UpdateStats runs a long time

Resolving The Problem

1. Tune the following file:

TWS v8.4:

<TWShome>/appserver/profiles/twsprofile/config/cells/DefaultNode/nodes/DefaultNode/servers/server1/server.xml

TWS v8.5.x:

<TWAhome>/eWAS/profiles/twaprofile/config/cells/DefaultNode/server.xml

TWS v8.6:

.../TWA/eWAS/profiles/TIPProfile/config/cells/TIPCell/nodes/TIPNode/servers/server1/server.xml

- Increase the initialHeapSize to 256

- Increase the maximumHeapSize according to the number of TWS jobs or jobstreams in your environment. (512 for 0-4000 objects, 1024 for about 10000, 2048 if higher)

The WebSphere Application server will need to be restarted for this to become effective.

(stopWas.sh / startWas.sh or conman stopappserver / startappserver)

2. For TWS v8.4 working with DB2 v9.1 only, create the following index in the TWS database as the DB2 user:

CREATE INDEX "MDL"."IDX_JOB_HISTORY_RU"
ON "MDL"."JHR_JOB_HISTORY_RUNS"
("JOD_ID" ASC,
"JHR_RUN_DATE" ASC )
PCTFREE 10
ALLOW REVERSE SCANS;
TERMINATE;

The created index can improve performance when the Production Plan has multiple instances of the same run for the same job, for example, when there are many jobs that utilize the "EVERY" option in the same plan.

This is resolved by APAR IZ66103 in TWS versions 8.5 and above.

3. Check to see if there are any very old jobstreams that have been carrying forward. If there are old jobstreams that do not need to still be part of the production plan, cancel them so they will not carry forward (and will be removed) the next time the Final jobstream runs. (See Technote # 1419631 in the "Related information" section below for further details about this.)

4. DB2 maintenance:

UpdateStats (logman) is using Java to process the job and jobstream statistics from DB2.

- Run the dbrunstats.sh script (before Jnextplan)

You can get more information about running dbrunstats in the TWS v8.6.01 Administration Guide:

http://www.ibm.com/support/knowledgecenter/SSGSPN_8.6.0.1/com.ibm.tivoli.itws.doc_8.6.0.1/awsadmaintmanualrun.htm?lang=en

This is the maintenance to run manually on DB2 that could help during the logman phase. From the earlier (8.5.x) Administration Guide:

"This routine has also been made available for you to run manually in the case either where you feel that the performance of DB2 has degraded, or because you have just added a large amount of data, and anticipate performance problems. The routine is imbedded in a tool called dbrunstats, which can be run to improve performance while DB2 is processing data without causing any interruption."