Where is the bottleneck? See http://spyced.blogspot.com/2010/01/linux-performance-basics.html
Recently we have upgraded our Cassandra servers to version 0.7.6-2 from 0.6.6
We are experiencing a severe performance issues – we must restart the Cassandra servers every half hour (otherwise our apps servers can't connect to the Cassandra servers).
Each Cassandra server has 2 SSD drives (one for data and one for commit log),
The amount of the read per second for each server is 500.
We haven't experienced those issues in 0.6.6
Do you have any ideas?
Co-Founder & CTO