Checklist for Tivoli Storage Manager server configuration

Evaluate key configuration settings and scheduling that can affect performance for the Tivoli® Storage Manager server.

Question Tasks, characteristics, options, or settings More information
Have you set up server schedules so that critical operations do not interfere with each other? Schedule operations that might otherwise be automatically started:
  • Disable automatic expiration by setting the EXPINTERVAL server option to 0.
  • Configure storage pools so that storage-pool migration, reclamation, and duplicate-identification operations are not automatically started.
  • Schedule each type of server data maintenance task with controlled start times and durations so that they do not overlap with each other.
Schedule node replication to avoid or minimize overlap with client backups.		 Tuning the schedule for daily operations
Are you running enough processes to handle data operations in your environment? Verify that the number of processes for an operation is enough to complete the workload. For example, if performance for reclamation seems slow, tune the number of parallel processes that are allocated for that operation.
Use the following commands and parameters to control processes for different operations:
  • Storage pool backup processes: MAXPROCESS parameter on the BACKUP STGPOOL command
  • Duplicate identification processes: NUMPROCESS parameter on the IDENTIFY DUPLICATES command.
  • Migration activity: MIGPROCESS parameter on the DEFINE STGPOOL command
  • Parallel expiration activity: RESOURCES parameter on the EXPIRE INVENTORY command
  • Reclamation processes: RECLAIMPROCESS parameter on the DEFINE STGPOOL command

Continue increasing parallel processes to the point where a resource on the server becomes saturated.

 For more information, see the following topics:
Are client backup schedules configured so that backups are spread across the available time? Schedule client backups in a backup window that is isolated from all data maintenance processes, such as storage-pool migration, reclamation, and duplicate identification processing.

If possible, schedule client backups so that they are not all started at one time. You might not have to stagger schedules if sufficient server resources are available to process all client backups.

Also, if you are using client-side data deduplication and there is commonality in the data that is being backed up, you might not have to stagger schedules.

 Avoiding contention for server resources during client operations
Are server option values updated from the defaults for optimum performance? Set the EXPINTERVAL server option to 0 and schedule inventory expiration processing.

Set the MAXSESSIONS server option to a value no higher than 1000, which is the maximum that was tested in IBM labs. Setting the value higher than what is required for the maximum number of sessions that you expect can unnecessarily consume memory on the server.

 Limits for the server database size and peak client sessions
Have you set up a schedule for your database backups?

Have you configured backup operations appropriately for the size of your database?

 When you set up a schedule for database backup, you have more control over when server resources are engaged. Schedule database backups to run after both the client backup, and if used, the storage pool backup.

Perform only full database backups, not incremental backups.

For databases over 500 GB, use multistreaming for database backups to improve performance.

Make the archive log directory for the database large enough so that you do not run out of space between database backups and so that only one database backup is required every 24 hours. Under normal conditions, do not back up the database at unscheduled times.

 For more information, see the following topics:
Have you formatted disk storage pool volumes sequentially if they are placed on the same file system? Formatting volumes sequentially helps avoid disk fragmentation and improves sequential read and write performance.

To format multiple sequential disk pool volumes, use the DEFINE VOLUME command and specify a value for the NUMBEROFVOLUMES parameter. Each volume is allocated sequentially to avoid fragmentation.

 Checklist for storage pools on disk