Re: KAIO

Root (I think I know him!) writes: >Whats the real story on KAIO and HP 10.20. Most posts list better >performance with KAIO. When post report decreased performance it >usually is fixed by correcting the configuration. Is there a split >camp on KAIO?? I personally did a benchmark on a large bank customer's installation where turning on KAIO increased average response time by 10%. They were running an HP, with around 600 concurrent users. I heard from an Informix tech who personally did his own BM and found the performance breakpoint was at about 3,000 (!) concurrent (simulated with Tuxedo(?)) users, after which KAIO significantly hurt performance. Up until that point it helped. One way KAIO can hurt performance is when it is no longer asynchronous. This can happen on some platforms when there exists a large number of IO operations being requested. The nice thing about this is that it is usually tunable. On HP there exists an environment variable called IFMX_HPKAIO_NUM_REQ, by default this value is set at 1000. If you are doing a lot of I/O this might not be high enough. On DEC I believe it need to be tuned in the kernel and for other platforms you will have to find out what the threshold is and how to tune it. (This from one of the Informix source code gods.) Be sure to read the release notes and machine notes. They contain information and warnings about KAIO on HP. (Don't get scared away by the warnings, just read and understand them.) The only way to know *for you* *for sure* is to try it yourself. The good news is that it is really easy to turn on and off. Be sure you have reliable, factual, numerical baseline data and that you can get comparable data after the change. If it's not in defensible, reproducible numbers then it's not science. ___________________________________________________________ Clem Akins (aka clem@informix.com) Informix Software, Inc (Standard disclaimers apply) International Technical Support Last seen: Motorcycling through rain and snow in California