|
Archives of the TeradataForumMessage Posted: Wed, 12 Oct 2016 @ 20:23:05 GMT
We tried to use OUC feature but it generated quite a bit of background noise (performance issues). I believe for our type of workload, the issues will be resolved in 15.10 (I am looking forward to use Stats Manager). For now, we still continue to use our old proven process. I have seen few sites follow some rules such as? - IF you have a metadata that maintains ETL interaction with the base table, you could use and calculate changed data in terms of DEL/INS/UPD and then refresh stats - Or you could use perm size monitoring (for this, you have to take daily snapshot of sizes), and then define rules around % growth in table size - not most efficient but it works - Lastly, you should ensure that you should have days logic (table stats older than 'N' number of days will be qualified and so forth) We have a daily as well as a weekend job. We typically allow fewer stats collected during weekdays and we relax some rules to refresh additional stats over the weekend (by changing % change in perm size and stale days logic). Of course, for all above to work, there should be statistics defined on base tables. Also, we regularly collect stats on smaller tables, dictionary, VOSI and NUSI's. There is more than one way to reach Rome and this could be one of them. HTH! Best, Vinay Bagare
| ||||||||||||||||||||||||||||||||||||||||||||||||
https: | ||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||
Copyright 2016 - All Rights Reserved | ||||||||||||||||||||||||||||||||||||||||||||||||
Last Modified: 24 Jul 2020 | ||||||||||||||||||||||||||||||||||||||||||||||||