Re: worker_instances vs worker_cores
Thanks a lot Andrew! Yeah I actually realized that later. I made a silly mistake here. On Mon, Oct 20, 2014 at 6:03 PM, Andrew Ash wrote: > Hi Anny, SPARK_WORKER_INSTANCES is the number of copies of spark workers > running on a single box. If you change the number you change how the > hardware you have is split up (useful for breaking large servers into <32GB > heaps each which perform better) but doesn't change the amount of hardware > you have. Because the hardware's the same, you're not going to see huge > performance improvements unless you were in the huge heap scenario. > > Typically you should configure the parameters so that SPARK_WORKER_CORES * > SPARK_WORKER_INSTANCES = the number of cores on your machine. If you have > an 8 core box, then you should lower SPARK_WORKER_CORES as you raise > SPARK_WORKER_INSTANCES. > > Cheers! > Andrew > > On Mon, Oct 20, 2014 at 3:21 PM, anny9699 wrote: > >> Hi, >> >> I have a question about the worker_instances setting and worker_cores >> setting in aws ec2 cluster. I understand it is a cluster and the default >> setting in the cluster is >> >> *SPARK_WORKER_CORES = 8 >> SPARK_WORKER_INSTANCES = 1* >> >> However after I changed it to >> >> *SPARK_WORKER_CORES = 8 >> SPARK_WORKER_INSTANCES = 8* >> >> Seems the speed doesn't change very much. Could anyone give an explanation >> about this? Maybe more details about work_cores vs worker_instances? >> >> Thanks a lot! >> Anny >> >> >> >> >> -- >> View this message in context: >> http://apache-spark-user-list.1001560.n3.nabble.com/worker-instances-vs-worker-cores-tp16855.html >> Sent from the Apache Spark User List mailing list archive at Nabble.com. >> >> - >> To unsubscribe, e-mail: user-unsubscr...@spark.apache.org >> For additional commands, e-mail: user-h...@spark.apache.org >> >> >
Re: worker_instances vs worker_cores
Hi Anny, SPARK_WORKER_INSTANCES is the number of copies of spark workers running on a single box. If you change the number you change how the hardware you have is split up (useful for breaking large servers into <32GB heaps each which perform better) but doesn't change the amount of hardware you have. Because the hardware's the same, you're not going to see huge performance improvements unless you were in the huge heap scenario. Typically you should configure the parameters so that SPARK_WORKER_CORES * SPARK_WORKER_INSTANCES = the number of cores on your machine. If you have an 8 core box, then you should lower SPARK_WORKER_CORES as you raise SPARK_WORKER_INSTANCES. Cheers! Andrew On Mon, Oct 20, 2014 at 3:21 PM, anny9699 wrote: > Hi, > > I have a question about the worker_instances setting and worker_cores > setting in aws ec2 cluster. I understand it is a cluster and the default > setting in the cluster is > > *SPARK_WORKER_CORES = 8 > SPARK_WORKER_INSTANCES = 1* > > However after I changed it to > > *SPARK_WORKER_CORES = 8 > SPARK_WORKER_INSTANCES = 8* > > Seems the speed doesn't change very much. Could anyone give an explanation > about this? Maybe more details about work_cores vs worker_instances? > > Thanks a lot! > Anny > > > > > -- > View this message in context: > http://apache-spark-user-list.1001560.n3.nabble.com/worker-instances-vs-worker-cores-tp16855.html > Sent from the Apache Spark User List mailing list archive at Nabble.com. > > - > To unsubscribe, e-mail: user-unsubscr...@spark.apache.org > For additional commands, e-mail: user-h...@spark.apache.org > >
worker_instances vs worker_cores
Hi, I have a question about the worker_instances setting and worker_cores setting in aws ec2 cluster. I understand it is a cluster and the default setting in the cluster is *SPARK_WORKER_CORES = 8 SPARK_WORKER_INSTANCES = 1* However after I changed it to *SPARK_WORKER_CORES = 8 SPARK_WORKER_INSTANCES = 8* Seems the speed doesn't change very much. Could anyone give an explanation about this? Maybe more details about work_cores vs worker_instances? Thanks a lot! Anny -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/worker-instances-vs-worker-cores-tp16855.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
