"Mindaugas" <ml@kilimas.com> writes:
> I execute simple query "select * from bigtable where From='something'".
> Query returns like 1000 rows and takes 5++ seconds to complete.
As you pointed out that's not terribly slow for 1000 random accesses. It
sounds like your drive has nearly 5ms seek time which is pretty common.
What exactly is your goal? Do you need this query to respond in under a
specific limit? What limit? Do you need to be able to execute many instances
of this query in less than 5s * the number of executions? Or do you have more
complex queries that you're really worried about?
I do have an idea of how to improve Postgres for this case but it has to wait
until we're done with 8.3 and the tree opens for 8.4.
> Ideas for improvement? Greenplum or EnterpriseDB? Or I forgot something
> from PostgreSQL features.
Both Greenplum and EnterpriseDB have products in this space which let you
break the query up over several servers but at least in EnterpriseDB's case
it's targeted towards running complex queries which take longer than this to
run. I doubt you would see much benefit for a 5s query after the overhead of
sending parts of the query out to different machines and then reassembling the
results. If your real concern is with more complex queries they may make sense
though. It's also possible that paying someone to come look at your database
will find other ways to speed it up.
--
Gregory Stark
EnterpriseDB http://www.enterprisedb.com
Get trained by Bruce Momjian - ask me about EnterpriseDB's PostgreSQL training!