Diagnosing backup and restore performance

Use the table to diagnose issues with backup and restore performance. The table includes links to diagnostic tasks.

Table 1. Explanation of the decisions and tasks for backup and restore performance
Step	Question	Diagnostic tasks
1	Do operating system tools show a disk bottleneck on the client or server? For more information, see Identifying disk bottlenecks for IBM Storage Protect servers.	Yes Fix any disk issues. If applicable, check the load on the system for applications that are not related to IBM Storage Protect. See Tuning disk storage for the server. No Collect client and server instrumentation. For more information, see Client instrumentation report. Go to question 2.
2	Does client instrumentation show high file I/O or directory process time?	Yes Go to question 3. No Go to question 4.
3	Do you have many files or a complex directory structure?	Yes Use file system or operating system tools to help improve backup and restore operations. See File space tuning. No Fix any disk issues. If applicable, check the load on the system for applications that are not related to IBM Storage Protect. See Tuning disk storage for the server.
4	Does server instrumentation show slow disk or tape I/O rates?	Yes Fix any disk issues. If applicable, check the load on the system for applications that are not related to IBM Storage Protect. See Tuning disk storage for the server. No Find the session thread for this node in server instrumentation. The thread might include the node name or you can see if the session thread ID is listed in the Activity Log to determine which node is affected. Go to question 5.
5	Does the client show high `Data Verb` time and does the server show high network time for that node’s session thread?	Yes Investigate network speeds and issues and fix any problems. See Tuning network performance. No Go to question 6.
6	Does server instrumentation show that most time is spent in database operations for that session thread?	Yes Investigate database issues, including disk and buffer pool settings. Fix any problems. See Identifying server performance problems. No Go to question 7.
7	Are you deduplicating data from an Oracle or SAP database, and the average extent size is less than 100 KB? Tip: To determine the average extent size, generate data deduplication statistics by using the GENERATE DEDUPSTATS command and view the statistics by using the QUERY DEDUPSTATS command. Based on the output, calculate the average extent size by using the following formula: `Total Protected Data`/(`Compressed Extent Count` + `Uncompressed Extent Count`)	Yes Consider specifying an extent size that is larger than the default value of 50 KB. For more information about changing the extent size, see the MINIMUMEXTENTSIZE parameter section of the REGISTER NODE and UPDATE NODE command documentation. No IBM® Support can help in diagnosing performance problems by requesting traces and other information from the environment. The comprehensive analysis of a performance problem is a billable service that is offered to IBM Storage Protect customers. For more information, see the IBM Software Support Handbook at http://www.ibm.com/support/customercare/sas/f/handbook/home.html. See also Collecting and analyzing data for performance problems.