I hate it when people say "I remember I had a problem with xxx but
can't remember what it was."
So, with regrets, I remember I had a problem with dbcp on JDO but
can't remember what it was. We now use c3p0 without any problems.
The issue is that c3p0 is LGPL so we can't distribute it, but it's
available from sourceforge as a maven download.
Anecdotally yours,
Craig
On Feb 4, 2009, at 12:57 PM, Michael Dick wrote:
Including DBCP in the lib directory with OpenJPA seems reasonable. We
already include Derby so adding another Apache binary should be fine.
I think I'd be interested in seeing how OpenJPA + DBCP performs
compared to
say Hibernate + DBCP or EclipseLink + DBCP. As I understand it
Hibernate
includes some rudimentary connection pooling (then again so do we).
If we're
still slower then we'll need to look elsewhere to even the
performance gap.
A lot of the comments on the blog you mentioned focused on whether the
entities were enhanced. If you've tried enhancing the entities do
you still
see a performance gap (without DBCP) ?
-mike
On Wed, Feb 4, 2009 at 2:46 PM, Rick Curtis <[email protected]>
wrote:
Evan -
I looked back over my notes and the inserts showed marginal
improvement(Sorry I don't have hard numbers, I ran this a couple
weeks
back). The largest gains where in the number of selects executed.
Rick
Evan Ireland wrote:
Rick,
If you enable DBCP and run the quoted benchmark using JPA, what
improvement
do you see with the insert throughput?
-----Original Message-----
From: Rick Curtis [mailto:[email protected]]
Sent: Thursday, 5 February 2009 9:21 a.m.
To: [email protected]
Subject: Improving out of the box performance
After reading this
http://terrazadearavaca.blogspot.com/2008/12/jpa-implementations-
comparison.html
blog post , I decided to do some performance(ish) testing on my
local
system
and I also came the conclusion that openJPA isn't the fastest out
of the
box(I don't think that's a surprise to anyone). It was easy to get
something
working, but it didn't work as fast as it could/should have. Ease
of use
is
a high priority when a developer is trying out a new technology,
but not
the
only priority.
A steaming pile that is easy to use, is still a steaming pile. :-)
In an attempt to make openJPA perform better out of the box, I'd
like to
see
DBCP packaged with openJPA and have the docs updated to stress
the use
DBCP
rather than a direct DB connection. This won't add much in the
way of
complexity, but it will help the performance greatly. I wrote a
blog
post
about this very topic that can be found
http://webspherepersistence.blogspot.com/2009/01/jpa-connection-
pooling.html
here .
Any thoughts?
--
View this message in context:
http://n2.nabble.com/Improving-out-of-the-
box-performance-tp2271261p2271261.html
Sent from the OpenJPA Developers mailing list archive at
Nabble.com.
--
View this message in context:
http://n2.nabble.com/Improving-out-of-the-box-performance-tp2271261p2271423.html
Sent from the OpenJPA Developers mailing list archive at Nabble.com.
Craig L Russell
Architect, Sun Java Enterprise System http://db.apache.org/jdo
408 276-5638 mailto:[email protected]
P.S. A good JDO? O, Gasp!
