jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Serge Huber <shub...@jahia.com>
Subject Re: PersitenceManager and Filesystem summary
Date Fri, 18 Mar 2005 14:58:52 GMT
Edgar Poce wrote:

> IMHO ...
>
> FileSystems
>
> - LocalFileSystem
> . Slow on window boxes
> . non transactional
> . If the jvm process is killed the repository might turn inconsistent
> . easy to configure
>
> - CQFS file system
> . fast
> . non transactional
> . if the jvm process is killed the repository might turn inconsistent
> . easy to configure
> . license issue
>
> - VFS
> . not an option
> . I think the vfs team will add transaction support. If this happens, 
> a cqfs provider for vfs will be an interesting option.
>
> -----
> PersistenceManagers
>
> ObjectPersistenceManager
> . simple
> . Not human readable
> . An inconsistency is hard to fix without a tool
> . easy to configure
> . Write operations are synchronized
>
> XMLPersistenceManager
> . not so simple but human readable
> . easy to configure
> . Write operations are synchronized
>
> OJBPersistenceManager
> . Unnecessary complexity
> . transactional
> . rdbm referencial integrity
> . not so easy to configure.
> . Multithreaded friendly. Write operations don't need to be synchronized.
>
> HibernatePersistenceManager
> . Need some work, the JCR-63 issue is closed but the problems are 
> still there.
> . not so easy to configure

I'm currently working on "porting" your OJB changes to the 
HibernatePersistanceManager, by also using the store(ChangeLog). But I'm 
still seeing quite a lot of test failures. Maybe I should put the code 
somewhere in case somebody wants to help out ?

Also I would add to your list of features the persistence managers that 
are cluster friendly. Shared filesystems are not trivial to setup 
properly, whereas share databases are quite trivial. Somehow I don't 
like the idea of a content repository relying solely on one machine :)

Regards,
  Serge...

Mime
View raw message