I've been asked a couple of times about stress testing an Exchange server sitting on a hardware virtualisation platform. What would you have to do differently? Well to be honest I haven't ever done it and a quick trawl of the Internet and internal Microsoft alias' shows that there still isn't much experience out there yet, when considering Exchange Server specifically. Of course the main reason being the supportability situation regarding running Exchange on a hardware virtualization platform. This, and the fact that there isn’t much of a precedent it seems, is what’s putting people off at the moment I think.
However thinking more about it I don't really think you need to do anything very differently. You still need to use a combination of jetstress and loadgen to verify the performance of the server. The only thing you need to do differently is consider the fact that you have multiple servers per host. There is clearly no point in testing a single instance in isolation. You need to run load and test I\O against an equivalent of a full complement of virtual machines that you expect to be supported on a single host. Not a lot of help unfortunately but I’ll blog more as I learn more.
For now it’s worth having a look at this VMWare test of Exchange 2007 running ESX on HP Proliant Servers. Unfortunately not Hyper-V but there you go…
‘Deploying Microsoft Exchange Server 2007 mailbox roles on VMware Infrastructure 3 using HP ProLiant servers and HP StorageWorks’
And some other links that might be useful:
There are some storage pointers here on Scott Schnoll’s blog specific to Hyper-V also…
Just found this if you are working in Hyper-V. Really useful free book 'Understanding Microsoft Virtualization Solutions - From the Desktop to the Datacenter' @ https://www.getvirtualnow.com/usevents/education/download/693371eBook.pdf