[horde] Horde Implementation Going South
Dave Cunningham
dcunningham at meccorp.mec.edu
Wed Sep 5 23:21:19 UTC 2007
I am not familiar with the statistics you refere to or how they would relate to Joins. Can you expand on this?
I am not sure of the difference between describe and explain.
I was initially running session in MYSQL. But, that was SOOO slow that I immediately disabled it, went back to php sessions and put them in /dev/shm.
Dave
--------
Is "DESCRIBE SELECT ..." the same as "EXPLAIN SELECT ..."?
EXPLAIN is the tool I have used to see how MySQL will process a query.
I recommend running ANALYZE TABLE to gather updated statistics on your
Horde tables. The table statistics can get stale as your usage ramps up.
I run the following command nightly:
mysqlcheck --all-databases --analyze --verbose
If your statistics are stale, MySQL may be doing JOINs inefficiently,
which is a big part of the datatree usage.
Are you storing sessions in MySQL as well? The session table can be a
huge burden on MySQL. A *lot* of data is stored in the session with Horde
3.x. I've moved sessions off into memcache (running on the same server as
MySQL) and it has been a big performance improvement.
Even still, MySQL is working pretty hard. I hope the improvements in
Horde 3.2 will help.
Andy
More information about the horde
mailing list