VMware Cloud Community

ESXi 4.1 -- high disk utilization chart for host object but not on any VMs?

What does it mean when the performance chart for a host (Disk) shows significant disk activity, but the same chart for all VMs on that host doesn't even come close to adding up to the disk utilization shown for the host?

The host in question is a whitebox Intel system (Core 2 Quad Q6600) with 4x750 GB eSATA disks in RAID-5 driven by an Adaptec 3405.   Yes, this is pretty low-end stuff but it manages to run a couple of FreeBSD systems and a couple of Windows systems without any issues.  All disks for all VMs is on the same LUN and are pure VMDKs -- no raw disk mappings, no direct eSATA disk.

Today I ran into a situation when doing some testing of a Windows 2008 R2 / Exchange 2010SP1 VM.  Guest OS responsiveness for this VM seemed to shrink to near zero, but the resource manager & task manager in the guest OS showed almost no disk activity, as did the VMs performance chart for CPU & Disk.  But the performance chart at the host level showed a sustained average of about 7-10 MB/sec with peaks otherwise.

Other VMs didn't really seem to suffer too much, but none of them do much, either.

I've had this VM running for a few months without issues, but I was really puzzled that no single VM could account for the disk I/O that the host was reporting.

0 Kudos
0 Replies