We are moving from Exchange 2007 to Exchange 2010. We are planning a DAG across our East and West datacentres with 4 mailbox servers (2 in each site) and 4 CAS/HT servers (2 in each site, in a CAS array in an NLB cluster). Most of our users are outside of the two datacentres, they are on WAN links that are well connected to both East and West. We will have clients connecting to OutlookWest.Company.com (which is the DNS name assigned to the VIP for the NLB) if their mailbox is on a server in the West datacentre. With Exchange 2007, we could configure a client (Outlook 2007) to point to any mailbox server, and when a client then connects, then Outlook would re-configure itself to the mailbox server where the user's mailbox existed. This has been useful because we use imaging to deploy Windows 7 with a E2007 Mailbox server name pre-configured. Now to my question; If we have a datacentre failure (e.g. "West"), then clients that are configured to use "OutlookWest.Company.com" will need to connect to the CAS array "OutlookEast.Company.com" so that they can connect to the DAG member that has got the secondary copy of their mailstore mounted as active. 1) Will this happen automatically? If there is a replica of their mailstore in the other datacentre, and this is brought online, will the client be informed that they need to change CAS array too? Most of our users are not in a datacentre. Will we need to manually change the DNS entry for the failed CAS array to point to the IP of the working CAS array? 2) For clients connecting for the first time, will they need to know which CAS array to connect to? If they connect to the wrong one, will the DAG attempt to do something stupid like mount the passive mailstore for that user? Can we leave our image pre-configured with just one of the 2 CAS arrays, or will we need to create a DNS alias over the top (e.g. Outlook.Company.com)? 3) For external (OutlookAnywhere) clients, will they be able to be re-configured to connect to the correct CAS array? Thanks for any useful advice or links

Basically all information you need to answer your questions is here: Designing a Site Resilient Exchange 2010 Solution (Part 1) Designing a Site Resilient Exchange 2010 Solution (Part 2) Designing a Site Resilient Exchange 2010 Solution (Part 3) and here: Datacenter SwitchoversJesper Bernle | Blog: http://xchangeserver.wordpress.com

If you are using Outlook 2007 or higher, then you should let autodiscover do its thing. That will ensure the client points to the correct hosts names. There should be no need to set Outlook with a host name as part of your image because that is what autodiscover is designed to do for you. If you aren't using an external load balancer, or WNLB with separate CAS servers then you will have to make a manual DNS change to send the traffic to the other site. You will not have to change the CAS array address, because the clients will simply connect to the CAS server you point the DNS host to. Basically everything is done by autodiscover and DNS. You don't have to touch the client configuration at all. The only key thing to have is the CAS array in place before you add users to the Exchange 2010 mailboxes, so they are using the CAS array host from the start. The CAS array host name would be something unique - outlook.example.local is the usual example. It can point to a CAS server directly. Simon.Simon Butler, Exchange MVP Blog | Exchange Resources | In the UK? Hire Me.