db-derby-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jim Newsham" <jnews...@referentia.com>
Subject RE: tuning/storage questions
Date Fri, 01 Jun 2007 21:38:59 GMT

> -----Original Message-----
> From: Mike Matrigali [mailto:mikem_app@sbcglobal.net]
> I'll answer below, but what I usually say in these cases is that you
> are starting too low.  I would not expect more than a 10% change for
> any of these low level stuff.  I would look more at the queries that
> are going to slow and see if the right indexes exist for these queries.
> Often you can get orders of magnitude changes in this area.
> also you don't say if you are using embedded vs. network.  Returning
> large result sets is likely to go much faster in embedded vs. network.

Appreciated.  I got about 15x speedup in the last few days with some recent
changes in our db access API and some of the client code. :)  Some of these
involved ensuring efficient queries (particularly join order) and using
Derby's statement cache, while others involved eliminating inefficient or
repetitive queries or other poor usage patterns.  I estimate about 3x
speedup in the db layer, and 5x speedup in the client layer.

We are using embedded mode, as the database is just a storage mechanism for
our desktop app.

Thank you for your other replies, they were also helpful.  If I can free up
some time I'll perform some tests with different page sizes, and possibly
different page cache sizes.


View raw message