RE: segment monitoring, stats, histograms

 Wolfgang has his reasons for not liking this approach, and I have to agree they are very significant and make sense. Basically it's a "if it's not broke, then don't fix it" approach. Sorta falls in with my love of KISS (Keep It Simple Stupid). But I'm sorry to say that I don't follow Wolfgang's beliefs in practice, mainly because with a high OLTP environment things get broken often enough to drive you crazy. Therefore Yes, I do monitor high transactional tables & gather the stale statistics, if needed, every morning. Keeps the machine well oiled. Also, I only gather histograms on indexed columns and do a full compute, stuff that sample size thing. Since that has been put in place here the 7am calls of a slow database have disappeared completely. They were a weekly occurrence before.

I would highly recommend evaluating both approaches at your location. Mine works here and Wolfgang's has been proven not to. Where Wolfgang is working his approach is working & I'd venture that mine would break things. Each site, each database instance, may function differently & consequently require a different approach. I see it as a part of the DBA job to apply the approach that works best in the respective environment. This list is a sounding board of the approaches that have been found to work, where they are implemented. That way at least you don't start with a blank piece of paper.

Dick Goulet
Senior Oracle DBA
Oracle Certified 8i DBA
-----Original Message-----
From: Christian Antognini [mailto:Christian.Antognini_at_trivadis.com]=20 Sent: Thursday, February 03, 2005 7:59 PM To: breitliw_at_centrexcc.com
Cc: oracle-l_at_freelists.org; jeremiah_at_ora-600.net Subject: RE: segment monitoring, stats, histograms

>In my opinion and experience, NO to all questions

Yeah, I know... on this topic I'll probably never share the same opinion =3D
with you!

--
http://www.freelists.org/webpage/oracle-l
--
http://www.freelists.org/webpage/oracle-l