Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have some questions below -- tried to word them the best I could. If you downvote my post, please make a comment explaining why you downvoted so that I can try to do better when I post in the future. Here are my questions (my big question is the second paragraph):

If we were to use Server 2012, would we be able to get High Availability (HA) and reliability without a SAN in a massive virtualized environment running VDIs and RDS?

We have high transaction applications with Microsoft SQL such as Exchange, SharePoint, etcetera needing HA and reliability -- is there another way to get it without using a SAN? (EDIT: Or I guess what I am trying to say is -- can we eliminate a SAN and still have HA and reliability for Microsoft SQL and Exchange? Does Hyper-V have anything to do with this in 2012?)

My supervisor had heard that with Server 2012 you can do something like that, and connect two servers together and achieve both HA and reliability -- is this true? How does it work? Thanks!

share|improve this question
up vote 5 down vote accepted

Yes, you can achieve high availability for all the services you listed without using a shared disk solution.

  • Exchange 2010: Database Availability Groups (minimum three servers where one is just a file share, remember to load-balance client requests properly.)
  • SQL 2012: AlywasOn availability groups. SQL 2008 R2 and downwards: Mirroring/log shipping
  • VHD files: Separate fileservers with DFS-R replication and a DFS namespace set up

Read up on those subjects, and you'll know if you want to use them, or if you need clustering to provide fault tolerance. All the techniques I listed above requires either manual intervention or uses a period of time to transition.

Hyper-V 2012 introduces live migrations and replicas between hosts using only regular TCP/IP. It is, however, NOT clustering. You will still get downtime when something breaks, as you would have to start up your replica.

Also - keep in mind that clustered services cannot be live migrated, as they will instantly BSOD when they get a pause in disk I/O. Exchange DAG and SQL AlwaysOn uses clustering to keep things in sync without shared storage.

share|improve this answer

BlueToast, what pauska is saying is that you can't have HA from your virtualization layer if you don't have shared storage. You can have HA at your application layer, in many cases. So instead of looking to Hyper-V to provide HA, look at the software within the VMs.

With that said, Server 2012 can use a file server as the necessary shared storage, as long at that file server is using SMB 3.0 (which comes with Server 2012.) To get true HA, you would need the file server to be implemented as file server cluster. You can do that with Server 2012, as long as the storage attached to the file server cluster is symmetrically attached. While this might make you wonder whether you need a SAN in the file server just to avoid a SAN attached to Hyper-V, it's not the case. You can use much cheaper shared SAS JBODs.

If you create that file server cluster with shared SAS, then you'll be able to use HA at the Hyper-V layer and you might then choose to forgo setting up HA at the application layer. Which solution makes the most sense to you should depend on the situation. Doing things at the application layer usually involves a richer set of choices, but forces you to use a different solution for every application. Doing things at the virtualization layer tends to allow you to use the same strategy for many applications, but with a somewhat more opaque view of what's going on in the VMs.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.