Quite a while back I setup a Hosted Exchange 2010 SP1 deployment in /Hosting mode with just one server. Now things have moved on and we have quite a large number of clients using this service.
We have have gone out and purchased new hardware for the sole purpouse of HA, just in case one bit of the jigsaw fails.
Now what I propose to do is install a second Exchange Server into the organization, setup a DAG and a CAS Array to provide failover / HA. I have installed this multiple times into our test lab and have got all of the configuration bang on I think. There is just one thing that I cannot confirm
Currently all clients are connected via Outlook Anywhere to exchange.globaldomain.com (RPC over HTTPS) or use Outlook Web Access
Autodiscover is working wonderfully, an HTTP redirect to autodiscover.globaldomain.com, it picks up all of their settings, the only thing required on the client domain is a CNAME
Now I have read that when you implement a CAS array, obviously the RPC name will change (will change from EX01.globaldomain.com to exchange.globaldomain.com). I have read that current Outlook clients can simply seem like they are connected but no longer recieve any email. Is there any truth in this ?
Online people say the solution is to simply recreate the Outlook profile and Autodiscover will point the new clients to CASArray.globaldomain.com, however in a hosted environment this is not an option, does anyone know if Autodiscover will repoint the already setup Outlook clients to the CAS Array ? And if not, however get around this problem ?
I am not Joe Hosted, but in normal operations, the CAS array LB VIP, be it NLB or HLB, assumes your address that you have established for autodiscover, OMA, EAS, OWA, EWS, etc. The the LB parcels out the traffic to a CASarray member. Nothing really changes except for the landing spot for the incoming IP. All of that may be out the window if you are doing hosted, but there is my $0.02.
That's a good answer, and very nice. But I thought that the CAS array DNS name in the Outlook profile had to match the actual CAS name otherwise no traffic will pass ?
Maybe that is just me making it up, but I swear I have read that somewhere ?
Not that simple. The CASArray name, eg, casarray.domain.com, is a logical construct that represents the CAS servers in the array. Each CAS has its' own FQDN. Which the clients never really see. So, the CASArray is built from, say, two CAS servers, cas1 and cas2. Internally, the client lands on the CASArray, which hands the traffic off to one of the two CAS servers. At the certificate level, and also NTLM and Kerberos, a domain-joined machine will trust the CAS server because the CAS server presents a identity that matches what the client is expecting. From the outside, the same is true. I like to build the CASarray with members that each have the exact same certificate for this reason. You can also go straight at an individual CAS and get the same performance. In fact, if you configure the OL profile for a single CAS server, the MAPI profile will get re-written with the CASArray name. When you get down to it, the CASArray name must be different from the CAS member servers because DNS returns the CASArray name, not the CAS member, although all three will exist in internal DNS. Externally, the client does not see the CAS servers in DNS, all they see is the CASarray.
|
My Subscription |
My Forums |
Address Book |
Member List |
Search |
FAQ |
Ticket List |
Log Out
Printable Version
All Forums >> [Microsoft Exchange 2010] >> General >> Autodiscover & New CAS Array 
Autodiscover & New CAS Array - 
RE: Autodiscover & New CAS Array - 
New Messages
No New Messages
Hot Topic w/ New Messages
Hot Topic w/o New Messages
Locked w/ New Messages
Locked w/o New Messages
Post New Thread
