Have you seen this before? There is a Microsoft support article over at:

http://support.microsoft.com/kb/898468

Increasing I come across this problem. Why? Well, I figured that out.

In the phyiscal world this scenario is unlikely unless it really is a DoS. However, if two VMs on the same phyiscal server transfer a high volume of files (some batch process for example) these symptoms present themself.

Technically it's an advantage - it means VMware vSwitching is so effiecient it is generating errors in Windows meant to protect the user from high volumes of I/O.

So, what do we do? We put HA affinity rules to keeps these VMs together (where possible), then increase the registry value discussed in the Microsoft article to a much higher number (don't go nutes, increase slowly over time until the problem goes away).

Enjoy!

Dave