Re: HELP: How much diskspace in /tmp per user

In <1994Jul15.053552.20462@hasler.ascom.ch> nbladt@autelca.ascom.ch (Norbert Bladt) writes: > Hello ! > > After getting my problem solved with the permissions of the Informix > files, I finally could do the performance tests for the three systems > we have. > During that I found that the performance depends on the amount of free > space on the root filesystem - no system did have a /tmp filesystem. > > Now, my questions are: > 1. Where does Informix put, of course, silently its temporary files ? > If it would not create the temporary files and delete them > immediately after the open, I could find out myself :-) > > 2. If it is in /tmp (or wherever it is) how much disk space should be > reserved for every user of the DB ? > Is it useful to have a /tmp-filesystem in this case ? > > I think answer for question #2 will depend on the size of the data one > query will deliver (more or less), but a first estimate would be nice. > I would like to have a guideline how to configure the UNIX system for > best Informix performance. > > A rule of thumb is very much appreciated. > > Thanks in advance for your help, > > Norbert. > P.S. My environment: ISC 4.01, Informix 5.0 with SE. Online might be > different completely, I guess. > > --- > Norbert Bladt, ascom Autelca AG VOICE: +41 31 999 6552 > Payphone Management Software FAX: +41 31 999 6544 > Worbstrasse 201 UUCP: uunet!mcsun!hslrswi!aut!nbladt > CH-3073 Guemligen, Switzerland EMAIL: nbladt@autelca.ascom.ch 1. Aw, that's an easy one. See the other posts. 2. This one's both easy and hard. The easy part: Perform a lengthy analysis of your source code and application in light of your database size and derive a set of statistics for your user's needs. Include: Largest possible request (don't forget sorting) and frequency Largest likely request and frequency Temp file size: min, max, avg, std dev, median and mode, etc Balance this information against the cost of disk space on your platform (remember I/O time, mean time between failure, admin time, not just initial cost) Develop a scale of values graphing the cost of solution against the usage likelyhood. Extend that graph to include the cost of your user's time, find the break even point, and compare that monetary value to your budget. Be sure to factor in expected growth and falling media costs/faster access times. The hard part: Actually doing this at your site in the span of a human lifetime. We all do this intuitively all the time, and I wonder to what degree our performance suffers in proportion to the intuitive/explicit mix. Maybe I should write my own DBA book! :) Good luck, __________________________________________________________________ | Clem Akins Standard Disclaimers Apply | | Reynolds Metals Co. "Climb High, Cave Deep!" | | Muscle Shoals, Alabama USA cwakins@leia.rmc.com | |________________________________________________________________|