Performance is an important question in my application (even if I chose
Java). Moreover, I don't think I have the option to collect objects for
batch transaction processing.
As for the select tests:
I got roughly the same results for Load() and getOQLQuery(). Probably
because they are all in the same transaction. It wouldn't be a big
problem for me, because I don't always want an object as a result,
therefore I can use select.
Norbert
> Why is everyone so worried about overhead? Of course, there is overhead.
> I am sure that there are improvements that can be made in the code to
> decrease the overhead. I have not seen any tests, in this thread, that
> paint a real live production picture. Castor gives you freedom and ease of
> use and less code to maintain. These are the things that you gain for the
> overhead that you experience.
>
> Adam
>
> -----Original Message-----
> From: Sudhir Bhojwani [mailto:[EMAIL PROTECTED]]
> Sent: Friday, September 07, 2001 4:10 PM
> To: [EMAIL PROTECTED]
> Subject: Re: [castor-dev] JDO performance tests
>
> any results for select kind of queries......
>
> Batch operations are always faster, even with direct
> JDBC calls you can see the difference. But still its not a good indicative
> of the Castor overheads.Any estimates on that.
>
> Thanks !
> Sudhir
>
> -----Original Message-----
> From: Neal Sanche [mailto:[EMAIL PROTECTED]]
> Sent: Friday, September 07, 2001 1:38 PM
> To: [EMAIL PROTECTED]
> Subject: Re: [castor-dev] JDO performance tests
>
> I found the largest time savings seemed to be when the records created were
> batched in the transaction. So don't commit the transaction until at least
> 500 records are created and you'll see dramatic improvements.
>
> -Neal
>
> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]
> Sent: Friday, September 07, 2001 4:31 PM
> To: [EMAIL PROTECTED]
> Subject: [castor-dev] JDO performance tests
>
> I've created a simple object such as
>
> class Person
> {
> int id;
> String name;
> }
>
> I used the following code to measure the update speed (again, I need to
> test for one transaction on one create())
>
> for(int c=0;c<MAX_OBJECTS;c++)
> {
> person = new Person();
> person.id = c;
> person.name = "name-" + new Integer(c).toString();
> db.begin();
> db.create( person );
> db.commit();
> }
>
> The speed was roughly 6 objects / sec.
>
> Then I run the following code:
>
> for(int c = 0;c < MAX_OBJECTS; c++)
> {
> execute("insert into test (id,value) values (" + c + ", 'name-" + c
> + "')");
> execute("commit");
> }
>
> where the method <execute> simply executes an sql Statement with the
> argument string. This test could give me 180 row inserts /sec.
>
> I have also made tests for large, complex objects. It turned out that
> larger objects don't need proportionally longer time.
> Did anyone get other results?
>
> Thanks in advance,
> Norbert
>
> > > I am using the pre-compiled, downloadable castor-0.9.3 version. I've
> > > made some tests with JDO on small object updates / queries / deletes
> > > testing single transaction speed. Comparing the results with the
> > > same amount of data transfer directly using JDBC (therefore
> > > eliminating > the object<->table mapping overhead and simply
> > > executing insert / > delete / select SQL commands), I've measured
> > > about 30 times more > performance.
> >
> > Would you provide more details about the tests you ran and your
> > results?
> >
> > Thanks
-----------------------------------------------------------
If you wish to unsubscribe from this mailing, send mail to
[EMAIL PROTECTED] with a subject of:
unsubscribe castor-dev
