Possible Duplicate:
Can Exchange 2010 high availability be achieved with only 2 servers?

We just recently migrated to an Exchange 2010 server. Currently all of the roles and mailboxes are installed on 1 server (we are a small company with less than 100 users). I am wanting to use DAG for replication however it seems most set ups for DAG requires at least 3 or 4 total servers. Is there anyway to make this work with just two servers and both of these servers would have all the roles and mailboxes? Maybe there is a better way to do this than DAG? I'm open for suggestions.

The goal here is to have some sort of replicated server so that if there is an issue with our primary Exchange server, another one can be brought up within an hour or so with all current information (not a backup). It doesn't necessarily have to be instantaneous.

  • Some Q's... Are your Exch2k10 hosts physical or virtual? What's your total store size? Sep 6 '12 at 20:01
  • They are virtual. Total store is about 700GB, currently using about 300. Also, I wouldn't consider this a duplicate because I'm not necessarily interested in HA.
    – Keith
    Sep 6 '12 at 21:28

Yes, this is possible, although somewhat tricky; see here for details.

You'll need a load balancer, and each server should have enough storage for a full copy of the mailbox database(s); also, an additional network adapter on both servers to handle replication traffic would be nice. On the software side, you'll need the Enterprise edition of the Windows operating system you're using, because failover clustering is not supported on the Standard edition, and DAG is based on failover clustering.

And yes, you can move your current Exchange environment to a two-servers solution with a DAG and a CAS array without no need to rebuild everything from scratch.

  • Yes, totally agree. Did exactly the same for one of my recent clients.
    – Vick Vega
    Sep 7 '12 at 18:04

Also, I saw this blog come in my mail the other day. I didn't get a chance to really read it in depth but I didn't notice anything regarding a hardware load balancer necessary:


Also, from the blog post itself:

"Please note that this solution does not give you High Availability, but it will provide you with a solution for site and server disaster."

I personally like the answer in Massimo's link better so far, just because it seems a HLB would make everything so much simpler (given of course you have the budget to buy one)

  • The article you linked makes the CAS array name be a DNS alias of one of the Exchange servers; or at least I think so, the explanation is not quite clear. However, if that server fails, the name will need to be manually redirected to the other server, so no automatic failover here. The proper way to do this is to load-balance the CAS servers and make the CAS array point to the balanced IP, but in this scanario you'll need a HLB for this, because Windows' software solution (NLB) is not compatible with failover clustering.
    – Massimo
    Sep 6 '12 at 20:54
  • @Massimo - Oh I agree. Hence the blog post about it not being HA. This is a poor man's way to have DR, not HA. The OP had mentioned it didn't need to be HA, so I thought the blog might be relevant if he can't afford an HLB.
    – TheCleaner
    Sep 6 '12 at 21:09
  • I'll look at this in more detail but this might be the way we'll go. It's not HA but it's something I should be able to get going within an hour or so.
    – Keith
    Sep 6 '12 at 21:20

A DAG on 2 servers is possible, you will need the 2 DAG servers and a file share witness to maintain a quorum. I maintain a 4 node DAG and a 2 node DAG so I know this is possible.

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