> Aaron, whenever I get a GCInspector event log, will it means that I'm having > a GC pause? Messages about ParNew are GS pauses that went over 200ms. CMS GC has to relatively small pauses during it's cycle.
All ParNew GC pauses the application threads, Cassandra is just logging the ones that take over 200ms. If you want to see them all enable the GC logging in /etc/cassanda/cassandra-env.sh Cheers ----------------- Aaron Morton Freelance Cassandra Consultant New Zealand @aaronmorton http://www.thelastpickle.com On 18/05/2013, at 10:29 AM, Sorin Manolache <sor...@gmail.com> wrote: > On 2013-05-16 02:58, Bill Hastings wrote: >> Hi All >> >> I am doing very small inserts into Cassandra in the range of say 64 >> bytes. I use a C++ Thrift client and seem consistently get latencies >> anywhere between 35-45 ms. Could some one please advise as to what >> might be happening? > > Sniff the network traffic in order to check whether you use the same > connection or you open a new connection for each new insert. > > Also check if the client does a set_keyspace (or "use keyspace") before every > insert. That would be wasteful too. > > In the worst case, the client would perform an authentication too. > > Inspect timestamps of the network packets in the capture file in order to > determine which part takes too long: the connection phase? The > authentication? The interval between sending the request and getting the > response? > > I do something similar (C++ Thrift, small inserts of roughly the same size as > you) and I get response times of 100ms for the first request when opening the > connection, authentifying, and setting the keyspace. But subsequent requests > on the same connection have response times in the range of 8-11ms. > > Sorin >
