5 Replies Latest reply on Jul 19, 2011 6:51 PM by clebert.suconic

    Most performant threading approach for enqueuing to an embedded HornetQ?

    collinvandyck

      I'm experiencing some high latency when send()'ing messages into HornetQ.  It is embedded and uses the in-JVM transport. Furthermore it's configured to use AIO, and all messages are durable.

       

      This web service typically has a load of 200 req/sec, whre each request borrows a producer from a local cache and send()'s string data into HornetQ. Would it be more performant to have a smaller number of threads push the same volume of data into HornetQ?  At lower loads (~100req/sec) I don't see the same periodic high latency, so I was wondering if perhaps the high threadcount is pulling the disk in too many directions at once.