2 Replies Latest reply on Jul 20, 2012 2:33 AM by satoyukitsukano

    How Infinispan MapReduce process multi-threads?

    satoyukitsukano
    satoyukitsukano Jul 15, 2012 11:37 PM

      Hi!

       

      I want to use Infinispan MapReduce.

      I am running Infinispan MapReduce in distribution mode(running multi processes).

      It is fast, but only single thread(Mapper, Reducer) runs on one process.

       

      How Infinispan MapReduce process multi-threads?

      I see [1][2]. But I don't know.

      For example, in Hadoop, multi threads run on one process.

       

      [1] https://docs.jboss.org/author/display/ISPN/Infinispan+Distributed+Execution+Framework

      [2] http://planet.jboss.org/post/introducing_distributed_execution_and_mapreduce_framework

       

       

      Best regards,

      Satoyuki Tsukano

      Acroquest Technology Co., Ltd.

        • 1. Re: How Infinispan MapReduce process multi-threads?
          vblagojevic
          vblagojevic Jul 17, 2012 1:08 PM (in response to satoyukitsukano)

          Hi Satoyuki,

           

          Each map function runs in a single thread per Infinispan node! Currently we do reduction on a master node in a single thread as well! However, starting with Infinispan 5.2 we will introduce distributed reduction phase - reduce phase per intermediate output key will be run on the cluster of Infinispan nodes.

           

          We plan to introduce multi-thread map and multi-thread reduce function execution in upcoming releases.

           

          Regards,

          Vladimir

          1 of 1 people found this helpful
            • Actions
          • 2. Re: How Infinispan MapReduce process multi-threads?
            satoyukitsukano
            satoyukitsukano Jul 20, 2012 2:33 AM (in response to vblagojevic)

            Hi Vladimir,

             

            Thank you for your response.

            I undertand current function.

             

            I hope that Infinispan MapReduce will grow up more useful!

             

            Regards,

            Satoyuki

              • Actions