ManagingDB2 RecoveryData Sets with SMS 71
directly to tape.
Backup
The archivelogs are a backup of the active logs. DB2 can create dualarchive
logs. There is no need for an additional backupof the archivelogs.

7.4.1 Storage Class

Storage Class SCDBARCH isan example of a Storage Class for archive logs.
This Storage Class has high availability and good performance.
• SCDBARCH

7.4.2 Management Class

Twodifferent Management Classes are used for the archivelogs. One is used for
the primary copy and the other forthe secondar y copy.Both allow migration of
the data sets.The reason for defining two separate Management Classes is to
enable a physicalseparation of the two copies.
The Management Class MCDBICM is usedfor the image copies retained longest
and fort hearc hivelogs. This ensures equivalent expiration dates for image
copies and archive logs.
The Management Class MCDBLV2is used for the secondary archives. This will
directly migratethe secondary copy to level2 of DFSMShsm and so ensure a
physicals eparationof the two archive copies.
• MCDBICM, used for primary archive log data sets
• MCDBLV2, used for secondary archivelog data sets

7.4.3 Storage Group

Primary and secondary archivelogs are allocated on volumes of the SGDBARCH
Storage Group.These data sets are migrated independently on differentdates.
This is determined by their Management Class.
• SGDBARCH
An alternative to the aboveS torageGroup could be a TMM Storage Group, but
only forthe secondar y copyof the archive logs. A TMM Storage Group simulates
a tape device on disk. Multiple data sets are placed together on the same tape.
This could havea performance impact if this archive log is required for a recovery
or a restart.

7.4.4 ACS Example

An exampleof ACS routines to allocate these SMS classes and Storage Group
for archivelogs is shown in Appendix B, section B.2, “Archive Logs” on page 191.
7.5 Image Copies
Image copies are thebackup of user and system data in a DB2 subsystem. For a
well managedbackup and recovery policy, the amount of data in image copy data
sets exceeds theamount of production data by at least a factor of three. This