More ADSM News from SHARE 84
1995-03-22 18:49:32
Subject: |
More ADSM News from SHARE 84 |
From: |
Melinda Varian <[email protected]> |
Date: |
Wed, 22 Mar 1995 18:49:32 EST |
The following is from the handout for another ADSM session at SHARE 84.
The presenter was Greg Tevis, of ADSM Technical Support.
Session 5071: New ADSM Disaster Recovery
Agenda
o Disaster recovery in ADSM today
o New roles of ADSM recovery utilities
o New disaster recovery functions
o Solutions and recommendations
Note: Support discussed in this presentation is subject to change.
Please consult announcement materials as they become available.
ADSM Database Protection Today
o Database and recovery log mirroring
+ continuous operations on media failure
+ improves read performance
- doubles storage requirements
o Online database dump
+ allows dump while server is online
- database audit required on recovery
o Consistent database dump
+ does not require database audit on recovery
- essentially stops server during dump
o Export of server definitions
- must be used with export of data
ADSM Storage Pool Protection Today
o Export of storage pool data
- extremely slow
- not incremental
o Manual backup of volumes
- out of synch with database
- tedious for tape volumes
- volume management nightmare
o Backup client data twice
- data sent twice across network
- management of offsite volumes (if sent to same server)
New Roles of ADSM Recovery Utilities
o Database and recovery log mirroring
- still great for availability and read performance
- recommended if not using rollforward
- recommended for recovery log if using rollforward
o Database salvage utilities
- still valuable as last resort to save database
o Export
- still useful for information interchange between servers
- can be used for backup of server definitions
o Online database dump
- obsolete
o Consistent database dump
- obsolete
ADSM Database Backup Series
o Full backup plus incrementals
o ADSM keeps internal backup series and volume numbers
o Full backup starts new backup series
o Incrementals limited to 32 to control
- recovery log size
- recovery time
o Used to restore to point in time or to most recent state (with
roll-forward)
ADSM Database Backup
o Full or incremental backups are run
- using BACKUP DB command
- using schedules
- automatically based on recovery log utilization
o Full backups are required at times
- new, restored, or resized database
- 32 incrementals reached
- log mode changed to ROLLFORWARD
- ensure faster recovery
o Full and incremental backups clear recovery log records
ADSM Database Recovery
o No database audit required
o Point in time recovery to any backup
- recovery log records needed
- storage pool volume audits needed for online and changed volumes
o Roll forward recovery to most recent state
- roll forward log mode increases recovery log size
- recovery log records used after last backup
- entire database or single volume can be restored
- no storage pool volume audits required
Storage Pool Backup and Recovery Highlights
o Provides means of protecting storage pool data
o Supports incremental backups of storage pools
o File level granularity enables many functions
o Manages primary and backup copies of files
- automatically uses backup copies of files
- primary copies can be recovered from backup copies
Primary_Pool_Disk + Primary_Pool_Tape = Copy_Pool_Tape ==> Offsite Vault
ADSM Storage Pool Backup
o A primary pool can use multiple copy pools
o Multiple primary pools can use one copy pool
o Only files new to primary storage pool are copied (incremental)
o Copies made automatically or by command
o Copy files not moved when primary files moved
o Reclamation (and Move Data) of offsite volumes copy from primary
pool
ADSM Storage Pool Recovery
o Copy file automatically accessed if primary file in error or volume
unavailable
o RESTORE VOLUME restores files for primary volumes
o RESTORE STGPOOL restores damaged files and all files on volumes
marked DESTROYED
o AUDIT VOLUME and read errors detect damaged files
Recovery from Media Failures -- Lost storage pool volume
o Use UPDATE VOLUME to mark UNAVAILABLE
o Use RESTORE VOLUME (with VOLUMESONLY) to preview which copy volumes
are needed
o Bring needed copy volumes onsite and mark READWRITE
o Issue RESTORE VOLUME command
o Mark copy volumes OFFSITE and return to vault
Recovery from Media Failures -- Lost database volume
o Bring back most recent database backup series
o Allocate database and log space if needed
o Issue DSMSERV RESTORE DB
- uses volume history, devclass and options files to determine
volumes to mount
- can specify point in time
- will automatically roll forward if able
Recovery from Disaster
o Move database and storage pool backups to recovery location
o Restore volume history, device class and options files and
allocate database
o Issue DSMSERV RESTORE DB
o Mark all primary volumes as DESTROYED using UPDATE VOLUMES
(wildcarding allowed)
o Define new volumes to primary storage pools
o Mark all copy volumes as READWRITE (restore services now available
to clients)
o Issue RESTORE STGPOOL for each primary pool
Recommendations
o Back up database and storage pools regularly
o Also back up volume history, device class and options files
o Specify volume history and device class files in options files
o Roll-forward recovery is valid alternative to database mirroring
- perform regular database backups
- continue recovery log mirroring
- no performance impact for roll-forward
o Use asynchronous storage pool backups for most cases
o Perform reclamation on offsite volumes periodically
o Run RESTORE STGPOOL periodically
|
<Prev in Thread] |
Current Thread |
[Next in Thread> |
- More ADSM News from SHARE 84,
Melinda Varian <=
|
|
|