Considerations for HA and BC/DR for Virtualized Exchange 2010

Microsoft updated their virtualization guidance for Exchange 2010 right around the time of TechEd, and I have started to get more and more questions from customers around virtualizing Exchange 2010 based on some of the solutions we have been creating. If you haven’t already seen the Microsoft whitepaper you can find it here:

There was also a presentation that was done by Jeff Mealiffe and Jim Lucey from the Exchange Product Group here. that delves into the highlights of the whitepaper.

With the increasing popularity of virtualizing Exchange 2010 a number of questions usually arise around High Avaliability and BC/DR options (aka Site Resiliency) and how they are effected by Virtualization. The good news is very little and with technologies like Hyper-V Live Migration, VM/HA, VMWare’s Site Recovery Manager you will have additional options that can provide other things to think about. Since these technologies can now have coexistence with DAG one must consider how or if these other options come into play.

Let’s start with local HA options. The use of Database Availability Groups (DAG) is certainly a popular option when virtualizing Exchange but in order to have a Microsoft supported configuration with DAG’s you need to follow a few rules. First Exchange server virtual machines, including Exchange Mailbox virtual machines that are part of a Database Availability Group (DAG), can be combined with host-based failover clustering and migration technology as long as the virtual machines are configured such that they will not save and restore state on disk when moved or taken offline. So all failover activity must result in a cold start when the virtual machine is activated on the target. Another important point is to make sure the VM’s housing your  active databases are not on the same physical hardware as the VM or VM’s housing the passive copies thus avoiding a single point of failure with the physical hardware. Overall using DAG’s is a virtualized environment are a good option that are supported by Microsoft as long as you follow a few simple rules.

Another option that customers have asked about in virtualized environment for the cost reduction and simplicity is to deploy stand alone mailbox servers and use the built in hypervisor HA mechanisms like VM/HA or Hyper-V Clustering to handle the local HA.  With this option only a single RAID protected copy of the database and logs is deployed locally and can help with the cost of deploying multiple DAG copies. It is important to understand, that Hypervisor based HA is not Application aware HA, but for some customers it can meet the requirements for the solution.

For many customers a single RAID protected copy that is being protected by a software or hardware based VSS backup solution is adequate protection based on the requirements. With this option you are protected from the loss of a disk with the RAID, from a physical server loss with  VM/HA or Hyper-V Clustering and database corruption with a VSS  snapshots. The biggest concern most customers have with this option is patching of the OS or Application, since it will require a reboot of the server. For those that have maintenance windows, this may be less of a concern if you have scheduled periods where you are able to reboot the servers.

When discussing BC/DR/Site Resiliency options, leveraging stretched DAG copies are a great option as well as technologies such as VMWare’s Site Recovery Manager (combined with a storage replication technology such as EMC RecoverPoint or EMC SRDF)  if you are looking to restart entire VMs and data from your entire data center into a DR site, as well as the database portability option built into Exchange (yes, it’s still around!). DB portability has taken a back seat to DAG these days, but it is still a very handy option if your BC/DR solution needs to address many more applications than just Exchange, and it’s a nice method to still consider.

Our customers have also been very interested in a recent EMC Proven Solution that we did to leverage a local, 2-copy DAG for HA and leveraging storage replication capability from EMC RecoverPoint to implement a DR solution in the remote site. This solution was discussed at EMC World 2011 and the whitepaper can be found here: Exchange 2010 DR Options with Cross-Site DAG and EMC RecoverPoint

Thanks to Chris Devine from EMC Strategic Solutions who contributed to this post.

Until next time,

Dustin