Re: optimizer_ ???

  • From: Wolfgang Breitling <breitliw@xxxxxxxxxxxxx>
  • To: Robyn <robyn.sands@xxxxxxxxx>
  • Date: Fri, 22 Apr 2005 15:46:59 -0600


I would say that the use of the system statistics this way is valid. 
Those are the real numbers Oracle is seeing while you collect the system 
statistics. You want to make sure the numbers are consistent before 
drawing conclusions (what you did) and not a one-time aberration because 
of special circumstances. But you want to do that anyways and not let 
the optimizer make decisions based on an erratic spike.

What EMC system did you have and what did you swap it for? If you don't 
mind me asking. You can e-mail me privately if you don't want to blast 
that information to the world.

Robyn wrote:

> Wolfgang,
> I don't know if this is an 'acceptable' use of system stats, but I had 
> gathered them on several different systems and noted that singleblock 
> and multiblock read times were consistently and noticeably slower on our 
> EMC systems.  I'd been arguing that our performance issues were related 
> to the durations of the reads, so I turned the numbers over to my boss 
> and the systems guys. When I got back to work, I realized I hadn't 
> gathered new system stats after the recent storage upgrade and I 
> recollected the numbers. The post SAN swap numbers did get much better 
> which I think proved the original point:


Wolfgang Breitling
Centrex Consulting Corporation

Other related posts: