Archives of the TeradataForum
Message Posted: Wed, 16 May 2001 @ 17:12:22 GMT
On the topic of RSS monitoring. I recommend that you keep a record of ResUsageSPMA data at least for each 10 or 15 minute period of each day. Then, plot each day's basic stats (ie. kernel and user cpu, io wait, and MB/sec IO) against the same time period for the same day of the week for the last 6 months or so. Note that IO wait is recorded in the CPUSysOh column.
Comparing the trend in this historical manner, allows you to detect when things diverge from normal. If you wait until you have a performance problem to look at RSS, you will be hard pressed to say if the measurements that you are seeing are good or bad. Often it is more useful to identify what has changed and especially WHEN it happened.
When I have done this in the field before, I implemented it in SAS. Although I think that it could be implemented in SQL with a lot of work, I have always thought it a good idea to not have performance tracking systems in the same environment with the system whose performance is being tracked. In the absence of SAS, your favorite development platform can be used.
Hopefully, it has a direct interface to Teradata.
|Copyright 2016 - All Rights Reserved|
|Last Modified: 28 Jun 2020|