-
Website
http://phildawes.net/blog/ -
Original page
http://www.phildawes.net/blog/2004/10/14/optimising-mysql-tables-for-rdf-store/ -
Subscribe
All Comments -
Community
-
Top Commenters
-
danja
4 comments · 4 points
-
phildawes
5 comments · 1 points
-
Dominic Sayers
1 comment · 1 points
-
David W.
2 comments · 1 points
-
ryantm
1 comment · 1 points
-
-
Popular Threads
-
Phil Dawes Stuff >> Idea for a global interpreter lock optimized for low contention
2 weeks ago · 3 comments
-
Phil Dawes Stuff >> Idea for a global interpreter lock optimized for low contention
In some ways using MySQL would have been the way to go, since it provides all the transaction support, remote APIs, etc, that we had to build into Kowari. On the other hand, we could never get it to load data very quickly, even without extensive indexing. I've also found that joining a table onto itself numerous times starts to make most relational databases slow down (including MySQL). Tracing a transitive predicate is an extreme example of this.
Basically It would be great to get all the MySQL features for free, but having a datastore that is purpose built for triples (or quads) seems to be the only way to get really high performance on both loads and queries. Depends on how big you need to scale I suppose.
Out of curiosity, have you considered different models in your system? Are you doing this with different "databases", tables, or something else?