside on the same or on different computers.
DFS provides client access to the resources in a
location-transparent manner.
DMZ
The Demilitarized Zone (DMZ) is a network inserted as a "neutral
zone" between a company's private network (intranet) and the
outside public network (Internet). It prevents outside users from
getting direct access to company servers in the intranet.
DNS server
In the DNS client-server model, this is the server containing
information about a portion of the DNS database that makes
computer names available to client resolvers querying for name
resolution across the Internet.
domain controller
A server in a network that is responsible for user security and
verifying passwords within a group of other servers.
DR image
Data required for temporary disaster recovery operating system
(DR OS) installation and configuration.
DR OS
A disaster recovery operating system is an operating system
environment in which disaster recovery runs. It provides Data
Protector a basic runtime environment (disk, network, tape, and
filesystem access). The OS has to be installed and configured
before the Data Protector disaster recovery can be performed.
DR OS not only hosts the Data Protector disaster recovery
process but is also a part of the restored system because it
replaces its own configuration data with the original
configuration data.
drive
A physical unit that receives data from a computer system and
can write it onto a magnetic medium (typically a tape drive). It
Glossary
can also read the data from the medium and send it to the
computer system.
drive-based
encryption
Data Protector drive-based encryption uses the encryption
functionality of the drive. While performing the backup, the
drive encrypts both the data and the meta-data that is written
to the medium.
drive index
A number that identifies the mechanical position of a drive inside
a library device. This number is used by the robotic control to
access a drive.
dynamic client
See client backup with disk discovery.
EMC Symmetrix
Agent (SYMA)
(EMC Symmetrix
specific term)
See Symmetrix Agent (SYMA).
emergency boot
file
(Informix Server specific term) The Informix Server configuration
file ixbar.server_id that resides in the directory
INFORMIXDIR/etc (on Windows) or INFORMIXDIR\etc (on
UNIX). INFORMIXDIR is the Informix Server home directory
and server_id is the value of the SERVERNUM configuration
parameter. Each line of the emergency boot file corresponds to
one backup object.
enhanced
incremental
backup
Conventional incremental backup backs up files that have
changed since a previous backup, but has certain limitations in
detection of changes. Unlike conventional incremental backup,
enhanced incremental backup reliably detects and backs up
also renamed and moved files, as well as files with changes in
attributes.
Enterprise Backup
Environment
Several cells can be grouped together and managed from a
central cell. The enterprise backup environment includes all
clients located in several Data Protector cells which are managed
and administered from a central cell using the
Manager-of-Managers concept.
See also MoM.
Event Log (Data
Protector Event
Log)
A central repository of all Data Protector related notifications.
By default, all notifications are sent to the Event Log. The Event
Log is accessible only to Data Protector users in the Admin group
Concepts guide
361
and to Data Protector users who are granted the Reporting
and notifications user rights. You can view or delete all
events in the Event Log.
362
Event Logs
(Windows specific term) Files in which Windows logs all events,
such as the starting or stopping of services and the logging on
and off of users. Data Protector can back up Windows Event
Logs as part of the Windows configuration backup.
Exchange
Replication Service
(Microsoft Exchange Server specific term) The Microsoft
Exchange Server service that represents storage groups that
were replicated using either Local Continuous Replication (LCR)
or Cluster Continuous Replication (CCR) technology.
See also cluster continuous replication and local continuous
replication.
exchanger
Also referred to as SCSI Exchanger.
See also library.
exporting media
A process that removes all data about backup sessions, such
as systems, objects, and file names, which reside on the media
from the IDB. Information about the media and their relation to
a pool is also removed from the IDB. The data on the media
remains unchanged.
See also importing media.
Extensible Storage
Engine (ESE)
(Microsoft Exchange Server specific term) A database technology
used as a storage system for information exchange in Microsoft
Exchange Server.
failover
Transferring of the most important cluster data, called group (on
Windows) or package (on UNIX) from one cluster node to
another. A failover can occur mostly because of software or
hardware failures or maintenance on the primary node.
failover
(HP StorageWorks EVA specific term) An operation that reverses
the roles of source and destination in CA+BC EVA
configurations.
See also CA+BC EVA.
FC bridge
See Fibre Channel bridge.
Fibre Channel
An ANSI standard for high-speed computer interconnection.
Using either optical or copper cables, it allows the high speed
Glossary
bi-directional transmission of large data files and can be
deployed between sites kilometers apart. Fibre Channel connects
nodes using three physical topologies: point-to-point, loop, and
switched.
Fibre Channel
bridge
A Fibre Channel bridge or multiplexer provides the ability to
migrate existing parallel SCSI devices, like RAID arrays, solid
state disks (SSD), and tape libraries to a Fibre Channel
environment. On one side of the bridge or multiplexer there is
a Fibre Channel interface while on the other side there are
parallel SCSI ports. The bridge or multiplexer enables SCSI
packets to be moved between the Fibre Channel and parallel
SCSI devices.
file depot
A file containing the data from a backup to a file library device.
file jukebox device
A device residing on disk consisting of multiple slots used to
store file media.
file library device
A device which resides on a disk emulating a library with several
media, hence containing multiple files, referred to as file depots.
File Replication
Service (FRS)
A Windows service that replicates the domain controller store
logon scripts and group policies. FRS also enables replication
of Distributed File System (DFS) shares between systems and
allows any server to perform replication activity.
file tree walk
(Windows specific term) The process of traversing a filesystem
to determine which objects have been created, modified, or
deleted.
file version
The same file can be backed up multiple times in case of full
backups and incremental backups (if the file changed). If the
log level ALL is selected for backup, Data Protector retains one
entry in the IDB for the filename itself and one for each version
(date/time) of the file.
filesystem
The organization of files on a hard disk. A filesystem is backed
up so that the file attributes and the file contents are stored on
the backup media.
first-level mirror
(HP StorageWorks Disk Array XP specific term) HP StorageWorks
Disk Array XP allows up to three mirror copies of a primary
volume and each of these copies can have additional two
copies. The three mirror copies are called first-level mirrors.
Concepts guide
363
See also primary volume and MU number.
364
flash recovery
area
(Oracle specific term) Flash recovery area is an Oracle 10g/11g
managed directory, filesystem, or Automatic Storage
Management disk group that serves as a centralized storage
area for files related to backup and recovery (recovery files).
See also recovery files.
fnames.dat
The fnames.dat files of the IDB contain information on the
names of the backed up files. Typically, these files occupy about
20% of the IDB, if filenames are stored.
formatting
A process that erases any data contained on a medium and
prepares it for use with Data Protector. Information about media
(medium ID, description, and location) is saved in the IDB as
well as on the respective media (media header). Data Protector
media with protected data are not formatted until the protection
expires or the media are unprotected/recycled.
free pool
An auxiliary source of media for use by media pools when they
run out of media. The media pools must be configured to use
free pools.
full backup
A backup in which all selected objects are backed up, whether
or not they have been recently modified.
See also backup types.
full database
backup
A backup of all data in a database, not only the data that has
been changed after the last (full or incremental) backup of the
database. A full database backup does not depend on any
other backup.
full mailbox
backup
A full mailbox backup is a backup of the entire mailbox content.
full ZDB
A ZDB to tape or ZDB to disk+tape session in which all selected
objects are streamed to tape, even if there are no changes from
the previous backup.
See also incremental ZDB.
global options file
A file that allows you to customize Data Protector. It explains
the global options, which cover various aspects of Data
Protector, typically time-outs and limits, and affect the entire
Data Protector cell. The file is located on the Cell Manager in
Glossary
the directory
Data_Protector_program_data\Config\Server\Options
(Windows Server 2008),
Data_Protector_home\Config\Server\Options (other
Windows systems), or /etc/opt/omni/server/options
(HP-UX or Solaris systems).
group
(Microsoft Cluster Server specific term) A collection of resources
(for example disk volumes, application services, IP names, and
addresses) that are needed to run a specific cluster-aware
applications.
GUI
A graphical user interface provided by Data Protector for easy
access to all configuration, administration, and operation tasks.
Besides the original Data Protector GUI that runs on Windows,
Data Protector also provides a Java-based graphical user
interface with the same look and feel, which runs on numerous
platforms.
hard recovery
(Microsoft Exchange Server specific term) A Microsoft Exchange
Server database recovery that is performed after a restore by
the database engine, using transaction log files.
heartbeat
A cluster data set with a time stamp carrying information about
the operational status of a particular cluster node. This data set
or packet is distributed among all cluster nodes.
Hierarchical
Storage
Management
(HSM)
A method for optimizing the use of expensive hard disk storage
by migrating less frequently used data to less expensive optical
platters. When needed, the data is migrated back to hard disk
storage. This balances the need for fast retrieval from hard disk
with the lower cost of optical platters.
Holidays file
A file that contains information about holidays. You can set
different holidays by editing the Holidays file on the Cell
Manager in the directory
Data_Protector_program_data\Config\Server\holidays
(Windows Server 2008),
Data_Protector_home\Config\Server\holidays (other
Windows systems), or /etc/opt/omni/server/Holidays
(UNIX systems).
host backup
See client backup with disk discovery.
Concepts guide
365
366
hosting system
A working Data Protector client used for Disk Delivery Disaster
Recovery with a Data Protector Disk Agent installed.
HP Operations
Manager
HP Operations Manager provides powerful capabilities for
operations management of a large number of systems and
applications in a network. Data Protector provides an integration
into this management product. This integration is implemented
as a SMART Plug-In for HP Operations Manager management
servers on Windows, HP-UX, Solaris, and Linux. Earlier versions
of HP Operations Manager were called IT/Operation,
Operations Center, Vantage Point Operations, and OpenView
Operations.
HP Operations
Manager SMART
Plug-In (SPI)
A fully integrated, out-of-the-box solution which "plugs into"
HP Operations Manager, extending the managed domain.
Through the Data Protector integration, which is implemented
as an HP Operations Manager SMART Plug-In, a user can have
an arbitrary number of Data Protector Cell Managers monitored
as an extension to HP Operations Manager.
HP StorageWorks
Disk Array XP
LDEV
A logical partition of a physical disk within an HP StorageWorks
Disk Array XP. LDEVs are entities that can be replicated in the
Continuous Access XP (CA) and Business Copy XP (BC)
configurations, or can be used as standalone entities.
See also BC, CA (HP StorageWorks Disk Array XP specific term),
and replica.
HP StorageWorks
EVA SMI-S Agent
A Data Protector software module that executes all tasks required
for the HP StorageWorks Enterprise Virtual Array integration.
With the EVA SMI-S Agent, the control over the array is
established through HP StorageWorks SMI-S EVA provider,
which directs communication between incoming requests and
CV EVA.
See also Command View (CV) EVA and HP StorageWorks SMI-S
EVA provider.
HP StorageWorks
SMI-S EVA
provider
An interface used for controlling HP StorageWorks Enterprise
Virtual Array. SMI-S EVA provider runs as a separate service
on the HP Storage Management Appliance system and acts as
a gateway between incoming requests and Command View
EVA. With the Data Protector HP StorageWorks EVA integration,
SMI-S EVA provider accepts standardized requests from the
EVA SMI-S Agent, communicates with Command View EVA for
Glossary
information or method invocation, and returns standardized
responses.
See also HP StorageWorks EVA SMI-S Agent and Command
View (CV) EVA.
HP StorageWorks
Virtual Array LUN
A logical partition of a physical disk within an HP StorageWorks
Virtual Array. LUNs are entities that can be replicated in the HP
StorageWorks Business Copy VA configuration, or can be used
as standalone entities.
See also BC VA and replica.
ICDA
(EMC Symmetrix specific term) EMC's Symmetrix Integrated
Cached Disk Arrays (ICDA) is a disk array device that combines
a set of physical disks, a number of FWD SCSI channels, an
internal cache memory, and control and diagnostic software
commonly referred to as the microcode.
IDB
The Data Protector Internal Database is an embedded database
located on the Cell Manager that keeps information regarding
which data is backed up, on which media it is backed up, how
backup and restore sessions are run, and which devices and
libraries are configured.
IDB recovery file
An IDB file (obrindex.dat) with information about IDB backups,
media, and devices used for the backup. This information can
significantly simplify IDB recovery. It is recommended to relocate
the file, together with IDB transaction logs, to a separate physical
disk from other IDB directories, and, additionally, to make an
additional copy of the file.
importing media
A process that re-reads all data about backup sessions which
are on the medium back into the IDB. This then allows for fast
and convenient access to the data on the media.
See also exporting media.
incremental
backup
A backup that selects only files that have changed since a
previous backup. Several levels of incremental backup are
available, which enables detailed control of restore chain length.
See also backup types.
incremental
backup
(Microsoft Exchange Server specific term) A backup of the
Microsoft Exchange Server data that has changed since the last
full or incremental backup. With the incremental backup, only
the transaction log files are backed up.
Concepts guide
367
See also backup types.
368
incremental
mailbox backup
An incremental mailbox backup backs up all the changes made
to the mailbox after the last backup of any type.
incremental1
mailbox backup
An incremental1 mailbox backup backs up all the changes made
to the mailbox after the last full backup.
incremental
(re)-establish
(EMC Symmetrix specific term) A BCV or SRDF control operation.
In BCV control operations, an incremental establish causes the
BCV device to be synchronized incrementally and to function
as an EMC Symmetrix mirrored medium. The EMC Symmetrix
devices must have been previously paired. In SRDF control
operations, an incremental establish causes the target (R2) device
to be synchronized incrementally and to function as an EMC
Symmetrix mirrored medium. The EMC Symmetrix devices must
have been previously paired.
incremental restore
(EMC Symmetrix specific term) A BCV or SRDF control operation.
In BCV control operations, an incremental restore reassigns a
BCV device as the next available mirror of the standard device
in the pair. However, the standard devices are updated with
only the data that was written to the BCV device during the time
of the original pair split, and the data that was written to the
standard device during the split is overwritten with data from
the BCV mirror. In SRDF control operations, an incremental
restore reassigns a target (R2) device as the next available mirror
of the source (R1) device in the pair. However, the source (R1)
devices are updated with only the data that was written to the
target (R2) device during the time of the original pair split, and
the data that was written to the source (R1) device during the
split is overwritten with data from the target (R2) mirror.
incremental ZDB
A filesystem ZDB to tape or ZDB to disk+tape session in which
only changes from the last protected full or incremental backup
are streamed to tape.
See also full ZDB.
Inet
A process that runs on each UNIX system or service that runs
on each Windows system in the Data Protector cell. It is
responsible for communication between systems in the cell and
for starting other processes needed for backup and restore. The
Inet service is started as soon as Data Protector is installed on
a system. The Inet process is started by the inetd daemon.
Glossary
Information Store
(Microsoft Exchange Server specific term) The Microsoft
Exchange Server service that is responsible for storage
management. Information Store in Microsoft Exchange Server
manages two kinds of stores: mailbox stores and public folder
stores. A mailbox store consists of mailboxes that belong to
individual users. A public folder store contains public folders
and messages that are shared among several users.
See also Key Management Service and Site Replication Service.
Informix Server
(Informix Server specific term) Refers to Informix Dynamic Server.
initializing
See formatting.
Installation Server
A computer system that holds a repository of the Data Protector
software packages for a specific architecture. The Installation
Server is used for remote installation of Data Protector clients.
In mixed environments at least two Installation Servers are
needed: one for UNIX systems and one for Windows systems.
instant recovery
(ZDB specific term) A process in which a replica, produced by
a ZDB-to-disk or a ZDB-to-disk+tape session, is used to restore
the contents of the source volumes to their states at the time at
which the replica was created, avoiding the need to perform a
restore from tape. Depending on the application or database
concerned, this may be all that is required, or other steps, such
as the application of transaction log files, may be required for
full recovery.
See also replica, zero downtime backup (ZDB), ZDB to disk,
and ZDB to disk+tape.
integration object
A backup object of a Data Protector integration, such as Oracle
or SAP DB.
Internet
Information
Services (IIS)
(Windows specific term) Microsoft Internet Information Services
is a network file and application server that supports multiple
protocols. Primarily, IIS transmits information in Hypertext
Markup Language (HTML) pages by using the Hypertext
Transport Protocol (HTTP).
IP address
An Internet Protocol address is a numeric address of a system
used to uniquely identify the system on the network. The IP
address consists of four groups of numbers separated by periods
(full stops).
Concepts guide
369
370
ISQL
(Sybase specific term) A Sybase utility used to perform system
administration tasks on Sybase SQL Server.
Java GUI Client
The Java GUI Client is a component of the Java GUI that contains
only user interface related functionalities and requires connection
to the Java GUI Server to function.
Java GUI Server
The Java GUI Server is a component of the Java GUI that is
installed on the Data Protector Cell Manager system. The Java
GUI Server receives requests from the Java GUI Client, processes
them and then sends the responses back to the Java GUI Client.
The communication is done through Hypertext Transfer Protocol
(HTTP) on port 5556.
jukebox
See library.
jukebox device
A device consisting of multiple slots used to store either optical
or file media. When being used to store file media, the jukebox
device is known as the “file jukebox device”.
keychain
A tool that eliminates the supply of a passphrase manually when
decrypting the private key. It needs to be installed and
configured on the Installation Server if you perform remote
installation using secure shell.
Key Management
Service
(Microsoft Exchange Server specific term) The Microsoft
Exchange Server service that provides encryption functionality
for enhanced security.
See also Information Store and Site Replication Service.
KMS
Key Management Server (KMS) is a centralized service that runs
on the Cell Manager and provides key management for the
Data Protector encryption functionality. The service is started as
soon as Data Protector is installed on the Cell Manager.
key store
All encryption keys are centrally stored in the key store on the
Cell Manager and administered by the Key Management Server
(KMS).
LBO
(EMC Symmetrix specific term) A Logical Backup Object (LBO)
is an object of data storage/retrieval in the EMC Symmetrix
environment. It is stored/retrieved by EMC Symmetrix as one
entity and can only be restored as a whole.
Glossary
library
Also called autochanger, jukebox, autoloader, or exchanger.
A library contains media in repository slots. Each slot holds one
medium (for example, DDS/DAT). Media are moved between
slots and drives by a robotic mechanism, allowing random
access to media. The library can contain multiple drives.
lights-out
operation or
unattended
operation
A backup or restore operation that takes place outside of normal
business hours without an operator. This implies that no operator
personnel is present to work with the backup application or
service mount requests, for example.
LISTENER.ORA
(Oracle specific term) An Oracle configuration file that describes
one or more Transparent Network Substrate (TNS) listeners on
a server.
load balancing
By default, Data Protector automatically balances the usage of
devices selected for backup, so that they are used evenly. Load
balancing optimizes the device usage by balancing the number
of objects written to each device. Since load balancing is done
automatically during backup time, you do not need to manage
how the data is actually backed up. You just specify the devices
to be used. If you do not want to use load balancing, you can
select which device will be used for each object in the backup
specification. Data Protector will access the devices in the
specified order.
local and remote
recovery
Remote recovery is performed if all Media Agent hosts specified
in the SRD file are accessible. If any of them fails, the disaster
recovery process fails over to the local mode. This means that
the target system is searched for locally attached devices. If only
one device is found, it is automatically used. Otherwise, Data
Protector prompts you to select the device, which will be used
for restore.
local continuous
replication
(Microsoft Exchange Server specific term) Local continuous
replication (LCR) is a single-server solution that creates and
maintains an exact copy (LCR copy) of a storage group. An LCR
copy is located on the same server as the original storage group.
When an LCR copy is created, it is kept up to date through
change propagation (log replay) technology. The replication
feature in LCR guarantees that logs that have not been replicated
are not deleted. The implication of this behavior is that running
backups in a mode that deletes logs may not actually free space
if replication is sufficiently far behind in its log copying.
Concepts guide
371
An LCR copy is used for disaster recovery because you can
switch to the LCR copy in a few seconds. If an LCR copy is used
for backup and if it is located on a different disk than the original
data, then the I/O load on a production database is minimal.
A replicated storage group is represented as a new instance of
Exchange writer called Exchange Replication Service and can
be backed up (using VSS) as a normal storage group.
See also cluster continuous replication and Exchange Replication
Service.
372
lock name
You can configure the same physical device several times with
different characteristics, by using different device names. The
lock name is a user specified string that is used for locking all
such device configurations to prevent collision if several such
devices (device names) are used concurrently. Use an identical
lock name for all device definitions which use the same physical
device.
log_full shell script
(Informix Server UNIX specific term) A script provided by ON-Bar
that you can use to start backing up logical log files when
Informix Server issues a logfull event alarm. The Informix Server
ALARMPROGRAM configuration parameter defaults to the
INFORMIXDIR/etc/log_full.sh, where INFORMIXDIR
is the Informix Server home directory. If you do not want logical
logs to be backed up continuously, set the ALARMPROGRAM
configuration parameter to INFORMIXDIR/etc/no_log.sh.
logging level
The logging level determines the amount of details on files and
directories written to the IDB during backup, object copying, or
object consolidation. You can always restore your data,
regardless of the logging level used during backup. Data
Protector provides four logging levels: Log All, Log Directories,
Log Files, and No Log. The different logging level settings
influence the IDB growth, backup speed, and the convenience
of browsing data for restore.
logical-log files
This applies to online database backup. Logical-log files are
files in which modified data is first stored before being flushed
to disk. In the event of a failure, these logical-log files are used
to roll forward all transactions that have been committed as well
as roll back any transactions that have not been committed.
Glossary
login ID
(Microsoft SQL Server specific term) The name a user uses to
log on to Microsoft SQL Server. A login ID is valid if Microsoft
SQL Server has an entry for that user in the system table syslogin.
login information
to the Oracle
Target Database
(Oracle and SAP R/3 specific term) The format of the login
information is user_name/password@service, where:
login information
to the Recovery
Catalog Database
(Oracle specific term) The format of the login information to the
Recovery (Oracle) Catalog Database is
user_name/password@service, where the description of
the user name, password, and service name is the same as in
the Oracle SQL*Net V2 login information to the Oracle target
database. In this case, service is the name of the service to
the Recovery Catalog Database, not the Oracle target database.
Note that the Oracle user specified here must be the owner of
the Oracle Recovery Catalog.
Lotus C API
(Lotus Domino Server specific term) An interface for the exchange
of backup and recovery information between Lotus Domino
Server and a backup solution, like Data Protector.
LVM
A Logical Volume Manager is a subsystem for structuring and
mapping physical disk space to logical volumes on UNIX
systems. An LVM system consists of several volume groups,
where each volume group has several volumes.
Magic Packet
See Wake ONLAN.
mailbox
(Microsoft Exchange Server specific term) The location to which
e-mail is delivered, which is set up by the administrator for each
user. If a set of personal folders is designated as the e-mail
delivery location, e-mail is routed from the mailbox to this
location.
• user_name is the name by which a user is known to Oracle
Server and to other users. Every user name is associated
with a password and both have to be entered to connect to
an Oracle Target Database. This user must have Oracle
SYSDBA or SYSOPER rights.
• password must be the same as the password specified in
the Oracle password file (orapwd), which is used for
authentication of users performing database administration.
• service is the name used to identify an SQL*Net server
process for the target database.
Concepts guide
373
374
mailbox store
(Microsoft Exchange Server specific term) A part of the
Information Store that maintains information in user mailboxes.
A mailbox store consists of a binary rich-text .edb file and a
streaming native internet content .stm file.
Main Control Unit
(MCU)
(HP StorageWorks Disk Array XP specific term) An HP
StorageWorks XP disk array that contains the primary volumes
for the CA and BC configurations and acts as a master device.
See also BC (HP StorageWorks Disk Array XP specific term),
CA (HP StorageWorks Disk Array XP specific term), and HP
StorageWorks Disk Array XP LDEV.
Manager-ofManagers (MoM)
See MoM.
make_net_
recovery
make_net_recovery is an Ignite-UX command, which allows
you to create a recovery archive over the network onto the
Ignite-UX server or any other specified system. The target system
can be recovered across subnets after booting either from a
bootable tape created by the Ignite-UX make_boot_tape
command or the system boots directly from the Ignite-UX server.
Booting directly from the Ignite-UX server can be automated with
the Ignite-UX bootsys command or interactively specified on
the boot console.
make_tape_
recovery
make_tape_recovery is a command on Ignite-UX which
creates a bootable recovery (installation) tape, customized for
your system and enables you unattended disaster recovery by
connecting the backup device directly to the target system and
booting the target system from the bootable recovery tape. The
backup device has to be locally connected to the client during
the creation of the archive and recovery of the client.
MAPI
(Microsoft Exchange Server specific term) The MAPI (Messaging
Application Programming Interface) is the programming interface
that lets applications and messaging clients interact with
messaging and information systems.
MCU
See Main Control Unit (MCU).
Media Agent
A process that controls reading from and writing to a device,
which reads from or writes to a medium (typically a tape). During
a backup session, a Media Agent receives data from the Disk
Agent and sends it to the device for writing it to the medium.
Glossary
During a restore session, a Media Agent locates data on the
backup medium and sends it to the Disk Agent. The Disk Agent
then writes the data to the disk. A Media Agent also manages
the robotics control of a library.
media allocation
policy
Determines in which sequence media are used for backup. The
Strict allocation policy directs Data Protector to prompt for a
specific medium. The Loose policy directs Data Protector to
prompt for any suitable medium. The Formatted First policy
directs Data Protector to give preference to unknown media,
even if unprotected media are available in the library.
media condition
The quality of a medium as derived from the media condition
factors. Heavy usage and age result in an increased number of
read and write errors with tape media. Media need to be
replaced when they are marked as POOR.
media condition
factors
The user-assigned age threshold and overwrite threshold used
to determine the state of a medium.
medium ID
A unique identifier assigned to a medium by Data Protector.
media label
A user-defined identifier used to describe a medium.
media location
A user-defined physical location of a medium, such as "building
4" or "off-site storage".
media
management
session
A session performing some action on a medium, such as
initializing, scanning the content, verifying data on a medium,
or copying a medium.
media pool
A set of media of the same type (such as DDS) used and tracked
as a group. Media are formatted and assigned to a media pool.
media set
The result of a backup session is data backed up on a group of
media called media set. Depending on the media usage policy,
several sessions can share the same media.
media type
The physical type of media, such as DDS or DLT.
media usage
policy
The media usage policy controls how new backups are added
to the already used media. It can be Appendable,
Non-Appendable, or Appendable for incrementals
only.
Concepts guide
375
376
merging
This defines one mode to resolve file conflicts during restore. If
the file to be restored already exists at the destination, the one
with the more recent modification date is kept. Files not present
on the disk are always restored.
See also overwrite.
Microsoft
Exchange Server
A “client-server” messaging and a workgroup system that offers
a transparent connection to many different communication
systems. It provides users with an electronic mail system,
individual and group scheduling, online forms, and workflow
automation tools. It provides a developer with a platform on
which to build custom information-sharing and messaging-service
applications.
Microsoft
Management
Console (MMC)
(Windows specific term) An administration model for
Windows-based environments. It provides a simple, consistent,
and integrated administration user interface allowing
management of many applications through the same GUI,
provided that the applications adhere to the MMC model.
Microsoft SQL
Server
A database management system designed to meet the
requirements of distributed "client-server" computing.
Microsoft Volume
Shadow Copy
Service (VSS)
A software service that provides a unified communication
interface to coordinate backup and restore of a VSS-aware
application regardless of its specific features. This service
collaborates with the backup application, writers, shadow copy
providers, and the operating system kernel to implement the
management of volume shadow copies and shadow copy sets.
See also shadow copy, shadow copy provider, replica, and
writer.
mirror (EMC
Symmetrix and HP
StorageWorks Disk
Array XP specific
term)
See target volume.
mirror rotation (HP
StorageWorks Disk
Array XP specific
term)
See replica set rotation.
Glossary
MMD
The Media Management Daemon process (service) runs on the
Data Protector Cell Manager and controls media management
and device operations. The process is started when Data
Protector is installed on the Cell Manager.
MMDB
The Media Management Database (MMDB) is a part of the IDB
that contains information about media, media pools, devices,
libraries, library drives, and slots configured in the cell, as well
as the Data Protector media used for backup. In an enterprise
backup environment, this part of the database can be common
to all cells.
See also CMMDB, CDB.
MoM
Several cells can be grouped together and managed from a
central cell. The management system of the central cell is the
Manager-of-Managers (MoM). The cells are called MoM clients.
The MoM enables you to configure and manage multiple cells
from a central point.
mount request
A screen prompt that tells you to insert a specific medium into
a device. Once you respond to the mount request by providing
the required medium and confirm the mount request, the session
continues.
mount point
The access point in a directory structure for a disk or logical
volume, for example/opt or d:. On UNIX, the mount points
are displayed using the bdf or df command.
MSM
The Data Protector Media Session Manager, which runs on the
Cell Manager and controls media sessions, such as copying
media.
MU number
(HP StorageWorks Disk Array XP specific term) Mirror Unit
number. An integer number (0, 1 or 2), used to indicate a
first-level mirror.
See also first-level mirror.
multi-drive server
A license that allows you to run an unlimited number of Media
Agents on a single system. This license, which is bound to the
IP address of the Cell Manager, is no longer available.
obdrindex.dat
See IDB recovery file.
Concepts guide
377
378
OBDR capable
device
A device that can emulate a CD-ROM drive loaded with a
bootable disk and can thus be used as a backup or boot device
for disaster recovery purposes.
object
See backup object.
object
consolidation
The process of merging a restore chain of a backup object,
consisting of a full backup and at least one incremental backup,
into a new, consolidated version of this object. The process is
a part of the synthetic backup procedure. The result is a synthetic
full backup of the specified backup object.
object
consolidation
session
A process that merges a restore chain of a backup object,
consisting of a full backup and at least one incremental backup,
into a new, consolidated version of this object.
object copy
A copy of a specific object version that is created during an
object copy session or a backup session with object mirroring.
object copy session
A process that creates an additional copy of the backed up data
on a different media set. During an object copy session, the
selected backed up objects are copied from the source to the
target media.
object copying
The process of copying selected object versions to a specific
media set. You can select object versions from one or several
backup sessions to be copied.
object ID
(Windows specific term) The object IDs (OIDs) enable access
to NTFS 5 files no matter where in the system the files reside.
Data Protector treats the OIDs as alternate streams of the files.
object mirror
A copy of a backup object created using object mirroring.
Object mirrors are often referred to as object copies.
object mirroring
The process of writing the same data to several media sets
during a backup session. Data Protector enables you to mirror
all or some backup objects to one or more media sets.
offline backup
A backup during which an application database cannot be
used by the application.
• For simple backup methods (non ZDB), the database is
generally put into a quiescent state that allows use by the
backup system, but not the application, for the whole backup
Glossary
period (several minutes or hours). For instance, for backup
to tape, until streaming of data to the tape is finished.
• For ZDB methods, the database is also put into the quiescent
state, but for the period of the data replication process only
(several seconds). Normal database operation can then be
resumed for the rest of the backup process.
See also zero downtime backup (ZDB) and online backup.
offline recovery
Offline recovery is performed if the Cell Manager is not
accessible, for example, due to network problems. Only
standalone and SCSI library devices can be used for offline
recovery. Recovery of the Cell Manager is always offline.
offline redo log
See archived redo log.
ON-Bar
(Informix Server specific term) A backup and restore system for
Informix Server. ON-Bar enables you to create a copy of your
Informix Server data and later restore the data. The ON-Bar
backup and restore system involves the following components:
• the onbar command
• Data Protector as the backup solution
• the XBSA interface
• ON-Bar catalog tables, which are used to back up dbobjects
and track instances of dbobjects through multiple backups.
ONCONFIG
(Informix Server specific term) An environment variable that
specifies the name of the active ONCONFIG configuration file.
If the ONCONFIG environment variable is not present, Informix
Server uses the configuration values from the onconfig file in
the directory INFORMIXDIR\etc (on Windows) or
INFORMIXDIR/etc/ (on UNIX).
online backup
A backup performed while a database application remains
available for use. The database is placed into a special backup
mode of operation for the time period that the backup
application requires access to the original data objects. During
this period, the database is fully operational, but there may be
a small performance impact and log files may grow very quickly.
• For simple backup methods (non ZDB), backup mode is
required for the whole backup period (several minutes or
hours). For instance, for backup to tape, until streaming of
data to tape is finished.
Concepts guide
379
• For ZDB methods, backup mode is required for the short
period of the data replication process only (several seconds).
Normal database operation can then be resumed for the
rest of the backup process.
In some cases, transaction logs may also have to be backed up
to allow a consistent database to be restored.
See also zero downtime backup (ZDB), and offline backup.
380
online redo log
(Oracle specific term) Redo logs that have not been archived,
but are either available to the instance for recording database
activity or are filled and waiting to be archived or reused.
See also archived redo log.
OpenSSH
A set of network connectivity tools used to access remote
machines securely, by using a variety of authentication and
encryption methods. It needs to be installed and configured on
the Installation Server and the client if you perform remote
installation using secure shell.
Oracle Data Guard
(Oracle specific term) Oracle Data Guard is Oracle’s primary
disaster recovery solution. Oracle Data Guard is able to
maintain up to nine standby databases, each of which is a
real-time copy of the production (primary) database, to protect
against corruptions, data failures, human errors, and disasters.
If a failure occurs on the production (primary) database, then
a failover to one of the standby databases which becomes the
new primary database is possible. In addition, planned
downtime for maintenance can be reduced because the
production processing can be moved from the current primary
database to a standby database and back quickly.
Oracle instance
(Oracle specific term) Each installation of an Oracle database
on one or more systems. One computer system can have several
instances of a database running.
ORACLE_SID
(Oracle specific term) A unique name for an Oracle Server
instance. To switch among Oracle Servers, specify the desired
ORACLE_SID. The ORACLE_SID is included in the CONNECT
DATA parts of the connect descriptor in a TNSNAMES.ORA file
and in the definition of the TNS listener in the LISTENER.ORA
file.
original system
The system configuration backed up by Data Protector before
a computer disaster hits the system.
Glossary
overwrite
An option that defines one mode to resolve file conflicts during
restore. All files are restored from a backup even if they are
older than existing files.
See also merging.
ownership
Backup ownership affects the ability of users to see and restore
data. Each backup session and all the data backed up within
it is assigned an owner. The owner can be the user that starts
an interactive backup, the account under which the CRS process
is running, or the user specified as the owner in the backup
specification options.
If a user starts an existing backup specification without modifying
it, the backup session is not considered as interactive.
If a modified backup specification is started by a user, the user
is the owner unless the following is true:
• The user has the Switch Session Ownership user right.
• The backup session owner is explicitly defined in the backup
specification, where the username, group or domain name,
and the system name are specified.
If a backup is scheduled on a UNIX Cell Manager, the session
owner is root:sys unless the above conditions are true.
If a backup is scheduled on a Windows Cell Manager, the
session owner is the user specified during the installation, unless
the above conditions are true.
P1S file
P1S file contains information on how to format and partition all
disks installed in the system during Enhanced Automated Disaster
Recovery (EADR). It is created during full backup and is saved
on backup medium and on Cell Manager into
Data_Protector_home\Config\Se ver\dr\p1s directory
on a Windows Cell Manager or in
/etc/opt/omni/server/dr/p1s directory on a UNIX Cell
Manager with the filename recovery.p1s.
package
(MC/ServiceGuard and Veritas Cluster specific term) A collection
of resources (for example volume groups, application services,
IP names and addresses) that are needed to run a specific
cluster-aware application.
pair status
(HP StorageWorks Disk Array XP specific term) A mirrored pair
of disks can have various status values depending on the action
performed on it. The three most important status values are:
Concepts guide
381
• COPY - The mirrored pair is currently re-synchronizing. Data
is transferred from one disk to the other. The disks do not
contain the same data.
• PAIR - The mirrored pair is completely synchronized and
both disks (the primary volume and the mirrored volume)
contain identical data.
• SUSPENDED - The link between the mirrored disks is
suspended. That means that both disks are accessed and
updated independently. However, the mirror relationship is
still maintained and the pair can be re-synchronized without
transferring the complete disk.
382
parallel restore
Restoring backed up data to multiple disks at the same time (that
is, in parallel) by running multiple Disk Agents, that receive data
from one Media Agent. For the parallel restore to work, select
data that is located on different disks or logical volumes and
during backup, the data from the different objects must have
been sent to the same device using a concurrency of 2 or more.
During a parallel restore, the data for multiple objects selected
for restore is read from media at the same time, thereby
improving performance.
parallelism
The concept of reading multiple data streams from an online
database.
physical device
A physical unit that contains either a drive or a more complex
unit such as a library.
post-exec
A backup option that executes a command or script after the
backup of an object or after the entire session completes.
Post-exec commands are not supplied by Data Protector. You
need to create your own. They can be written as executables
or batch files on Windows and as shell scripts on UNIX.
See also pre-exec.
pre- and post-exec
commands
Pre- and post-exec commands are used to perform additional
action before and after a backup or restore session. They are
not supplied by Data Protector. You need to create your own
commands. They can be written as executables or batch files
on Windows and as shell scripts on UNIX.
prealloc list
A subset of media in a media pool that specifies the order in
which media are used for backup.
Glossary
pre-exec
A backup option that executes a command or script before the
backup of an object or before the entire session is started.
Pre-exec commands are not supplied by Data Protector. You
need to create your own. They can be written as executables
or batch files on Windows and as shell scripts on UNIX.
See also post-exec.
primary volume
(P-VOL)
(HP StorageWorks Disk Array XP specific term) Standard HP
StorageWorks Disk Array XP LDEVs that act as a primary volume
for the CA and BC configurations. The P-VOL is located in the
MCU.
See also secondary volume (S-VOL) and Main Control Unit
(MCU).
protection
See data protection and also catalog protection.
public folder store
(Microsoft Exchange Server specific term) The part of the
Information Store that maintains information in public folders.
A public folder store consists of a binary rich-text .edb file and
a streaming native internet content .stm file.
public/private
backed up data
When configuring a backup, you can select whether the backed
up data will be:
• public, that is visible (and accessible for restore) to all Data
Protector users
• private, that is, visible (and accessible for restore) only to
the owner of the backup and administrators
RAID
Redundant Array of Inexpensive Disks.
RAID Manager
Library
(HP StorageWorks Disk Array XP specific term) The RAID
Manager Library is used internally by Data Protector on Solaris
systems to allow access to HP StorageWorks Disk Array XP
configuration, status, and performance data and to key HP
StorageWorks Disk Array XP features through the use of function
calls translated into a sequence of low level SCSI commands.
RAID Manager XP
(HP StorageWorks Disk Array XP specific term) The RAID
Manager XP application provides an extensive list of commands
to report and control the status of the CA and BC applications.
The commands communicate through a RAID Manager instance
with the HP StorageWorks Disk Array XP Disk Control Unit. This
Concepts guide
383
instance translates the commands into a sequence of low level
SCSI commands.
384
rawdisk backup
See disk image backup.
RCU
See Remote Control Unit (RCU).
RDBMS
Relational Database Management System.
RDF1/RDF2
(EMC Symmetrix specific term) A type of SRDF device group.
Only RDF devices can be assigned to an RDF group. An RDF1
group type contains source (R1) devices and an RDF2 group
type contains target (R2) devices.
RDS
The Raima Database Server process (service) runs on the Data
Protector Cell Manager and manages the IDB. The process is
started when Data Protector is installed on the Cell Manager.
Recovery Catalog
(Oracle specific term) A set of Oracle tables and views that are
used by Recovery Manager to store information about Oracle
databases. This information is used by Recovery Manager to
manage the backup, restore, and recovery of Oracle databases.
The recovery catalog contains information about:
• The physical schema of the Oracle target database
• Data file and archived log backup sets
• Data file copies
• Archived Redo Logs
• Stored scripts
Recovery Catalog
Database
(Oracle specific term) An Oracle database that contains a
recovery catalog schema. You should not store the recovery
catalog in your target database.
recovery files
(Oracle specific term) Recovery files are Oracle 10g/11g
specific files that reside in the flash recovery area: the current
control file, online redo logs, archived redo logs, flashback logs,
control file autobackups, datafile copies, and backup pieces.
See also flash recovery area.
RecoveryInfo
When backing up Windows configuration files, Data Protector
collects the information about the current system configuration
(information on disk layout, volume, and network configuration).
This information is needed for disaster recovery.
Glossary
Recovery Manager
(RMAN)
(Oracle specific term) An Oracle command-line interface that
directs an Oracle Server process to back up, restore, or recover
the database it is connected to. RMAN uses either the recovery
catalog or the control file to store information about backups.
This information can be used later in restore sessions.
recycle
A process that removes the data protection from all backed up
data on a medium, allowing Data Protector to overwrite it during
one of the next backups. Data that belongs to the same session(s)
but resides on other media is also unprotected. Recycling does
not actually alter the data on the medium.
redo log
(Oracle specific term) Every Oracle database has a set of two
or more redo log files. The set of redo log files for the database
is known as the database's redo log. Oracle uses the redo log
to record all changes made to data.
Remote Control
Unit (RCU)
(HP StorageWorks Disk Array XP specific term) The Remote
Control Unit (RCU) acts as a slave of an MCU in a CA
configuration. In bidirectional configurations, the RCU can act
as an MCU.
Removable
Storage
Management
Database
(Windows specific term) A Windows service used for managing
removable media (such as tapes and disks) and storage devices
(libraries). Removable Storage allows applications to access
and share the same media resources.
reparse point
(Windows specific term) A system-controlled attribute that can
be associated with any directory or file. The value of a reparse
attribute can have user-controlled data. The format of the data
is understood by the application that stored the data and a
filesystem filter that was installed to interpret the data and
process such files. Whenever the filesystem encounters a file
with a reparse point, it attempts to find the filesystem filter
associated with the data format.
replica
(ZDB specific term) An image, at a particular point in time, of
the data in source volumes that contain user-specified backup
objects. Depending on the hardware or software with which it
is created, the image may be an independent exact duplicate
(clone) of the storage blocks at the physical disk level (for
example, a split mirror or snapclone), or a virtual copy (for
example, a snapshot). From perspective of a basic operating
system, the complete physical disk containing backup objects
Concepts guide
385
is replicated. However, if a volume manager is used on UNIX,
the whole volume or disk group containing a backup object
(logical volume) is replicated. If partitions are used on Windows,
the whole physical volume containing the selected partition is
replicated.
See also snapshot, snapshot creation, split mirror, and split
mirror creation.
386
replica set
(ZDB specific term) A group of replicas, all created using the
same backup specification.
See also replica and replica set rotation.
replica set rotation
(ZDB specific term) The use of a replica set for regular backup
production: Each time the same backup specification requiring
the use of a replica set is run, a new replica is created and
added to the set, until the maximum number for the set is
reached. After that, the oldest replica in the set is replaced,
maintaining the maximum number of replicas in the set.
See also replica and replica set.
restore chain
All backups that are necessary for a restore of a backup object
to a certain point in time. A restore chain consists of a full
backup of the object and any number of related incremental
backups.
restore session
A process that copies data from backup media to a client.
resync mode
(HP StorageWorks Disk Array XP VSS provider specific term)
One of two XP VSS hardware provider operation modes. When
the XP provider is in the resync mode, the source volume (P-VOL)
and its replica (S-VOL) are in the suspended mirror relationship
after a backup. The maximum number of replicas (S-VOLs per
a P-VOL) rotated is three provided that MU range is 0-2 or 0,
1, 2. Restore from a backup in such a configuration is possible
only by re-synchronization of an S-VOL with its P-VOL.
See also VSS compliant mode, source volume, primary volume
(P-VOL), replica, secondary volume (S-VOL), MU number, and
replica set rotation.
RMAN (Oracle
specific term)
See Recovery Manager.
RSM
The Data Protector Restore Session Manager controls the restore
session. This process always runs on the Cell Manager system.
Glossary
RSM
(Windows specific term) Removable Storage Manager (RSM)
includes a media management service that facilitates
communication among applications, robotic changers, and
media libraries. It enables multiple applications to share local
robotic media libraries and tape or disk drives and to manage
removable media.
scan
A function that identifies the media in a device. This synchronizes
the MMDB with the media that are actually present at the
selected locations (for example, slots in a library).
scanning
A function which identifies the media in a device. This
synchronizes the MMDB with the media that are actually present
at the selected locations (for example, slots in a library). It is
useful to perform a scan and check the actual media in the
device if someone has manually manipulated media without
using Data Protector to eject or enter, for example.
Scheduler
A function that controls when and how often automatic backups
occur. By setting up a schedule, you automate the start of
backups.
secondary volume
(S-VOL)
(HP StorageWorks Disk Array XP specific term) secondary
volumes, or S-VOLs, are XP LDEVs that act as a secondary CA
or BC mirror of another LDEV (a P-VOL). In the case of CA,
S-VOLs can be used as failover devices in a MetroCluster
configuration. The S-VOLs are assigned separate SCSI
addresses, different from the addresses used by the P-VOLs.
See also primary volume (P-VOL) and Main Control Unit (MCU)
session
See backup session,media management session, and restore
session.
session ID
An identifier of a backup, restore, object copy, object
consolidation, or media management session, consisting of the
date when the session ran and a unique number.
session key
This environment variable for the pre-exec and post-exec script
is a Data Protector unique identification of any session, including
preview sessions. The session key is not recorded in the
database, and it is used for specifying options for the omnimnt,
omnistat, and omniabort commands.
Concepts guide
387
388
shadow copy
(Microsoft VSS specific term) A volume that represents a
duplicate of the original volume at a certain point in time. The
data is then backed up from the shadow copy and not from the
original volume. The original volume continues to change as
the backup process continues, but the shadow copy of the
volume remains constant.
See also Microsoft Volume Shadow Copy Service and replica.
shadow copy
provider
(Microsoft VSS specific term) An entity that performs the work
on creating and representing the volume shadow copies.
Providers own the shadow copy data and expose the shadow
copies. Providers can be software (for example, system
providers) or hardware (local disks, disk arrays).
See also shadow copy.
shadow copy set
(Microsoft VSS specific term) A collection of shadow copies
created at the same point in time.
See also shadow copy and replica set.
shared disks
A Windows disk on another system that has been made
available to other users on the network. Systems with shared
disks can be backed up without a Data Protector Disk Agent
installed.
SIBF
The Serverless Integrations Binary Files (SIBF) is a part of the
IDB that stores raw NDMP meta data. This data is necessary to
perform restore of NDMP objects.
single instancing
(IAP specific term) The process of recognizing redundancy of
data, at both a whole object and a chunk level. It computes a
strong hash for each data chunk and uses it as a unique content
address needed to determine whether attempts to store
duplicates are being made.
See also backup to IAP.
Site Replication
Service
(Microsoft Exchange Server specific term) The Microsoft
Exchange Server 2000/2003 service that permits compatibility
with Microsoft Exchange Server 5.5 by emulating the Exchange
Server 5.5 directory service.
See also Information Store and Key Management Service.
slot
A mechanical position in a library. Each slot can hold a medium,
such as a DLT tape. Data Protector references each slot with a
Glossary
number. To read a medium, a robotic mechanism moves the
medium from a slot into the drive.
SMB
See split mirror backup.
smart copy
(VLS specific term) A copy of the backed up data created from
the virtual tape to the physical tape library. The smart copy
process allows Data Protector to distinguish between the source
and the target medium thus enabling media management.
See also Virtual Library System (VLS).
smart copy pool
(VLS specific term) A pool that defines which destination library
slots are available as smart copy targets for a specified source
virtual library.
See also Virtual Library System (VLS) and smart copy.
SMBF
The Session Messages Binary Files (SMBF) part of the IDB stores
session messages generated during backup, restore, object
copy, object consolidation, and media management sessions.
One binary file is created per session. The files are grouped by
year and month.
snapshot
(HP StorageWorks VA and HP StorageWorks EVA specific term)
A form of replica produced using snapshot creation techniques.
A range of snapshot types is available, with different
characteristics, depending on the arrays/techniques used. Such
replicas are dynamic and may be either virtual copies, still reliant
upon the contents of the source volumes, or independent exact
duplicates (clones), depending on the snapshot type and the
time since creation.
See also replica and snapshot creation.
snapshot backup
(HP StorageWorks
VA and HP
StorageWorks EVA
specific term)
See ZDB to tape, ZDB to disk, and ZDB to disk+tape.
snapshot creation
(HP StorageWorks VA and HP StorageWorks EVA specific term)
A replica creation technique, in which copies of source volumes
are created using storage virtualization techniques. The replicas
are considered to be created at one particular point in time,
without pre-configuration, and are immediately available for
Concepts guide
389
use. However background copying processes normally continue
after creation.
See also snapshot.
390
source (R1) device
(EMC Symmetrix specific term) An EMC Symmetrix device that
participates in SRDF operations with a target (R2) device. All
writes to this device are mirrored to a target (R2) device in a
remote EMC Symmetrix unit. An R1 device must be assigned to
an RDF1 group type.
See also target (R2) device.
source volume
(ZDB specific term) A storage volume containing data to be
replicated.
sparse file
A file that contains data with portions of empty blocks. Examples
are: a matrix in which some or much of the data contains zeros,
files from image applications, and high-speed databases. If
sparse file processing is not enabled during restore, it might be
impossible to restore this file.
split mirror
(EMC Symmetrix and HP StorageWorks Disk Array XP specific
term) A replica created using split mirror techniques. Such a
replica provides an independent, exact duplicate, or clone of
the contents of the source volumes.
See also replica and split mirror creation.
split mirror backup
(EMC Symmetrix
specific term)
See ZDB to tape.
split mirror backup
(HP StorageWorks
Disk Array XP
specific term)
See ZDB to tape, ZDB to disk, and ZDB to disk+tape.
split mirror
creation
(EMC Symmetrix and HP StorageWorks Disk Array XP specific
term) A replica creation technique, in which a pre-configured
set of target volumes (a mirror) is kept synchronized with a set
of source volumes until the time at which a replica of the contents
of the source volumes is required. Then, the synchronization is
stopped (the mirror is split) and a split mirror replica of the
source volumes at the time of the split remains in the target
volumes.
See also split mirror.
Glossary
split mirror restore
(EMC Symmetrix and HP StorageWorks Disk Array XP specific
term) A process in which data backed up in a ZDB-to-tape or a
ZDB-to-disk+tape session is restored from tape media to a split
mirror replica, which is then synchronized to the source volumes.
Individual backup objects or complete sessions can be restored
using this method.
See also ZDB to tape, ZDB to disk+tape, and replica.
sqlhosts file
(Informix Server specific term) An Informix Server connectivity
information file (on UNIX) or registry (on Windows) that contains
the names of each of the database servers and any aliases to
which the clients on a host computer can connect.
SRD file
The Data Protector System Recovery Data (SRD) file contains
system information required for installing and configuring the
operating system in case of a disaster. The SRD file is an ASCII
file, generated when a CONFIGURATION backup is performed
on a Windows client and stored on the Cell Manager.
SRDF
(EMC Symmetrix specific term) The EMC Symmetrix Remote
Data Facility is a business continuation process that enables
effective, real-time data replication of SLDs between dislocated
processing environments. These environments could be situated
within the same root computer environment or separated by
long distances.
SSE Agent
(HP StorageWorks Disk Array XP specific term) A Data Protector
software module that executes all tasks required for a split mirror
backup integration. It communicates with the HP StorageWorks
Disk Array XP storing system using the RAID Manager XP utility
(HP-UX and Windows systems) or RAID Manager Library (Solaris
systems).
sst.conf file
The file /usr/kernel/drv/sst.conf is required on each
Data Protector Sun Solaris client to which a multi-drive library
device is connected. It must contain an entry for the SCSI address
of the robotic mechanism of each library device connected to
the client.
st.conf file
The file /kernel/drv/st.conf is required on each Data
Protector Solaris client with a backup device connected. It must
contain device information and a SCSI address for each backup
drive connected to that client. A single SCSI entry is required
Concepts guide
391
for a single-drive device and multiple SCSI entries are required
for a multi-drive library device.
392
stackers
Devices with multiple slots for media storage usually with only
one drive. A stacker selects media from the stack sequentially.
In contrast, a library can randomly select media from its
repository.
standalone file
device
A file device is a file in a specified directory to which you back
up data.
Storage Group
(Microsoft Exchange Server specific term) A collection of mailbox
stores and public folder stores that share a set of transaction
log files. Exchange Server manages each storage group with a
separate server process.
StorageTek ACS
library
(StorageTek specific term) Automated Cartridge System is a
library system (also known as Silo) consisting of one Library
Management Unit (LMU) and one to 24 Library Storage Modules
(LSM) connected to the unit.
storage volume
(ZDB specific term) A storage volume represents an object that
may be presented to an operating system or some other entity
(for example, a virtualization mechanism) upon which volume
management systems, file systems, or other objects may exist.
The volume management systems, file systems are built on this
storage. Typically, these can be created or exist within a storage
system such as a disk array.
switchover
See failover.
Sybase Backup
Server API
(Sybase specific term) An industry-standard interface developed
for the exchange of backup and recovery information between
a Sybase SQL Server and a backup solution like Data Protector.
Sybase SQL Server
(Sybase specific term) The server in the Sybase “client-server”
architecture. Sybase SQL Server manages multiple databases
and multiple users, keeps track of the actual location of data on
disks, maintains mapping of logical data description to physical
data storage, and maintains data and procedure caches in
memory.
Symmetrix Agent
(SYMA)
(EMC Symmetrix specific term) The Data Protector software
module that prepares the EMC Symmetrix environment for
backup and restore operations.
Glossary
synthetic backup
A backup solution that produces a synthetic full backup, an
equivalent to a conventional full backup in terms of data, without
putting stress on the production servers or the network. A
synthetic full backup is created from a previous full backup and
any number of incremental backups.
synthetic full
backup
The result of an object consolidation operation, where a restore
chain of a backup objects is merged into a new, synthetic full
version of this object. A synthetic full backup is equivalent to a
conventional full backup in terms of restore speed.
System Backup to
Tape
(Oracle specific term) An Oracle interface that handles the
actions required to load, label, and unload correct backup
devices when Oracle issues a backup or restore request.
system databases
(Sybase specific term) The four system databases on a newly
installed Sybase SQL Server are the:
• master database (master)
• temporary database (tempdb)
• system procedure database (sybsystemprocs)
• model database (model).
System State
(Windows specific term) The System State data comprises the
Registry, COM+ Class Registration database, system startup
files, and the Certificate Services database (if the server is a
certificate server). If the server is a domain controller, Active
Directory services and the SYSVOL directory are also contained
in the System State data. If the server is running the Cluster
service, the System State data also includes resource registry
checkpoints and the quorum resource recovery log, which
contains the most recent cluster database information.
system
volume/disk/
partition
A volume/disk/partition containing operating system files.
Microsoft terminology defines the system volume/disk/partition
as the volume/disk/partition containing files required for the
initial step of the boot process.
SysVol
(Windows specific term) A shared directory that stores the server
copy of the domain’s public files, which are replicated among
all domain controllers in the domain.
Concepts guide
393
394
tablespace
A part of a database structure. Each database is logically
divided into one or more tablespaces. Each tablespace has data
files or raw volumes exclusively associated with it.
tapeless backup
(ZDB specific term)
See ZDB to disk.
target database
(Oracle specific term) In RMAN, the target database is the
database that you are backing up or restoring.
target (R2) device
(EMC Symmetrix specific term) An EMC Symmetrix device that
participates in SRDF operations with a source (R1) device. It
resides in the remote EMC Symmetrix unit. It is paired with a
source (R1) device in the local EMC Symmetrix unit and receives
all write data from its mirrored pair. This device is not accessed
by user applications during normal I/O operations. An R2
device must be assigned to an RDF2 group type.
See also source (R1) device.
target system
(disaster recovery specific term) A system after a computer
disaster has occurred. The target system is typically in a
non-bootable state and the goal of disaster recovery is to restore
this system to the original system configuration. The difference
between a faulty system and a target system is that a target
system has all faulty hardware replaced.
target volume
(ZDB specific term) A storage volume to which data is replicated.
Terminal Services
(Windows specific term) Windows Terminal Services provide a
multi-session environment that allows clients to access a virtual
Windows desktop session and Windows-based programs
running on the server.
thread
(Microsoft SQL Server specific term) An executable entity that
belongs to only one process. It comprises a program counter,
a user-mode stack, a kernel-mode stack, and a set of register
values. Several threads can run at a time within one process.
TimeFinder
(EMC Symmetrix specific term) A business continuation process
that creates an instant copy of single or multiple Symmetrix
Logical Devices (SLDs). The instant copy is created on specially
preconfigured SLDs called BCVs and is accessible via a separate
device address to the system(s).
TLU
Tape Library Unit.
Glossary
TNSNAMES.ORA
(Oracle and SAP R/3 specific term) A network configuration
file that contains connect descriptors mapped to service names.
The file may be maintained centrally or locally, for use by all
or individual clients.
transaction
A mechanism for ensuring that a set of actions is treated as a
single unit of work. Databases use transactions to keep track of
database changes.
transaction backup
Transaction backups generally use fewer resources than
database backups, so they can be created more frequently than
database backups. By applying transaction backups, you can
recover the database to a specific point in time prior to when
a problem occurred.
transaction backup
(Sybase and SQL specific term) A backup of the transaction log
providing a record of changes made since the last full or
transaction backup.
transaction log
backup
Transaction log backups generally use fewer resources than
database backups so they can be created more frequently than
database backups. By applying transaction log backups, you
can recover the database to a specific point in time.
transaction log
files
Files that record transactions of the database modifications, and
provide fault tolerance in case of a database disaster.
transaction logs
(Data Protector specific term) Keep track of IDB changes. The
archiving of transaction logs should be enabled to prevent you
from losing the transaction log files that are created after the
last IDB backup and are necessary for IDB recovery.
transaction log
table
(Sybase specific term) A system table in which all changes to
the database are automatically recorded.
transportable
snapshot
(Microsoft VSS specific term) A shadow copy that is created on
the application system and can be presented to the backup
system where a backup can be performed.
See also Microsoft Volume Shadow Copy Service (VSS).
TSANDS.CFG file
(Novell NetWare specific term) A file that allows you to specify
the names of containers where you want backups to begin. It
is a text file located in the SYS:SYSTEM\TSA directory on the
server where TSANDS.NLM is loaded.
Concepts guide
395
396
UIProxy
The Java GUI Server (UIProxy service) runs on the Data
Protector Cell Manager. It is responsible for communication
between the Java GUI Client and the Cell Manager, moreover,
it performs business logic operations and sends only important
information to the client. The service is started as soon as Data
Protector is installed on the Cell Manager.
unattended
operation
See lights-out operation.
user account (Data
Protector user
account)
You can use Data Protector only if you have a Data Protector
user account, which restricts unauthorized access to Data
Protector and to backed up data. Data Protector administrators
create this account specifying a user logon name, the systems
from which the user can log on, and a Data Protector user group
membership. This is checked whenever the user starts the Data
Protector user interface or performs specific tasks.
User Account
Control (UAC)
A security component in Windows Vista and Windows
Server 2008 that limits application software to standard user
privileges until an administrator authorizes an increase in
privilege level.
user disk quotas
NTFS quota-management support enables an enhanced tracking
mechanism and control over disk space usage on shared storage
volumes. Data Protector backs up user disk quotas on the whole
system and for all configured users at a time.
user group
Each Data Protector user is member of a User Group. Each User
Group has a set of user rights that are given to every user in
that User Group. The number of User Groups with their
associated user rights can be defined as desired. Data Protector
provides three default user groups: admin, operator, and user.
user profile
(Windows specific term) Configuration information retained on
a user basis. This information includes desktop settings, screen
colors, network connections, and so on. When the user logs on,
the user profile is loaded and the Windows environment is set
accordingly.
user rights
User rights or access rights are the permissions needed to
perform specific Data Protector tasks. Configuring a backup,
starting a backup session, or starting a restore session are typical
Glossary
user rights. Users have the access rights of the user group to
which they belong.
vaulting media
The process of storing media to a safe and remote place. The
media are brought back to the data center when they are
needed for restore or are ready for reuse in subsequent backups.
The vaulting procedure depends on your company's backup
strategy and policies for data protection/reliability.
verify
A function that lets you check whether the Data Protector data
on a specified medium is readable. Additionally, consistency
within each block can be checked if the backup was performed
with the cyclic redundancy check (CRC) option ON.
Virtual Controller
Software (VCS)
(HP StorageWorks EVA specific term) The firmware that manages
all aspects of storage system operation, including communication
with Command View EVA through the HSV controllers.
See also Command View (CV) EVA.
Virtual Device
Interface
(Microsoft SQL Server specific term) This is a SQL Server
programming interface that allows fast backup and restore of
large databases.
virtual disk
(HP StorageWorks EVA specific term) A unit of storage allocated
from an HP StorageWorks Enterprise Virtual Array storage pool.
Virtual disks are the entities that are replicated using the HP
StorageWorks Enterprise Virtual Array snapshot functionality.
See also source volume and target volume.
virtual full backup
An efficient type of synthetic backup where data is consolidated
using pointers instead of being copied. It is performed if all the
backups (the full backup, incremental backups, and the resulting
virtual full backup) are written to a single file library that uses
distributed file medium format.
Virtual Library
System (VLS)
A disk-based data storage device hosting one or more virtual
tape libraries (VTLs).
virtual server
A virtual machine in a cluster environment defined in a domain
by a network IP name and address. Its address is cached by
the cluster software and mapped to the cluster node that is
currently running the virtual server resources. This way all
requests for a particular virtual server are cached by a specific
cluster node.
Concepts guide
397
398
virtual tape
(VLS specific term) An archival storage technology that backs
up data to disk drives in the same way as if it were being stored
on tape. Benefits of virtual tape systems include improved backup
and recovery speed and lower operating costs.
See also Virtual Library System (VLS) and Virtual Tape Library.
Virtual Tape
Library (VTL)
(VLS specific term) An emulated tape library that provides the
functionality of traditional tape-based storage.
See also Virtual Library System (VLS).
VMware
management client
(VMware integration specific term) The client that Data Protector
uses to communicate with VMware Virtual Infrastructure. This
can be a VirtualCenter Server system (VirtualCenter environment)
or an ESX Server system (standalone ESX Server environment).
volser
(ADIC and STK specific term) A VOLume SERial number is a
label on the medium to identify the physical tape used in very
large libraries. A volser is a naming convention specific to
ADIC/GRAU and StorageTek devices.
volume group
A unit of data storage in an LVM system. A volume group can
consist of one or more physical volumes. There can be more
than one volume group on the system.
volume mount
point
(Windows specific term) An empty directory on a volume that
can be used to mount another volume. The volume mount point
acts as a gateway to the target volume. Provided that the volume
is mounted, users and applications can refer to the data on the
mounted volume by the full (merged) filesystem path as if both
volumes are one integral part.
Volume Shadow
Copy Service
See Microsoft Volume Shadow Copy Service.
VSS
See Microsoft Volume Shadow Copy Service.
VSS compliant
mode
(HP StorageWorks Disk Array XP VSS provider specific term)
One of two XP VSS hardware provider operation modes. When
the XP provider is in the VSS compliant mode, the source volume
(P-VOL) and its replica (S-VOL) are in simplex, unpaired state
after a backup. Therefore the number of replicas (S-VOLs per a
P-VOL) rotated is not limited. Restore from a backup in such a
configuration is possible only by switching of the disks.
Glossary
See also resync mode, source volume, primary volume (P-VOL),
replica, secondary volume (S-VOL), and replica set rotation.
VxFS
Veritas Journal Filesystem.
VxVM (Veritas
Volume Manager)
A Veritas Volume Manager is a system for managing disk space
on Solaris platforms. A VxVM system consists of an arbitrary
group of one or more physical volumes organized into logical
disk groups.
Wake ONLAN
Remote power-up support for systems running in power-save
mode from some other system on the same LAN.
Web reporting
The Data Protector functionality that allows you to view reports
on backup, object copy, and object consolidation status and
Data Protector configuration using the Web interface.
wildcard character
A keyboard character that can be used to represent one or many
characters. The asterisk (*), for example, typically represents
one or more characters, and the question mark (?) typically
represents a single character. Wildcard characters are often
used in operating systems as a means of specifying more than
one file by name.
Windows
CONFIGURATION
backup
Data Protector allows you to back up Windows
CONFIGURATION, including Windows Registry, user profiles,
Event Logs, and WINS and DHCP server data (if configured on
a system) in one step.
Windows Registry
A centralized database used by Windows to store configuration
information for the operating system and the installed
applications.
WINS server
A system running Windows Internet Name Service software that
resolves Windows networking computer names to IP addresses.
Data Protector can back up WINS server data as part of the
Windows configuration.
writer
(Microsoft VSS specific term) A process that initiates change of
data on the original volume. Writers are typically applications
or system services that write persistent information on a volume.
Writers also participate in the shadow copy synchronization
process by assuring data consistency.
Concepts guide
399
400
XBSA interface
(Informix Server specific term) ON-Bar and Data Protector
communicate with each other through the X/Open Backup
Services Application Programmer's Interface (XBSA).
XCopy engine
(direct backup specific term) A SCSI-3 copy command that allows
you to copy data from a storage device having a SCSI source
address to a backup device having a SCSI destination address,
thus enabling direct backup. The data flows from a source device
(either block or streaming, that is, disk or tape) to the destination
device (either block or streaming) through XCopy. This releases
the controlling server of reading the data from the storage device
into memory and then writing the information to the destination
device.
See also direct backup.
ZDB
See zero downtime backup (ZDB).
ZDB database
(ZDB specific term) A part of the IDB, storing ZDB related
information such as source volumes, replicas and security
information. The ZDB database is used for ZDB, instant recovery,
and split mirror restore.
See also zero downtime backup (ZDB).
ZDB to disk
(ZDB specific term) A form of zero downtime backup where the
replica produced is kept on the disk array as a backup of the
source volumes at a specific point in time. Multiple replicas,
produced using the same backup specification at different times,
can be kept in a replica set. A replica from ZDB to disk can be
restored using the instant recovery process.
See also zero downtime backup (ZDB), ZDB to tape, ZDB to
disk+tape, instant recovery, and replica set rotation.
ZDB to disk+tape
(ZDB specific term) A form of zero downtime backup where the
replica produced is kept on the disk array as a backup of the
source volumes at a specific point in time, in the same way as
with ZDB to disk. However, data in the replica is also streamed
to a backup medium, as with ZDB to tape. If this backup method
is used, data backed up in the same session can be restored
using the instant recovery process, the standard Data Protector
restore from tape, or on split mirror arrays, split mirror restore.
See also zero downtime backup (ZDB), ZDB to disk, ZDB to
tape, instant recovery, replica, and replica set rotation.
Glossary
ZDB to tape
(ZDB specific term) A form of zero downtime backup where
data in the replica produced is streamed to a backup medium,
typically tape. Instant recovery is not possible from such a
backup, so the replica need not be retained on the disk array
after backup completion. The backed-up data can be restored
using standard Data Protector restore from tape. On split mirror
arrays, split mirror restore can also be used.
See also zero downtime backup (ZDB), ZDB to disk, instant
recovery, ZDB to disk+tape, and replica.
zero downtime
backup (ZDB)
A backup approach in which data replication techniques,
provided by a disk array, are used to minimize the impact of
backup operations on an application system. A replica of the
data to be backed up is created first. All subsequent backup
operations are performed on the replicated data rather than the
original data, while the application system can be returned to
normal operation.
See also ZDB to disk, ZDB to tape, ZDB to disk+tape, and instant
recovery.
Concepts guide
401
402
Glossary
Index
A
adding data to media during backups,
148
ADIC (EMASS/GRAU) AML, 162
admin user group, 185
alarms, 208
alternative disaster recovery methods,
131
operating system vendors, 131
third-party tools, 131
ANSI X3.27 labels, 146
any-to-any connectivity, 171
Application Agents, 42
Application Response Measurement
real time alerts, 209
application client
snapshot backup, 277
split mirror backup, 266
Application Response Measurement,
207, 208
real time alerts, 208
response time, 208
transactions, 208
architecture
backup devices, 40
Cell Managers, 40
cells, 40
archive log backup
snapshot backup, 278
split mirror backup, 267
ARM 2.0, 208
audience, 23
auditing, 208
autoloaders, 162
See also libraries
automated smart media copying, 124
automated media copying, 123
examples, 332
automated object consolidation
sessions, 233
automated object copy sessions, 230
automated operation, 37, 110
B
backed up data
hiding from other users, 76
visibility, 76
backing up data, 102 - 110
procedure, 102
backup
IDB operation, 194
to disk, 253
Backup Agents, 42
backup client
snapshot backup, 277
backup concurrency, 329
backup configuration, 106
backup generations, 324
backup interfaces, 240
backup object, 103
Backup Session Manager, 221
backup specifications, 309, 326
backup strategy planning
data protection, 61
device configuration, 61
media management, 62
Concepts guide
403
backup with disk discovery, 225
Backup Agents, 42
backup client
split mirror backup, 267
backup client as failover server
snapshot backup, 286
split mirror backup, 268
backup concurrency, 157, 311
backup devices, 49, 68
overview, 153
backup duration
example calculations, 307, 324
backup environment growth
database growth and performance
key factors, 198
database growth and performance
key factors, 198
backup environments, 299, 313
backup generations, 144, 307, 331
backup options, 310, 328
backup overview, 38
backup ownership, 79
backup performance, 157
backup policies, 45, 152
enterprise environment, 45
backup process
destination, 38
source, 38
backup scenarios (company ABC), 313,
330
backup scenarios (company XYZ), 299,
313
backup session
definition, 105, 220
ownership, 78
backup sessions, 43, 102, 106, 220 - 225
backup configuration, 106
interactive, 221
mount requests, 224
scheduled, 221
timeout, 224
backup specifications, 49, 103, 103
backup strategy, 57
404
backup strategy planning
system availability, 60
backup strategy factors, 60
backup strategy planning, 57 - 131
backup policies, 61
catalog protection, 61
data encryption, 76
data types, 61
defining requirements, 58
definition, 58
scheduling backups, 61
backup strategy requirements, 302, 315
backup types, 107
full, 71, 91, 92
incremental, 71, 91, 92
planning performance , 71
backups
adding data to media, 148
automated, 110
backup specifications, 103
backup objects, 103
configuring, 70
devices, 153
direct, 68
disk image, 71
disk discovery vs. standard backup,
225
filesystem, 71
lights-out, 110
local, 67
network, 67, 68
scheduled, 105
scheduling policies, 105
sessions, 106
staggering, 107
standard backup vs. disk discovery,
225
unattended, 110
barcode support, 163
barcodes, 163
benefits
disk backup, 254
synthetic backup, 258
Volume Shadow Copy service, 293
benefits of online integrations, 240
block size
backup devices, 158
default, 158
devices, 158
performance, 158
broadcasts, 207
browsing files, 101
BSM, 221
C
cache memory, 72, 238
Catalog Database
location, 192
records, 191
Catalog Database growth factors
catalog protection, 100
catalog protection, 310
as an IDB key tunable parameter, 201
backup generations, 332
Catalog Database, 191
do not log any details, 100
filename size and growth, 191
log all detailed information, 100
log directory names only, 100
log level of information, 104
size and growth for CDB Records
other than filenames, 192
Catalog Database growth factors
level of details, 100
catalog protection, 100
browsing files, 101
expired, 201
IDB size and growth, 188
impact on backup performance, 201
restoring data when catalog
protection expires, 201
catalog protection as an IDB key
tunable parameter, 201
CDB location
Catalog Database, 192
CDB records
Catalog Database, 191
CDB.
See Catalog Database
Cell Request Server, 220
Cell Managers, 64
high availability, 82
optimizing the load, 224
cells
backup operation, 42
Cell Managers, 41
logical view, 41
mixed environment, 66
multiple, 45, 62
physical view, 41
planning, 62
planning security, 74
remote, 66
restore operation, 42
single-point management, 46
splitting, 45
UNIX environment, 64
Windows environment, 65
Windows 2000 environment, 65
Windows domains, 65
Windows workgroups, 66
Centralized Media Management
Database, 47
centralized licensing, 47
Centralized Media Management
Database, 319
Centralized Media Management
Database, 189
character encoding standards, 341
checkpoints, 239
cleaning tape detection, 163
cleaning tape support, 164
magazine devices, 161
magazines, 161
Concepts guide
405
client systems, 42
clients, 42
installing, 64
maintaining, 64
cluster heartbeat, 80
cluster (definition), 79
cluster integrations
overview, 82
cluster node, 80
clustering, 79 - 91
automatic restart, 82
Cell Manager availability, 82
device sharing, 180
failover, 81
floating drives, 181
group, 81
heartbeat, 80
load balancing, 82
MC/Service Guard, 79
Microsoft Cluster Server, 79
nodes, 80
package, 81
primary node, 81
secondary node, 81
shared disks, 80
Veritas Cluster, 79
virtual cluster node backup, 84, 86,
89
virtual server, 81
CMMDB, 47, 319
CMMDB.
See Centralized Media Management
Database
code sets, 341
collision, 160
commands
omniclus command, 90
post-exec, 223, 239
pre-exec, 223, 239
company backup policies, 152
comparison
disk-based devices, 255
complete filesystem restore, 312
406
complete filesystem restore, 330
compression
hardware, 68, 70
software, 70
concepts
snapshot backup, 276
split mirror backup, 265
concurrency, 156
concurrent sessions
backup, 222
media management, 236
object consolidation, 234
object copy, 231
restore, 227
configuring cells, 304, 319
configuring devices, 153
configuring backup specifications, 103
configuring devices
large libraries, 162
magazines, 161
standalone devices, 160
consolidating a restore chain, 118
control files, 238
conventional incremental backup, 93
conventions
document, 31
copying media
automated, 123
copying objects
to implement disk staging, 118
copying backed up data, 112
copying media, 122
smart media copying, 124
copying objects, 113
for vaulting purposes, 116
to consolidate a restore chain, 118
to demultiplex a medium, 117
to free a medium, 116
to migrate to another media type, 118
creating backup specifications, 103
creating cells
Windows 2000 environment, 65
creating cells
mixed environment, 66
UNIX environment, 64
Windows domains, 65
Windows workgroups, 66
Windows environment, 65
CRS, 220
D
daily maintenance
IDB operation, 196
data
hiding from other users, 76
visibility, 76
data encoding, 76
data encryption, 76
data protection, 310
Data Protector architecture
Cell Managers, 40
Data Protector features, 35
Data Protector GUI, 50
data encoding, 76
data files, 238
data protection, 99
Data Protector concepts
cells, 40
Data Protector Inet, 219
Data Protector services, 219 - 236
Data Protector Inet, 219
Data Protector architecture
cell, 40
client systems, 40
devices, 40
logical view, 41
physical view, 41
Data Protector concepts
Cell Managers, 40
clients, 40
devices, 40
Data Protector functionality, 35
Data Protector GUI
Data Protector Java GUI, 52
Data Protector Java GUI, 52
Data Protector operation, 219 - 236
Data Protector processes, 219 - 236
Cell Request Server, 220
Data Protector Inet, 219
Media Management Daemon, 220
Raima Database Server, 220
Data Protector services
Cell Request Server, 220
Media Management Daemon, 220
Raima Database Server, 220
Data Protector setup, 55
Data Protector user accounts, 74
Data Protector user groups, 75
Data Protector user interfaces, 42, 50
Data Protector user rights (definition),
75
Data Source Integration, 209
database
advantages, 187
architecture, 189
Catalog Database, 191
catalog protection, 188
Detail Catalog Binary Files, 192
growth and performance, 197
IDB management, 196
in the Manager-of-Managers
environment, 189
Media Management Database, 190
on the Windows Cell Manager, 188
operation, 194
Serverless Integrations Binary Files,
194
Session Messages Binary Files, 193
size and growth, 188
UNIX Cell Managers, 189
database growth and performance key
tunable parameters
catalog protection, 201
usage of logging level and catalog
protection, 201
Concepts guide
407
database on the UNIX Cell Managers
IDB format, 189
IDB location, 189
database on the Windows Cell
Manager
IDB format, 188
IDB location, 188
database architecture, 189
database growth and performance key
factors
filesystem dynamics, 197
database growth and performance key
factors, 197
database growth and performance key
tunable parameters, 198
logging level, 199
database growth and performance key
factors
backup environment growth, 198
backup environment growth, 198
database growth and performance key
tunable parameters, 199
database in the MoM environment
Centralized Media Management
Database, 189
database in the MoM environment, 189
Database Library, 240
database on the Windows Cell
Manager, 188
database operation, 237
database size estimation, 203
408
databases, 237
backup interfaces, 240
cache memory, 238
Centralized Media Management
Database, 47
checkpoints, 239
control files, 238
data files, 238
dbspaces, 237
files, 237
online backups, 239
segments, 237
tables, 237
tablespaces, 237
transaction logs, 238
dbspaces, 237
DC binary file
Detail Catalog Binary Files, 192
IDB operation, 194
DC directory
Detail Catalog Binary Files, 192
DCBF information
Detail Catalog Binary Files, 192
DCBF location
Detail Catalog Binary Files, 193
DCBF size and growth
Detail Catalog Binary Files, 192
DCBF.
See Detail Catalog Binary Files
default block size, 158
default media pools, 136
demultiplexing media, 117
Detail Catalog Binary Files, 192
DC directory, 192
Detail Catalog Binary Files
DC binary file, 192
DCBF size and growth, 192
information, 192
location, 193
device chaining, 155
device configuration, 153
device chains, 161
device collision, 160
device lists, 155
device locking, 159
device sharing in clusters, 180
device sharing in SAN, 175
drives, 177
robotics, 177
device streaming (definition), 156
devices, 49, 68, 153 - 181
disk-based, 255
ADIC (EMASS/GRAU) AML, 162
autoloaders, 162
cleaning tape support, 164
concurrency, 156
configuring, 153
device chaining, 155
device lists, 155
device locking, 159
device streaming, 156
exchangers, 162
GRAU/EMASS, 162
HP StorageWorks DAT Autoloaders,
324
HP StorageWorks DAT24
Autoloaders, 307
HP StorageWorks DLT 4115w
Libraries, 307
HP StorageWorks DLT 4228w
Libraries, 323
jukeboxes, 162
library management console,
support, 154
load balancing, 155
lock names, 159
multiple devices, 155
number of buffers, 159
overview, 153
physical device collision, 160
planning performance, 68
SCSI libraries, 162
segment size, 157
selecting for restore, 126
standalone, 160
StorageTek/ACSLS, 162
TapeAlert support, 154
Direct Backup
overview, 243
supported configurations, 250
Direct backup
requirements, 249
Direct Backup, 243
Concepts guide
409
Direct Library Access, 179
dirty drive detection, 164
disaster, 128
Disaster Recovery
Phase 3, 129
disaster recovery, 129
Disaster Recovery
concepts, 128
overview, 128
Phase 1, 129
Phase 0, 129
Phase 2, 129
disaster recovery
alternative methods, 131
disaster recovery
alternative , 131
Disk Agent concurrency, 329
disk performance, 72
Disk Agent concurrency, 157, 311
Disk Agents, 42
disk backup, 253
benefits, 254
disk based devices
overview, 253
disk discovery (definition), 225
disk discovery vs. standard backup, 225
disk fragmentation, 72
disk image backups, 71, 72
disk image vs. filesystem backups, 71
disk performance, 72
cache memory, 72
compression, 72
disk image backups, 72
disk staging, 118
disk-based devices
comparison, 255
do not log any details
Catalog Database, 100
document
conventions, 31
related documentation, 23
410
documentation
HP website, 23
providing feedback, 34
Drive Servers, 42
drives, 177
connecting to multiple systems, 165
floating, 181
static, 180
duplicating backed up data, 112
E
e-mail, 207
EMC Symmetrix, 266
encoding, 76
encryption, 76
drive-based, 76, 77
encryption key, 76
Key Management Server, 76
software-based, 76
encryption key
Key Management Server, 76
end-user user group, 185
enhanced incremental backup, 93
enterprise environment, 45
enterprise reporting, 47
environment
enterprise, 45
Manager-of-Managers, 45
mixed, 66
network, 39
UNIX, 64
Windows, 65
examples
backup scenarios, 297
media pool usage, 140
reporting and notification, 212
scheduling policies, 108
using data provided by Data
Protector, 214
vaulting usage, 152
examples of media usage policies, 148
exchangers, 162
See also libraries
expired catalog protection, 201
exporting media, 102
IDB operation, 195
removed objects, 195
F
factors affecting restore duration, 125
factors influencing backup strategies,
60
failover, 81, 82
FC-AL, 173
features of Data Protector, 35
fibre channel
planning performance, 73
Fibre Channel (definition), 172
Fibre Channel Arbitrated Loop, 173
Fibre Channel topologies, 173
loop topology, 173
point-to-point, 173
switched topology, 174
file jukebox device, 255
file library device, 255
file versions purge, 196
Filename Handling, 341
filename size and growth
Catalog Database, 191
fnames.dat file, 191
filenames purge
IDB operation, 196
filesystem backup
Volume Shadow Copy service, 295
filesystem dynamics
database growth and performance
key factors, 197
filesystem backup, 71
Volume Shadow Copy service, 293
filesystem vs. disk image backups, 71
floating drives, 181
fnames.dat file
filename size and growth, 191
formatting media, 135
fragmentation, 72
freeing media, 116
full and incremental backups, 91 - 98
full backups, 71
staggering, 107
functionality of Data Protector, 35
further information, 331
G
General Media Agent, 165
geographically remote cells, 66
GRAU/EMASS, 162
group, 81
H
hardware compression, 68, 70
heartbeat, 80
help
obtaining, 33
high availability, 36, 82
snapshot backup, 275
split mirror backup, 267
HP
technical support, 33
HP Operations Manager software, 208,
209, 210
HP Performance Agent, 206, 208
HP Performance Agent integration, 209
HP StorageWorks Disk Array XP, 266
HP StorageWorks DAT24 Autoloaders,
307, 324
HP StorageWorks DLT 4115w Libraries,
307
HP StorageWorks DLT 4228w Libraries,
323
HP StorageWorks Enterprise Virtual
Array, 276
HP StorageWorks Virtual Array, 276
HTML, 207
Concepts guide
411
I
IDB, 187
advantages, 187
architecture, 189
Catalog Database, 191
Detail Catalog Binary Files, 192
in the Manager-of-Managers
environment, 189
management, 196
Media Management Database, 190
on the UNIX Cell Managers, 189
on the Windows Cell Manager, 188
operation, 194
Serverless Integrations Binary Files,
194
Session Messages Binary Files, 193
size and growth, 188
IDB architecture
IDB parts scheme, 190
IDB growth and performance
key factors, 197
IDB management
IDB configuration, 196
IDB maintenance, 197
IDB recovery, 197
overview, 196
IDB operation
exporting media, 195
filenames purge, 196
IDB size and growth
catalog protection, 188
IDB advantages, 187
IDB architecture, 189
Catalog Database, 191
Detail Catalog Binary Files, 192
IDB parts, 189
Media Management Database, 190
Serverless Integrations Binary Files,
194
Session Messages Binary Files, 193
412
IDB configuration
creating a backup specification for
the IDB backup, 196
IDB management, 196
IDB format
UNIX Cell Managers, 189
Windows Cell Manager, 188
IDB growth and performance
backups as key factors, 197
IDB growth and performance, 197
database size estimation, 203
key tunable parameters, 198
IDB in the MoM environment
Centralized Media Management
Database, 189
IDB location
UNIX Cell Managers, 189
Windows Cell Manager, 188
IDB maintenance
IDB management, 197
IDB management
setting up backup environment, 196
IDB operation, 194
backup, 194
daily maintenance, 196
DC binary file, 194
media position record, 194
restore, 195
session messages binary files, 194
IDB parts
architecture, 189
IDB parts scheme
IDB architecture, 190
IDB recovery
IDB management, 197
IDB size and growth, 188
logging level, 188
incremental backup types
conventional incremental backups,
93
enhanced incremental backups, 93
leveled incremental backups, 94
incremental backups, 71
types, 93
Indirect
Storage Are Networks, 178
Indirect Library Access
Library Access, 178
Indirect Library Access, 179
influence of logging level and catalog
protection on IDB growth scheme, 199
initializing media, 135
media ID, 145
Installation Servers, 42, 64
instant recovery
snapshot backup, 279
split mirror backup, 267
integration with database applications,
38, 237 - 241
integrations, 210
Volume Shadow Copy service, 293
interactive backup sessions, 221
interactive object consolidation sessions,
233
interactive object copy sessions, 230
interactive smart media copying, 124
internal database.
See IDB
Internationalization, 340
IT management, 206
J
Java GUI Client, 53
Java GUI Server, 53
java reporting, 213
java-based online reporting, 213
jukeboxes, 162
See also libraries
K
Key Management Server, 76
KMS
See Key Management Server
L
labeling media, 145
labels, 146
LAN-free backups, 175
large libraries, 162 - 170
level 1 incremental backups, 310, 326
leveled incremental backups, 94
libraries, 47
barcode support, 163
cleaning tape support, 164
connecting to multiple systems, 165
drives, 165
entering and ejecting mail slots, 163
HP StorageWorks DAT Autoloaders,
324
HP StorageWorks DAT24
Autoloaders, 307
HP StorageWorks DLT 4115w
Libraries, 307
HP StorageWorks DLT 4228w
Libraries, 323
management console, support, 154
media handling, 162
multiple slots, 163
sharing, 163
silo, 162
size, 162
slot range, 162
slots, 162
Library Access
Direct, 179
library management console, support,
154
library sharing, 164
library size, 162
life cycle, media, 134
lights-out operation, 37, 110
LIP, 173
load balancing, 70, 82, 104, 155
load balancing (definition), 155
Localization, 340
location fields, 146
Concepts guide
413
lock names, 159, 177
log all detailed information
Catalog Database, 100
log directory names only
Catalog Database, 100
log level of information, 104
logging level
IDB size and growth, 188
logging level
enabling restore, 200
impact on ability to browse for
restore, 200
impact on IDB speed and backup
processes, 200
impact on restore speed, 200
Log All, 199
Log Directories, 200
Log File, 199
No Log, 200
Loop Initialization Primitive (Protocol),
173
loop topology, 173
M
magazine devices
cleaning, 161
management console
See library management console
Manager-of-Managers, 46, 320
enterprise reporting, 47
remote cells, 67
sharing libraries, 47
MC/Service Guard, 79
414
media
age, 151
barcode support, 163
barcodes, 163
catalog segments, 157
cleaning tape support, 164
copying, 122
copying, automated, 123
data segments, 157
device errors, 151
ejecting mail slots, 163
encrypting, 77
entering mail slots, 163
estimating quantity of needed media,
144
exporting, 102
file marks, 157
formatting, 135
header segments, 157
initializing, 135, 145
labeling, 145, 163
location fields, 146
mail slots, 163
number of overwrites, 150
object distribution, 71
preparing, 134
retiring, 134
selecting for backup, 147
selecting for restore, 126
smart copying using VLS, 124
vaulting, 134, 151
Media Agents
NDMP Media Agent, 165
media condition
poor, 148
media copies, 122
media life cycle, 134
media management
copies, 122
selecting media, 147
Media Management Database, 190
media management session (definition),
235
media pools, 135, 309, 325
default, 136
properties, 136
usage examples, 140
Media Agents, 42
General Media Agent, 165
media allocation policies, 144
media allocation policies, 136, 147
loose, 147
strict, 147
media condition, 150
calculating, 150
fair, 148
good, 148
media condition factors, 150
media description, 145
media handling, 144, 162
media location, 145
media location priority, 126
media management, 48, 133 - 153
adding data to media, 148
copying media, 122
labeling media, 145
media condition, 148
media life cycle, 134
media pools, 48, 135
media rotation policies, 143
media allocation policies, 147
media copies, 122
pre-allocation policies, 147
vaulting, 151
media management after backing up,
151
media management before backing up,
145
media management concepts, 48
Media Management Daemon, 220
Media Management Database
location, 191
records, 190
size and growth, 191
media management during backing up,
147
media management functionality, 48,
133
media pool properties
append incrementals only, 136
appendable, 136
media allocation policy, 136
media pool usage examples, 140
large library configuration, 141
multiple devices/multiple pools, 143
multiple devices/single pool, 142
one device/one pool, 140
media pools, 48, 49
definition, 135
usage examples, 136
media recognition, 163
media rotation policies, 143
media rotation policy (definition), 143
Media Session Managers, 235
media set
definition, 105
selection algorithm, 126
media usage, 134
media usage policies, 148
appendable, 148
appendable of incrementals only, 148
examples, 148
non-appendable, 148
media vaulting, 134
Microsoft Cluster Server, 79
migrating to another media type, 118
mirroring objects, 119
miscellaneous information, 331
mixed environment, 66
MMD, 220
MMDB location
Media Management Database, 191
MMDB records
Media Management Database, 190
MMDB size and growth
Media Management Database, 191
MMDB.
See Media Management Database
MoM, 46
Concepts guide
415
monitoring, 38, 210, 211
mount prompt handling, 111
mount requests, 224, 232, 235
automating, 224
notification, 224
responding, 224, 228
mount requests (restore sessions), 228
MSM, 235
multiple cells, 45, 62
multiple devices, 155
multiple slots, 163
N
NDMP Media Agent, 165
network environment, 39
node
cluster, 80
primary, 81
secondary, 81
notification, 38
number of concurrent sessions
media management, 236
object consolidation, 234
object copy, 231
number of buffers, 159
number of cells, 62
considerations, 62
number of concurrent sessions
backup, 222
restore, 227
O
object distribution to media, 71
object consolidation sessions
mount requests, 235
object consolidation sessions, 233
queuing, 234
object copy sessions
mount requests, 232
object copy sessions, 229
queuing, 232
416
object copy tasks, 116
object copying, 113
object mirroring, 119
omniclus command, 90
online backup of databases, 239
online database backup
archive log backup, split mirror, 267
archive log backup, snapshot, 278
snapshot backup, 278
split mirror backup, 267
online integrations, 240
online reporting, 213
operator user group, 185
optimizing the load on Cell Managers,
224
overview
backup, 38
Direct Backup, 243
Disaster Recovery, 128
IDB management, 196
restore, 39
snapshot backup, 275
split mirror backup, 265
synthetic backup, 257
Volume Shadow Copy service, 290
ownership, 79
backup sessions, 78
restore sessions, 78
P
package, 81
parallel restore vs standard restore, 228
parallel restores, 228
parallelism, 69
physical device collision, 160
planning performance
cache memory, 72
compression, 72
planning security
cells, 74
planning cells, 62 - 67
Cell Managers, 64
Installation Servers, 64
number of cells, 62
planning performance, 67 - 73
backup types, 71
compression, 68
devices, 68
direct backups, 68
disk fragmentation, 72
disk performance, 72
fibre channel, 73
hardware compression, 70
infrastructure, 67
load balancing, 70
local backups, 67
network backups, 67
parallelism, 69
software compression, 70
planning security, 73 - 76
data encoding, 76
Data Protector user accounts, 74
Data Protector user groups, 75
visibility of backed up data, 76
point-to-point topology, 173
post-backup object copying, 114
post-backup media copying, 123
post-exec commands, 239
post-exec commands, 223
post-exec scripts, 104
pre-exec commands, 239
pre-exec and post-exec scripts, 223
pre-exec commands, 223
pre-exec scripts, 104
predefined user groups, 184, 185
preparing a backup strategy plan, 60
preparing media, 134
preventing collision, 160
primary node, 81
processes, 219
backup, 38
Backup Session Manager, 221
restore, 39
Restore Session Managers, 226
properties of media pools, 136
protection types
catalog, 100
data, 99
purging
file versions, 196
filenames, 196
Q
queuing
object consolidation sessions, 234
object copy sessions, 232
restore sessions, 227
R
RAID
snapshot backup, 276
split mirror backup, 269
Raima Database Server, 220
RDS, 220
real time alerts, 209
real time alerts, 208
recovery, 129
disaster recovery, 129
recycling media, 134
related documentation, 23
remote cells, 66
replica
snapshot backup, 276
split mirror backup, 265
replica set
snapshot backup, 279
split mirror backup, 268
replica set rotation
snapshot backup, 279
split mirror backup, 268
Concepts guide
417
reporting, 38, 211
reporting and notification, 311, 329
broadcasts, 207
e-mail, 207
examples, 212
HTML, 207
SNMP, 207
requirements
Direct backup, 249
response time, 208
restore policies, 125
end users, 128
restore by query, 312, 330
restore chain, 96
restore duration, 125
factors affecting, 125
parallel restore, 125
restore options, 311
restore overview, 39
restore policies
operators, 127
Restore Session Managers, 226
restore sessions, 44, 78, 225 - 229
definition, 225
mount requests, 228
queuing, 227
timeout, 227
restores, 125, 225
complete filesystem restore, 312, 330
configuring, 70
duration, 125
end users
end-user user group, 128
IDB operation, 195
media location priority, 126
operators, 127
optimizing, 107
parallel, 228
restore by query, 312, 330
selecting devices, 126
selecting media, 126
vaulting, 153
Volume Shadow Copy service, 294
418
restoring data, 125 - 128
restoring from media in a vault, 153
retiring media, 134
robotics, 177
RSM, 226
S
SAN
See Storage Area Networks
scheduled object copying, 114
scheduled backup sessions, 221
scheduled backups, 105
scheduled media copying, 123
scheduling
backup configuration, 106
scheduling policies, 105, 107
scheduling policy examples, 108
scheduling tips and tricks, 106
scripts
post-exec, 104
pre-exec, 104
pre-exec and post-exec, 223
secondary node, 81
security
data encoding, 183
definition, 73
unauthorized access of data, 183
user groups, 183
user-related, 183
visibility of backed up data, 183
security features, 74
segment size, 157
segments, 237
selecting backup objects, 103
selecting media for backup, 147
Serverless Integrations Binary Files, 194
data, 194
location, 194
size and growth, 194
Service Management, 37, 205 - 215
Application Response Measurement,
207
monitor, 210
notification, 211
operative analyses of trends, 206
overview, 205
reporting, 211
service management applications, 206
HP Performance Agent, 206
Service Management examples, 214
service monitoring, 210
services, 219
Session Messages Binary Files, 193
Session Messages Binary Files
location, 193
records, 193
size and growth, 193
sessions
backup, 43, 220
media management, 235
object consolidation, 233
object copy, 229
restore, 44, 225
setting catalog protection
usage of logging level and catalog
protection, 201
setting up backup environment
IDB management, 196
setting up Data Protector (overview), 55
shadow copy, 290
shadow copy set, 290
shadow copy provider, 290
shared disks, 80
sharing devices in SAN, 175
drives, 177
robotics, 177
sharing libraries, 47, 162, 163, 164
SIBF data
Serverless Integrations Binary Files,
194
SIBF location
Serverless Integrations Binary Files,
194
SIBF size and growth
Serverless Integrations Binary Files,
194
silo libraries, 162
single file restore, 229
size
libraries, 162
size and growth for CDB Records other
than filenames
Catalog Database, 192
slot range, 162
slots, 162
smart media copying, 124
SMBF location
Session Messages Binary Files, 193
SMBF records
Session Messages Binary Files, 193
SMBF size and growth
Session Messages Binary Files, 193
SMBF.
See Session Messages Binary Files
snapclones, 280
snapshot backup
instant recovery, 279
overview, 275
RAID, 276
replica set, 279
ZDB to disk, 278
ZDB to disk+tape, 278
Concepts guide
419
snapshot backup, 275
application client, 277
archive log backup, 278
backup client, 277
backup client as failover server, 286
concepts, 276
configuration, Campus Cluster with
LVM Mirroring, 286
configuration, multiple application
hosts - single backup host, 283
configuration, multiple disk arrays dual host, 282
configuration, single disk array - dual
host, 281
configuration, disk arrays - single
host, 284
configuration, LVM mirroring, 285
configuration, other, 286
configurations, 281
high availability, 275
online database backup, 278
replica, 276
replica set rotation, 279
source volume, 276
target volume, 276
ZDB to tape, 278
snapshot configurations, 281
Campus Cluster with LVM Mirroring,
286
disk arrays - single host, 284
LVM mirroring, 285
multiple application hosts - single
backup host, 283
multiple disk arrays - dual host, 282
other, 286
single disk array - dual host, 281
snapshots
types of, 279
snapshots with the preallocation of disk
space, 279
snapshots without the preallocation of
disk space, 280
SNMP, 207
420
software compression, 70
solutions for backup scenarios, 303, 317
source volume
snapshot backup, 276
split mirror backup, 265
split mirror backup
archive log backup, 267
overview, 265
RAID, 269
source volume, 265
ZDB to disk, 268
ZDB to tape, 268
split mirror configurations
other configurations, 273
remote mirror, 270
split mirror backup
application client, 266
backup client as failover server, 268
backup client, 267
concepts, 265
configuration, local mirror-single
host, 270
configuration, local/remote mirror,
272
configuration, other, 273
configuration, remote mirror, 270
configuration, local mirror-dual host,
269
configurations, 269
high availability, 267
instant recovery, 267
online database backup, 267
replica, 265
replica set, 268
replica set rotation, 268
target volume, 266
ZDB to disk+tape, 268
split mirror configurations, 269
local mirror-dual host, 269
local mirror-single host, 270
local/remote mirror, 272
stacker devices, 161
staggering full backups, 107
standalone devices, 161
standalone devices, 160
standalone file device, 255
standard restore vs parallel restore, 228
standard backup vs disk discovery, 225
static drives, 180
Storage Area Networks
any-to-any connectivity, 171
LAN-free backups, 175
Storage Area Networks, 170 - 181
concepts, 171
device sharing, 175
device sharing in clusters, 180
Direct Library Access, 179
Fibre Channel, 172
Fibre Channel topologies, 173
Indirect Library Access, 178
LAN-free backups, 177
lock names, 177
sharing devices, 175
storage duration of backed up data, 99
- 102
storage virtualization, 275
StorageTek/ACSLS, 162
Subscriber's Choice, HP, 34
supported configurations for Direct
Backup , 250
switched topology, 174
synthetic backup, 257
benefits, 258
operation, 258
media space consumption, 260
restore, 260
synthetic full backup, 257
systems to be backed up, 42
systems with backup devices, 42
T
Target System, 129
target volume
split mirror backup, 266
technical support
HP, 33
technical support
service locator website, 34
timeout, 224
timeout (restore sessions), 227
transaction logs, 238
transactions, 208
types of incremental backups, 93
leveled incremental backups, 94
U
unattended operation, 37, 110, 161
usage of logging level and catalog
protection, 201
specifics for large cells, 203
specifics for small cells, 202
using different logging levels in the
same cell, 202
usage of logging level and catalog
protection
setting catalog protection, 201
usage of media pools, 136
user groups, 184
admin, 185
end-user, 185
operator, 185
predefined, 184, 185
user interfaces, 42, 50
Data Protector GUI, 50
Data Protector Java GUI, 52
user rights, 184, 185
user-related security, 183
users, 184
users and user groups, 183 - 185
tablespaces, 237
TapeAlert support, 154
target volume
snapshot backup, 276
Concepts guide
421
V
Z
vaulting, 134, 151 - 153, 311, 329
definition, 151
restoring, 153
restoring from a vault, 312, 330
vaulting usage example, 152
Veritas Cluster, 79
virtual cluster nodes, 84, 86, 89
virtual full backup, 258
virtual server, 81
visibility of backed up data, 76, 183
Volume Shadow Copy service (VSS)
shadow copy set, 290
Volume Shadow Copy service (VSS)
backup, 293
overview, 289
writer, 290
Volume Shadow Copy service (VSS)
backup model, 291
benefits, 293
filesystem backup, 293
filesystem backup and restore, 295
integration with Data Protector , 293
restore, 294
shadow copy, 290
shadow copy provider, 290
VSS
See Volume Shadow Copy service
VSS backup, 293
VSS backup model, 291
ZDB to disk+tape
snapshot backup, 278
split mirror backup, 268
ZDB to disk
snapshot backup, 278
split mirror backup, 268
ZDB to tape
snapshot backup, 278
split mirror backup, 268
Zero Downtime Backup
snapshot backup, 275
Zero Downtime Backup
split mirror backup, 265
W
websites
HP Subscriber's Choice for Business,
34
HP , 34
product manuals, 23
Windows domains, 65
Windows workgroups, 66
writer, 290
Writer Metadata Document (WMD), 294
422