On Mon, Apr 30, 2012 at 3:32 PM, Ronald Hahn, DOCFOCUS INC.
<rhahn@docfocus.ca> wrote:
> We went to the SQL2k server (On the same hardware) and ran the selects
> again. When bringing back on an int32 PG was faster with the fetch and the
> row coming back in 1-5 ms and SQL2k coming back in 500-700 ms. This tells me
> that the problem is not related to PG index or Disk. When bringing back 400
> bytes PG fetch time would be 1-2 ms but the results would take 2-3 s but
> SQL2k would it bring back in 700-900 ms. Working with 866 bytes, PG fetch
> time is 1-3 ms with the results coming back in 9 - 11 s and SQL2k bringing
> the results back in 2-3 s.
I think the opposite. I'm thinking it's quite probable that it's disk
access the one killing you. Remember, two different database systems
means two different access patterns.
To figure it out, you have to provide a lot more information on your
system and your query. Check out how to post "Slow Query Questions"
[0]. Only after getting all that information the people of the list
will be able to have a clue as to what your problem is.
[0] http://wiki.postgresql.org/wiki/SlowQueryQuestions