Hello!

I'm testing a DAG failover in my test lab (Exch1 and Exch2 = both CAS and MB, test database is being replicated from Exch1 to Exch2, CAS roles are fault-tolerant by using DNS round-robin) and would like to have a couple of questions clarified:

1) When only one server is online - be it Exch1 in case Exch2 has not been started yet or Exch2 when Exch1 is turned off simulating failover - the mailflow becomes extremely slow: I can send/receive mail both in OWA and Outlook 2013 but mail is being RECEIVED (NOT SENT!) very slowly. Navigating between various ECP panes becomes very slow as well.

Q1: Should there be any performance degradation when one of the servers is not working (I think we can cut the extra time needed for reconnecting from Exch1 to Exch2 out of equation)?

Q2: Suppose the following situation: Exch1 hosts the active db copy, Exch2 hosts the passive one. Both servers are turned off. I turn on Exch2 only. Should in this case failover succeed?

Thank you in advance,

Michael

Q1: Unless you're heavily loaded, there should not.

Q2: That isn't a failover, technically, it's a restart.  When Exch2 comes back up, it should be able to find the file share witness and mount databases, I believe.  But I would never do it that way.

Hi,

Great advice from Ed.
Heres an blog about Exchange 2010 High Availability Misconceptions Addressed, for your reference:
http://blogs.technet.com/b/exchange/archive/2011/05/31/exchange-2010-high-availability-misconceptions-addressed.aspx

Thanks