RE: Re: Optimizing..

In article <6gdhm8$75m$1@news.xmission.com>, "Billy Wheeler" <billy@west.co.za> wrote: > >On 7 Apr 98 at 13:38, jamienash@cim.co.uk wrote: > >> Whats the point in people sendin gin problems they have like the one >> below, when you get dickheads that answer peoples queries saying >> 'Attend a course' or 'read the manual' how does that help someone?. >> The person with the problem may have already tried to fin d the >> answer in the manual and may not understand it, or can't find it. >> Why not give them a decent answer to the question. I thought this >> mailing list was for people to share knowledge and experience, not >> take the piss out of people that know less than us. > >Er no. If people send in questions like "how do I optimise a query >like this (SELECT yadda-yadda...) given table structures like this >(CREATE TABLE foo...) and record volumes like this with indexes like >this on this database engine, dickheads can give very specific >observations. > >What would you say if I said "how do I design a database?" Am I using >Informix? Am I using VTAM? Am I using UNIX text files? Is it a data >warehouse? > >My advice for tuning a SE 4.x database is very different from my >advice for tuning OnLine 7.23. > >Open-ended questions lead to open-ended answers. > >I didn't consider it to be taking the piss, and Abel, my apologies if >that's what you understood my reply to mean. > >And my observations stand. All the information on performance tuning >*I* know could easily be found in the docs for the platform. If you >are the kind of person that does better with structured training, the >courses are excellent. (I know, I give them all the time.) > >I could quite easily have flamed Abel for not providing sufficient >detail on his queries, which is my standard reply to such vague >queries. > >I certainly didn't expect the Spanish Inquisition. > >-- > >/Ciao, >\\Billy >/ >\\Group Managing Director, The West Solutions Group: http://www.west.co.za >/ Chairman, South African Informix User Group: http://www.west.co.za/~saiug >\\ Drivel @ http://www.west.co.za/tasteless/ >/ >\\ "Granted, Mr Wheeler's ideas are stupid and unreasonable, but he does own the company and I >/ think we should go along with him..." >\\________________________________________________________________________________________________ Nobody expects the Spanish Inquisition. Performance tuning recommendations in the docs are *very* basic, and sometimes at odds with experience. One specific example - the number of CPU virtual processors. Docs specifically state you should not configure more CPU VPs into the engine than physical CPUs in the box. Through trial and error, we have found 50% more CPU VPs than physical CPUs provides faster throughput, measured in the sense of "How many minutes does query XXX take to run?". On our 4 CPU V2200 box we get best performance configuring 6 CPU VPs (we use KAIO, don't know if that impacts results.) (yet ;) ) I saw a posting to this effect from someone at hp.com a while back, and tried it. He was correct. Ray Garrison