Hi Kelly, I used m1.large instances. I tried both EBS and local storage. I did IO test on the EBS and local devices. It is not bad. I also tested HDFS on these instances. it showed much better aggregate writing throughput. So I guess it is not disk IO problems. Yeah, EC2 disk IO does have fluctuations, but the throughput is not so bad.
Best, Xiaofei On Wed, Nov 23, 2011 at 2:33 PM, Kelly Kane <[email protected]> wrote: > On Wed, Nov 23, 2011 at 12:07, Xiaofei Du <[email protected]> wrote: >> >> I installed Ceph on 10 ec2 instances. > > Can you go into more detail about your EC2 instances and where you are > storing your data? If you are storing them on standard EBS then you > are competing for non-guaranteed bandwidth. The information contained > on the AWS product description page is basically a lie ( > http://aws.amazon.com/ebs/ ) unless things have changed substantially > since I last used them. If you are storing data on the ephemeral > disks then unless you are on a "whole machine" instance > (m1.xlarge/c1.xlarge/m2.4xlarge) you are competing for SATA resources > on the ephemeral disks. If your machine neighbors are doing some heavy > disk workload you may simply be starved for resources. > > Kelly > -- Xiaofei (Gregory) Du Department of Computer Science University of California, Santa Barbara -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
