Deploying Exchange can be a challenge. Particularly when you are all ready to validate your servers & storage with Jetstress and you realize that even though we suggest that you should always run Jetstress prior to going into production, you discover that we don’t support running Jetstress in a virtual machine on that fancy new virtual platform you just deployed. Ouch. Now what?
First, some background. You might be wondering why we don’t support running Jetstress in a virtual machine. The reason is actually quite straightforward. Over the years as we have worked with customers and partners who were either deploying new hardware for Exchange or validating Exchange storage solutions in the Exchange Solution Reviewed Program (ESRP), we saw a number of examples of Jetstress test results where the reported IO latency numbers were wildly inaccurate. Given the lack of trust in the reported performance metrics, we had to ensure that Jetstress was not run in this configuration. This resulted in the guidance that customers deploying on virtual infrastructure should validate storage performance by running Jetstress in the root rather than in a guest virtual machine. While this was a feasible workaround with Hyper-V, it’s not a realistic solution for other hypervisors.
Just as the Exchange product has matured, the hypervisor products that some of our customers use to manage their Exchange infrastructure have matured as well, and we decided that the time had come to do some new testing and see if those strange performance results of the past would come to haunt us again. After weeks of automated testing with multiple hypervisors and well over 100 individual Jetstress tests completed in various configurations, we’ve reached a conclusion…
Effective immediately, we support running the Microsoft Exchange Server Jetstress 2010 tool in virtual guest instances which are deployed on one of the following hypervisors:
- Microsoft Windows Server 2008 R2 (or newer)
- Microsoft Hyper-V Server 2008 R2 (or newer)
- VMware ESX 4.1 (or newer)
Additionally, we are removing the restriction in the ESRP v3.0 program on using virtual machines, so from this point on our storage partners will be able to submit ESRP solutions for Exchange Server 2010 where the validation testing was performed on a virtual machine.
As a reminder, the best place to learn about supportability for Exchange Server 2010 virtualization is on TechNet in the Hardware Virtualization section of the System Requirements topic. Additionally, we have published a Best Practices for Virtualizing Exchange Server 2010 with Windows Server 2008 R2 Hyper-V whitepaper that contains many helpful deployment recommendations. The best resource for understanding how to properly use Jetstress for storage and solution validation is theJetstress Field Guide, which has been recently updated to include this change to our support for guest virtual machines.
I hope this is good news for some of you and that this will result in simpler, easier, and more thorough pre-production validation of your Exchange deployments.
0 comments:
Post a Comment