Re: 7.22.UC1X2 Performance tuning questions....

Billy, May I first say thanks for your response, We will be installing KAIO tomorrow night as it requires ansyncdsk to be installed and linked into the kernel first. Re numaiovps ... it says in the manual that if NUMAIOVPS is not in the onconfig the the default will be (2 * number of chunks (43chunks) = 86) or 6, whichever is greater. My question is, If this is the default config, Are we under by a long way or over by a long way ? and should we care about the default config for this parameter ? About update statistics... I checked the release notes and under update statistics guidelines it has the following information... 1) Update statistics medium for all columns in a table. 2) Update statistics HIGH for all columns that head an index. 3) Update statistics LOW for all multi-column index columns. Which we have done, I more concerned with ensuring I can implement a update statistics strategy that I can I use for our database withoutaffecting any processing, remembering this is a 24x7 system. To help guide you in your answers our system is about 60% OLTP and 40% DSS the OLTP is about 400 users. I look forward to your ideas, Once again thanks for the help ! Cheers, Tony. ---------- From: Billy Wheeler To: tony edwards; informix-list@rmy.emory.edu Subject: Re: 7.22.UC1X2 Performance tuning questions.... Date: Thursday, 3 April 1997 17:41 On 3 Apr 97 at 2:46, tony edwards wrote: > We don't have KAIO set because processor affinity is not supported on > HP. The two are not dependent on each other - set KAIO on and reduce the no of AIO VPs. > 4) System not using indexes correctly > > - Have run update statistics medium for all tables, update statistics > high on first column of > each index and run update statistics low on all other index fields. I thought "official policy" was USH for first index columns, USM for other index columns and USL for the rest. > 6) If we changed our optcompind to zero would we need to run so many > update stats? Yes. UPDATE STATS must be run early and often (depending on your database activity, of course) Regarding your checkpoints, aren't your log size perhaps too large, otherwise decrease the checkpoint interval. HTH. -- Ciao, Billy Nose to the ground and ear to the grindstone in: Johannesburg "Wow, that's a big word for an eight-year old!!"