CODEWITHSUNDEEP
cassandrathrifttitanastyanaxrexster
Remis Haroon - رامز
Remis Haroon - رامز

Reputation: 3572

TITAN : Cassandrathrift vs Astyanax performance

I am using a TITAN-0.4.3, REXSTER 2.4 over Cassandra(6 node cluster) & Elasticsearch. My use case requires to bulk upload vertices and edges into the graph at a time. Presently the rate at which adding new vertex+edge is very slow ( ~45 / sec ), and the targeted is ~1000 vertices / sec. Presently I am using Astyanax connecter to cassandra ( which has load balancing and failover detection capabilities). Is there any other advantages or dis-advantages for Astyanax over Thrift (Cassandrathrift) connecter or vice-versa?

Please advise!

Astyanax

    <graph>
        ...
        <properties>
            <storage.backend>cassandra</storage.backend>
            ...

Cassandrathrift

    <graph>
        ...
        <properties>
            <storage.backend>cassandrathrift</storage.backend>
            ...

Upvotes: 0

Views: 561

Answers (1)

stephen mallette
stephen mallette

Reputation: 46206

There are some important differences between these two settings. You touched on the first set of differences by mentioning "load balancing and failover detection" when using cassandra. The other set of differences relate to how each keeps data in memory and how it is sent over the wire. For bulk loading scenarios, I think that you will find a generally better experience with cassandra as your setting, as bulk-loading under cassandrathrift might hit GC issues depending on your loading strategy. Of course, this is just a generality and somewhat dependent on your loading strategy. I'd suggest that you simply try both settings given your load and see what works best.

Upvotes: 2

Related Questions