Thank you very much for the links. I knew already DavidsModel-Page, but it was a bit theoretical, so I hoped, that I could get some "real-life" experiences of what would be the right approach.
Best regards, Ulrich -----Ursprüngliche Nachricht----- Von: Alexander Klimetschek [mailto:[email protected]] Gesendet: Mittwoch, 17. März 2010 11:33 An: [email protected] Betreff: Re: Workspace - concept Hi, see previous discussions: http://jackrabbit.markmail.org/thread/vmbqjcecolt5ar63 http://jackrabbit.markmail.org/thread/eso35utn3k3572nv http://jackrabbit.markmail.org/thread/idajqo3mwxlnsuyr and David's model, rule #3: http://wiki.apache.org/jackrabbit/DavidsModel#Rule_.233:_Workspaces_are_for_clone.28.29.2C_merge.28.29_and_update.28.29. Regards, Alex On Wed, Mar 17, 2010 at 08:52, Cech. Ulrich <[email protected]> wrote: > Hi to all, > > I just thinking about the best way of using the workspaces in conjunction > with user/rights management for an application using JCR/Jackrabbit as > storage: > 1. Every user gets his/her own workspace > Advantage: all data is clearly separated, the user/rights management must > only deals with workspace-wide constraints (perhaps nearly the node-wide > access management can be left out); no problem with workspace-wide indexing > (I see that the workspace contains the index of the data; if I think correct, > taht would mean, that anyone can search for key words, but some keywords > perhaps should not be seen by all users) > Disadvantage: this can be perhaps many thousends workspaces (I have no idea > at the moment if this can be handled) > > 2. One global workspace, all users connect through this workspace > Advantage: links to other nodes in another tree-part can be established; > easier handling of this one workspace > Disadvantage: can all users search for "keywords" which they should not see? > (if the whole index is shared by all users); more handling of ACL-constraints > on the per-node-basis; very big workspace with big index > > So, I would like to know, if someone had some experiences, which concept to > use in practice with very big data (could be about 2TB in the next 2-5 years) > and many thousend users. > > Thanks in advance, > Ulrich > > -- Alexander Klimetschek [email protected]
