VM vs Host vs Storage system reported response time

[asyntax]

I am seeing some strage behavior that I can't quite pin down.

2008 R2 VM (current tools), ESX 4.1 (491557), FC Storage (Compellent, or NetApp)

My Windows Server is running Exchagne 2010, has multiple disk presented to it split across multiple LSI SAS controllers. Under high workload I see response time repored in Windows to be upwards of 1 second. But in the VSphere Client and on the Storge system I see a resposne time of 50ms.

The storage and VSphere are reporting about the same response time but Windows is way out of wack.

Any ideas on this?

[RyanPead]

I'm getting a similar issue, just that my vSphere client is reporting alarms for the VM's aswell...
"Total disk latency"

Also "SolarWinds Storage Response Time Monitor" is reporting the same latency...

Currently using Compellent Series 40 over 10Gb iSCSI with jumbo frames enabled at every point in the chain.

VM's are spread evenly across multiple datastores. With a 1:1 ratio relationship [IOW 1 datastore to 1 LUN]

The guest OS VMDK's [c:\] and data [e:\ & f:\ etc] are on seperate datastores

Any advice?