jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefan Guggisberg <stefan.guggisb...@gmail.com>
Subject Re: the 1mn test: first jackrabbit performance and scalability tests
Date Wed, 10 Nov 2004 14:39:54 GMT
On Wed, 10 Nov 2004 12:31:08 +0100, Ugo Cei <ugo@apache.org> wrote:
> Il giorno 10/nov/04, alle 12:14, Stefan Guggisberg ha scritto:
> 
> > here's the (not too sophisticated ;) brute force test code:
> 
> Thanks. I see that there are three persistence managers currently
> available: InMem, Object and XML. Can you give us some data on how they
> perform with the code you pasted?
> 
> 

ok, i did a couple of quick test runs, using the above code (adding
1000 child nodes and calling save() afterwards, invoked in a loop) :

1. InMem, not surprisingly ;), yielded the best results; ca. 3 sec./1000 nodes
2. Object with CQFileSystem scored 2nd: ca. 5-6 sec./1000 nodes
3. Object with LocalFileSystem, not surprisingly either, showed the worst
    performance: ca. 30 sec./1000 nodes

there was no significant difference between Object and XML since this test
is 'write only'; Object will be significantly faster than XML when reading 
operations are taken into account also (because Object has no parsing 
overhead for reading operations).

i also tested a couple of b-tree based persistence managers, they scored
somewhere between InMem and Object w/CQFileSystem

because InMem obviously should only be used in testing environments
i would currently recommend using Object with CQFileSystem for best
performance.

cheers
stefan

> 
>        Ugo
> 
> --
> Ugo Cei - http://beblogging.com/
> 
> 
>

Mime
View raw message