Any easy way to correlate IOPS with virtual disks?

[JerkyChew]

Management has given me a list of VMs and asked me to generate a report of how many IOPS each machine's virtual hard drives has been using. However, I can't find an easy way in the interface to do this. I can get VM Disk IOPS Counters, but those only point to the hba instance. We have a medium sized ESX farm with MetaLUNs on an EMC SAN, and I'm having a difficult time taking the vmhbax:blah and corroborating it with the VM's C drive, D drive etc. Is there a way to get this information from the nworks MP? Thanks in advance.

[chrisdearden]

you should be able to get this from the reporting pack?

[vBPav]

Ahh, interesting request. Are you asking if we are able to show the following?:

VMGuest - vDISK0 = vmhba:1:0:1 = OSGuest Partitions; C:

Likewise, you may have:

VMGuest - vDISK0 = vmhba:1:0:1 = OSGuest Partitions; C:, D:

[Alec King]

vCenter does not supply the disk trafffic such as IOPS for every VM partitioned drive (such as C:, D: etc)

From the vCenter API we can only obtain the VM traffic to it's datastore file (vmdk) via the Host vmhbax:x:x:x. If there are multiple guest OS disks partitioned in the VM datastore file, then vCenter simply aggregates their traffic.
In fact, even if one VM has multiple vmdk on the same datastore, then they are also aggregated by vCenter. This is the VM's vmhbax:blah metric seen in the MP (and I agree, it is a bit 'blah' and not user-friendly, but that's VMware's fault )

In vCenter 4.1 and later VMware added the virtual disk capability to the API metrics, so we will be able to add each virtual disk (vmdk) as a separate instance for I/O in our next update. Meaning that if a VM has 2 vmdk files, even on the same datastore, we will be able to separate the traffic for them. However vCenter still does not supply I/O metrics per VM guest partitioned disk inside the virtual disk. So that's a feature request for VMware

For the moment, you would have to check how your disks are configured for each VM - for every Datastore that a VM is using, there will be a vmhbax:x:x:x metric showing the I/O. This will combine the metrics for all partitioned guest disks on that datastore, and you can report in the MP on that.

Hope that helps!

Cheers,
Alec