I've been browsing the net hoping to stumble across some ideas or a solution to an issue that I've been noticing on one of my new ESX clusters.
Problem: Drive performance on x86 or x64 Windows Server VM is terribly slow...and erratic. Vms running on older clusters not using NFS volumes on SATA perform much better. Our oldest ESX cluster runs iscsi connected volumes for our datastores (Netapp 3050). Our second cluster uses NFS volumes on FibreChannel drives for hosting datastores (Netapp 3040). The third and problem cluster is running its datastores on NFS volumes using SATA drives (same 3050 as first cluster).
Hardware: 2 x Sunfire x4150 ( 2x QC E5345 Intel Xeon CPUs @ 2.3Ghz, 32GB memory)
ESX Platform: 3.5.0, 82663 with HA and DRS enabled.
Storage Platform: Netapp FAS3050 Cluster running Ontap 7.2.4
I'm unsure if this issue is a VMWare problem or if it's a Netapp problem... I ran a few drive tests on the windows systems on different clusters. The first two clusters perform fairly well and drive tests show consistant performance on par with physical disks. The third problem cluster reports VERY erractic performance, which is also terrible. Attached are disk test reports.
I noticed this issue about a week or so ago...I hadn't noticed such slow performance before that. The systems are just very unresponsive. It takes 10 seconds for the Program menu to show up once I click the start button...something just isn't right.
Any insight or comments are most appreciated! Thanks!
Okay...upon further testing it has been determined to be one of our Netapps. We have a 3050 cluster and the NFS volumes on the second filer head is acting odd. We created an NFS volume and mounted it as an NFS datastore on the same VM cluster. We then created a new VM on the new datastore on the first filer...and it flew like the wind. The disk tests were great.
So it appears to be one of our Netapps...lovely.