Archives of the TeradataForum
Message Posted: Wed, 17 Jan 2001 @ 22:37:27 GMT
On my last project, we implemented a fairly robust implementation of Priority Scheduler on two different production systems. Both implementations were pretty successful. We created multiple partitions and used numerous performance groups. We also took advantage of the ABSolute and RELative policies, which significantly impact overall workload performance. These policies give you a powerful tool to control certain types of workloads by limiting the resources available.
A couple of suggestions:
1) Make sure you are at least at PDE 3.03.34 and and PTDBMS 3.03.66. There are a couple of critical patches in these point releases.
2) If creating multiple partitions, make sure that the Default Partition is weighted highest. Most system work is done in this partition.
3) Keep performance group wieghtings in ascending order, just like the default partition.
4) You can dynamically change your entire priority scheduler configuration on the fly through the use if UNIX scripts and the schmon command. This is great for changing your priority schemes based on time of day, day of week, system events, etc.
There was supposed to be an article in the February issue of Teradata Review describing this Priority Scheduler implementation that was co-authored by myself and my customer, but it looks like the article will get bumped to the summer issue.
Hope this helps.
Thomas F. Stanek
|Copyright 2016 - All Rights Reserved|
|Last Modified: 23 Jun 2019|