On Wed, Sep 23, 2009 at 9:57 AM, Brian Aker <[email protected]> wrote: > Hi! > > So why did you run the tests with such a low concurrency? > > On Sep 23, 2009, at 9:38 AM, MARK CALLAGHAN wrote: > >> For the non-Drizzle tests, about 66% of the overhead in mysqld is from >> executing the query, about 10% from parsing and about 12% from >> pre-execution (between parse and execution -- table locking, access >> control). > > Did you test the full life cycle? Connect/disconnect?
connect run 1M statements disconnect The intent for the test is to measure statement processing overhead. I have yet to look at connect/disconnect efficiency. That is very interesting, but a different project. > >> * I reran the tests to use utf8 (5138-innodb-plugin-utf8) to confirm >> that the only cause for the performance difference is not utf8 > > One note about this, MySQL currently only supports 3 byte UTF-8, we support > 4. > >> * tests for drizzle used the default value -- pool_of_threads_size=8 >> (pool8 below), but changing it to 16 (pool16 below) did not improve >> performance much > > Did you confirm you were using the pool scheduler? We have found, depending > on the test, that the 1=1 scheduler will still turn out better in some > cases. How do I determine that? SHOW VARIABLES reports 'pool_of_threads_size=8' as a default value. Nothing is reported that matches '%sched%'. Do I get the 1:1 scheduler when I figure out how to disable pool of threads? I suspect that 1:1 will provide better performance for this workload. -- Mark Callaghan [email protected] _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
