Re: CRIT: RBLIVE statspack report

On Fri, 18 May 2001 15:43:19 +0100, "andrew_webby at hotmail" <spam_at_no.thanks.com> wrote:

>Seeing as it was my idea, here's a statspack report for anyone who fancies
>looking over it. I'm prepared to accept any criticism on it. Except any that
>suggest I collect my P45 ;-)
>

Darn, you've done most of the work already! :-)

I'd agree completely that KCADATA is a "hot" area. Majority of reads seem to be coming from there. And waits. Split it would be a priority, I'd say. Maybe as a first approach, move tables/indexes off it based on size? As in, the larger move out first. Or if you are patient enough, use V$BH as you say to find out the hot ones and move them off.

I'd say you will also benefit from KEEP/RECYCLE/DEFAULT. I've managed to convert a <80% hit ratio database to 99.99/97/97 respectively by just enabling them and moving tables to each class. (how to make RECYCLE hit 97? Easy, make it larger than recommended and put the occasional table scans and "temp" tables there!).

RLBs are weird!? Why would optimal be set so high? I'm not aware of any recommendation for such. If anyone can provide info on that I'd appreciate it.

And with the SQL I agree too, however:
Could the cursor_sharing=force have side effects? Like stuffing up something else?

Cheers
Nuno Souto
nsouto_at_bigpond.net.au.nospam
http://www.users.bigpond.net.au/the_Den/index.html Received on Sat May 19 2001 - 06:37:27 CDT