When I did some performance testing on Cassandra 0.7.6, I was getting 10,000 - 20,000 inserts per second on a *single *Cassandra node, on real hardware (a consumer desktop PC with 4 GB RAM). Cassandra has got substantially faster since then. I was inserting 1KB columns each on a new row, if I remember right, using multiple clients on localhost (each in its own process).
So unless your values are much larger than 1KB, you should be able to get *much *greater write throughput out of your 4-node cluster. Several folks have pointed out issues - you need a physical machine per node, and you need multiple clients in order to exploit the concurrency in Cassandra (even when testing a single node). Ideally more RAM per node would be good too. On 22 January 2012 13:10, Gustavo Gustavo <dotnetdev.java...@gmail.com>wrote: > > > Cassandra x 4 > 2012-01-21 11:45:38,044 #6 [Logger.Log] INFO - >> Inserted 10000 > positions for 1000 vehicles (10000000 inserts): > 2012-01-21 11:45:38,082 #6 [Logger.Log] INFO - >> Total Time: > 2:37:03,359 > 2012-01-21 11:45:38,085 #6 [Logger.Log] INFO - >> Throughput: > 1061 inserts/s > > And for MySQL x 2 > 2012-01-21 14:26:25,197 #6 [Logger.Log] INFO - >> Inserted 10000 > positions for 1000 vehicles (10000000 inserts): > 2012-01-21 14:26:25,250 #6 [Logger.Log] INFO - >> Total Time: > 2:06:25,914 > 2012-01-21 14:26:25,263 #6 [Logger.Log] INFO - >> Throughput: > 1318 inserts/s > > Is there something that I'm missing here? Is this excepted? Or the problem > is somewhere else and that's hard to say looking at this description? > > Cheers, > Gustavo > >
