I have a server that has several hundred table in a few different databases comprising almost a gig of data, all running on a rather old (3.23) version of mysql. I have used the slow query log to identify queries and have optimized the queries significantly. At this point the entries in the slow-query log (with long query time set around at 3 seconds) usually examine between 1k and 10k rows. When I run the query to test them the query time is generally under .1 second.
The server is basically running with a my-small.cnf and I think that most of the rest of performance I can pull out of the server will come from tuning the mysql server variables for table cache and temporary table size. I am concerned that I might have issues with ram usage. With this in mind: 1. Is there some way to use the general query log to test different server configurations with a real world assortment of queries? Perhaps some way to use mysqlslap? Any other suggestions for benchmarking tools? 2. I see entries in the slow query log where the number of row examined does not correlate with EXPLAIN's that I run of the queries on the production server. Is this likely a situation where mysql needs index hints, or could something else be in play? Thank you in advance for any help. RTFM welcomed, just point out what page ;) -- Rob Wultsch
