Reputation: 35
Let's say I have a Titan cluster with Cassandra as storage backend and an application that talks with RexPro (or the new Gremlin Server). How should I distribute the queries from my applications to the Titan nodes without knowing which node actually holds the data? Is a simple round-robin a good choice here?
Upvotes: 0
Views: 153
Reputation: 46206
On a really large graph, round-robin isn't the best option, but there really aren't other strategies ready for use I'm afraid. Experiments were done with more advanced query routing with reasonable success, but none have made it to production Titan/TinkerPop yet.
Upvotes: 0