jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Philippe Girolami" <philippe.girol...@digiplug.com>
Subject RE: Performance issues of using "object" repository rather than "orm"
Date Wed, 18 May 2005 12:41:14 GMT
Stephan,

> <disclaimer>
> i have to admit that i am biased. i love simplicity and easy
> deployment and
> i am not a huge fan of ORM.
> </disclaimer>
I'm all for simplicity. Whether that means ORM or not depends on the case :)

> Object PM
> ----------------
> pro:
> - very easy (i.e. zero) deployment
> - very simple yet efficient
> - excellent performance on a decent file system
> - no network transport overhead
>
> con:
> - scalability for very large data sets?
Hummm... I'm talking of millions of "nt:file" nodes here. Each file is
linked to one object with a number of attributes some of which will be
references to other nodes.
Because of this I'm thinking of writing our own custom PM if we use ORM as
we will not want to store the binaries in BLOBs !

> - non-transactional
meaning XA or is the Object PM not even atomic when using
versionning/locking ?
Ok, I could look at the source myself to figure it out but if you have the
answer for everyone to share that would be great.

Thanks
Philippe


Mime
View raw message