Regarding the performance: I have tried to tweak Babudb, but could not get any more out of it than you did. I guess when you put 5M entries at once, there is only so much tweaking one can do. (At least for babudb). It seems odd though, since babudb is a less complicated framework, with less features. We should contact the babudb guys and ask why we see the results that we see.
Regarding keys: Having one entry in the db for each value will give the fastest update speed, but will be slower for a large number of nodes with the same value. Having an entry with all the ids will make a large entry, causing more bytes to be written to disk for every update. On Fri, Jun 11, 2010 at 1:51 PM, Tobias Ivarsson <tobias.ivars...@neotechnology.com> wrote: > On Fri, Jun 11, 2010 at 12:54 PM, Mattias Persson <matt...@neotechnology.com >> wrote: > >> >> BabuDb has a >> >> db.prefixLookup( "key|value|" ) >> >> method so that's the one I'm using. >> > > I wonder how well that scales. I'd like to see some performance figures from > indexing massive volumes and then doing lookup based on prefixLookup. > > -- > Tobias Ivarsson <tobias.ivars...@neotechnology.com> > Hacker, Neo Technology > www.neotechnology.com > Cellphone: +46 706 534857 > _______________________________________________ > Neo4j mailing list > User@lists.neo4j.org > https://lists.neo4j.org/mailman/listinfo/user > _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user