Seems like our performance issues with cassandra where mostly due to different settings for rpc and query timeouts.