I think this is a very domain specific question. If you data is naturally
very disjoint, then there can be value in having separate databases, and
benefit from increased total database size, easy clustering/partitioning,
etc. But, as you say below, you cannot (or at least not easily) traverse the
total dataset, only within each database. Then if you need the occasional
inter-database traversal, you can create special proxy nodes in each
database representing the inter-database connections, and at application
traverser level write the code for skipping over to the next database when
needed. Obviously this is all very domain- and app-specific, and will not
perform like in-database traversing (obviously :-)
I believe the Neo4j guys are investigating generalized partitioning schemes
for clustering transparently, but I suspect that will take some time to
mature.
On Sun, Apr 11, 2010 at 2:50 PM, wrote:
> The fact that you'd lose all capability for sharing any common
> relationships, tagging, and other benefits, plus the extra resource
> demands, would seem to suggest it doesn't make any sense to use Neo4J in
> a "database per user" manner. Would you do that with any other type of
> data store? Almost certainly not, so I wouldn't do it with Neo.
>
>
> Original Message
> Subject: [Neo] a lot of small embedded databases or one big?
> From: TuX RaceR
> Date: Sun, April 11, 2010 7:00 am
> To: Neo user discussions
>
> Hello Neo4j Users!
>
> I'd like to store tree data attached to users of a web application.
> As each user tree depends only on the user, I have the choice of store
> the user data on a user file.
> (This is some kind of easy sharding)
> Is there some kind of problem in creating many small database files?
>
> Thanks
> TuX
> ___
> Neo mailing list
> User@lists.neo4j.org
> https://lists.neo4j.org/mailman/listinfo/user
>
> ___
> Neo mailing list
> User@lists.neo4j.org
> https://lists.neo4j.org/mailman/listinfo/user
>
___
Neo mailing list
User@lists.neo4j.org
https://lists.neo4j.org/mailman/listinfo/user
