Hi,
I hope I'm posting this question in the correct section.
I've recently encountered an issue my organization is having with our Exchange 2010 HA solution. The solution implemented uses the Active/Active configuration which is best explained in this article:
http://www.msexchange.org/articles_tutorials/exchange-server-2010/high-availability-recovery/designing-site-resilient-exchange-2010-solution-part3.html
The only difference is there are no NLB's in place, nor was a CAS Array configured for the solution. Otherwise there is only one CAS server per site, and two Mailbox servers per site with a two stretched DAG's and both have witness servers configured the opposite
server. The Active and Passive mailbox servers are also clustered.
The setup would look something like
Site A:
US-CAS, US-MB1, US-MB2
Site B:
EU-CAS, EU-MB1, EU-MB2
US-MB1 and EU-MB1 are members of DAG1 and are clustered.
US-MB2 and EU-MB2 are members of DAG2 and are clustered.
DAG1 witness server: US-CAS
DAG2 witness server: EU-CAS
Given the topology, when I perform a failover of a mailbox database from US-MB1 to EU-MB1, the solution works as intended, Outlook clients configured to use US-CAS are redirect to their mailboxes mounted on EU-MB1.
Now.. when the US-CAS is taken offline, what happens with the outlook clients configured to use US-CAS as their mailserver?
I have considered a few solutions, none of which I am sure would be suitable.
1. I could repair all the users Outlook 2010 profiles.. far from ideal.
2. I could redirect the DNS of US-CAS to point to EU-CAS, I have not tried this but I believe it may work in theory?
We recently had to take all exchange servers (US-CAS,US-MB1,US-MB2) offline for maintenance. Just before the shutdown of the CAS/Mailbox servers we initiate a mailbox failover to EU-MB1 and EU-MB2, now all mailbox databases are mount correctly in the new site,
however once they were taken offline the mailbox databases dismounted and would refuse to mount.
Based on what I can find out reading online, the current configuration is not built to handle a complete site failure, and what I would like to find out is:
1. Is there a way of providing HA for a complete site failover given the current configuration
2. What steps I can take to ensure high availability?
3. What should I expect in the event of complete site failure (planned or unplanned), and what the typical procedure would be to follow to ensure availability.
4. What ways can I improve the Exchange HA, CAS Array, NLB's etc?
Thanks
Derek