On Dec 2, 2003, at 3:48 PM, Ari Davidow wrote:
We have just put up a new MySQL 3.23.x on a modest Sun V120 with 1GB RAM and a few gig of hard disk space. We're running Solaris 2.8.
Why are you using an old version of MySQL? MySQL4.0 is the recommend production version currently.
We seem to be clueless (I am certainly clueless) about testing various configuration options so that we are reasonably optimizing the resources available to this server. We ran through (approximately) the four configuration files included with the MySQL distro, and then added more memory (we had, after all, 1GB).
We set up a script to repeated parse some common documents with lots of queries. Then we tried the sort of exercise where we'd create temporary tables and copy back and forth. Finally, we ran a load of SELECTs using full-text search (3 explicitly joined tables).
We did not record statistically significant results.
In truth, we have just begun using MySQL, and don't know what our pattern of usage is. Is there an idealized generalized configuration model and a tool or method available to test our server against that ideal? The online reference doesn't say much about optimization, and it is surely messier when we don't really know what we are optimizing for (or how to test, assuming we did know).
Any suggestions? stress testers that have matched up against specific types of configuration?
If you're interested in seeing how it stacks up with a multi-threaded client, or just want to test how efficiently it works with various numbers of threads running, you can use the mysqlsyseval utility, which is part of a tarball available at this site:
http://www.wiley.com/legacy/compbooks/pachev/
You can modify the mysqlsyseval.c source to add different types of queries to test.
- Gabriel
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
