The latest plan for migrating from 2010 to 2013 is to create a new 3 member DAG with all the active mailbox databases on the 2 DAG members inside our office and the 3rd DAG member hosting passive copies in a remote data center.

We want to keep the active database copies on the LAN since most email activity during office hours is generated by Outlook clients on the LAN and having all databases except one of them kept local to the LAN should reduce traffic across the WAN due to less database replication happening.

The DAG member in the data center will also do the additional role of CAS server for remote users accessing their mail from the Internet.

If there is a power outage requiring the all servers in our local office including these two DAG members to all be quickly powered down before UPS batteries die, what is going to happen with a 3 member DAG vs 4 member DAG?

I assume that with a 4 member DAG with FSW in the data center, the databases in the DAG should all become active automatically, but with a 3 member DAG, with a power outage in the office containing two DAG members, we would have to access the data center server using an outside internet connection (mobile data plus VPN etc)  and perform some kind of action to manually activate the databases there.   Is that true?  Anything else to consider about this layout?

We have to weigh the costs of adding an additional server to be the 2nd DAG member in the data center increasing the total multi-site DAG from 3 to 4 members and then also increasing the replication traffic on the WAN in the process vs saving that expense and having reduced replication traffic, but having the longer outage, and extra work to manually activate the passive databases in the data center.

• Edited by MyGposts Tuesday, February 17, 2015 5:42 AM

With a 3 member DAG if you have a failure of your Primary Datacenter (the location with 2 nodes) then the DAG will shutdown and you will need to do a datacenter fail over manually.  Remember the DAG will only continue to function as long as it has quorum ( meaning more than 50% of the nodes need to be available) so if you lose 2 out of your 3 nodes, then only 33% of the dag will be functional and it will shutdown.

With a 4 member DAG, the only way to achieve an automatic failure would be putting the file share witness in a 3rd site that both the primary and secondary datacenter can access.  In this setup the third site would act as a tie breaker and if you lost your primary site you would still have 3 active members (including the witness) and the Datacenter failover would happen automatically. 

Here's a good link explaining HA in Exchange 2013.

http://msexchangeguru.com/2013/05/23/e2013-ha-demystified/

Hi,

Agree with Hinte. If you want to achieve that the DAG will failover to DR site automatically, you can place the FSW server in a third site.

With the 3rd site, a failure at the Primary Data Center would allow for an automatic failover to the Secondary Node, as cluster keeps a majority of the voters (Secondary Node and FSW).

For more details, you can look at the following blog.

http://blogs.technet.com/b/exchange/archive/2007/04/25/3402084.aspx

Best regards,