[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: ozone & modularity
- To: ozone-users@ozone-db.org
- Subject: Re: ozone & modularity
- From: Andreas Monitzer <a@blackbox.net>
- Date: Tue, 6 Feb 2001 12:28:50 +0100
- Delivered-To: softw7-ozone-db:org-ozone-users@ozone-db.org
On Tuesday, February 6, 2001, at 11:31 AM, Falko Braeutigam wrote:
> > This is somewhat similar to the Lightweight Objects approach discussed in
> > <http://ozone-db.org/ozone-users/11-2000/msg00141.html>. You really should
> > read this mail and the discussion that followed.
> I perfectly agree ;)
Ok, lets redesign again :-)
> However, besides a better, ozone-friendly design there seems to be another
> problem in your test Andreas. What do you mean by "read object from the
> database"? Just one call to the database?
It's the following:
* get iterator of all message proxies
* while iter.hasNext()
* iter.next().getHead()
* convert & store into local list
getHead() returns the object itself, so it's only one RMI call.
btw, I forgot to mention that it's from a remote database (on the same machine), so it's another VM.
I tested it with a local database now, it's .2-.4 sec (-> 30sec total for 95 messages), still too much.
> Then 0.4-0.8s are way to slow
> for a simple database roundtrip. You should get values around 3ms or 0.003s!
Even those 0.05sec for post processing are much too slow, but maybe that's because my profiling code hangs there (total is 30sec with profiling code, 21sec without).
> What ozone version do you use? Please checkout latest CVS.
I checked out last Saturday I think (maybe +/- 1 day).
andy
--
We fucked up. We fucked up big time.
-- Steve Jobs