Comments on: VMware’s Comparison of Storage Protocol Performance http://andyleonard.com/2008/02/08/vmwares-comparison-of-storage-protocol-performance/ qstat -u aleonard -s z Tue, 06 Jan 2009 03:24:33 +0000 http://wordpress.org/?v=2.7 hourly 1 By: Andy http://andyleonard.com/2008/02/08/vmwares-comparison-of-storage-protocol-performance/comment-page-1/#comment-9 Andy Sat, 09 Feb 2008 22:44:35 +0000 http://andyleonard.com/2008/02/08/vmwares-comparison-of-storage-protocol-performance/#comment-9 Thanks for the comment, Dan. Are you able to give more details about the configuration of your NFS systems where you saw poor performance? I wouldn't be surprised if VMware wants to get out of the filesystem business, although I think they're going to be supporting vmfs for a long time to come, given the installed base. If it is the case that VMware is looking to NFS as their way out of developing filesystems, it's odd that we haven't heard much of anything about NFSv4.1 from them. Andy Thanks for the comment, Dan. Are you able to give more details about the configuration of your NFS systems where you saw poor performance?

I wouldn’t be surprised if VMware wants to get out of the filesystem business, although I think they’re going to be supporting vmfs for a long time to come, given the installed base. If it is the case that VMware is looking to NFS as their way out of developing filesystems, it’s odd that we haven’t heard much of anything about NFSv4.1 from them.

Andy

]]> By: Dan http://andyleonard.com/2008/02/08/vmwares-comparison-of-storage-protocol-performance/comment-page-1/#comment-8 Dan Sat, 09 Feb 2008 06:28:28 +0000 http://andyleonard.com/2008/02/08/vmwares-comparison-of-storage-protocol-performance/#comment-8 While I agree with their finding that NFS is just as good for vmware, I don't agree with their testing methods. iometer tests using 100% read and write setting are not a good indicator for vmware performance. Random IO is the key.. We have NFS storage systems that are very fast at sequential IO and poor random IO and the VM's are very slow. Also, 256M VM may have been an issue in the numbers... I believe there is a larger issue going here... I think vmware want's to get out of the filesystem business and HPC NFS systems are their way out. While I agree with their finding that NFS is just as good for vmware, I don’t agree with their testing methods.

iometer tests using 100% read and write setting are not a good indicator for vmware performance. Random IO is the key.. We have NFS storage systems that are very fast at sequential IO and poor random IO and the VM’s are very slow.

Also, 256M VM may have been an issue in the numbers…

I believe there is a larger issue going here… I think vmware want’s to get out of the filesystem business and HPC NFS systems are their way out.

]]>