• Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 3643
  • Last Modified:

Exchange 2010: CAS Array using Round Robin DNS, no NLB

We're currently running Exchange 2003 SP2 and Outlook 2003 in our environment and looking to migrate to Exchange 2010/Outlook 2010.  We're looking to go with a 2 node CAS array for one of our offices.  For the CAS array, someone suggested using round robin DNS only, (no NLB), and in the event that one of the CAS nodes in the array went down, Outlook 2010 clients would be sophisticated enough to automatically reconnect to the other CAS server.  Can someone verify that in a round robin DNS configuration, that Outlook 2010 wouldn't try to connect to a CAS server that is down?  My thinking is that it's hit or miss.  I always thought one of the shortcomings of round robin DNS is the possibility of connecting to a host that is down.  
0
bsohn417
Asked:
bsohn417
1 Solution
 
tigermattCommented:

>> in the event that one of the CAS nodes in the array went down, Outlook 2010 clients would be sophisticated enough to automatically reconnect to the other CAS server

As you suspected, that's not correct. DNS could just as easily return the record for a downed CAS server as that of an active one, and there's no automated detection mechanism to prevent this. Should a CAS go down, you'd have a window of restricted service until someone manually removed the DNS entry for the downed server and the record had expired from caches on downlevel DNS servers/client workstations.

At the bare minimum, use an NLB array with its own virtual name and IP address. An NLB array is fault tolerant to a point - it will skip servers in the list which are offline. However, beware that it is not application aware -- if the host CAS is up but Exchange Services are not, traffic might still be routed to that CAS server. Whether you accept this possibility is something you need to work out dependent on your implementation strategy. If you need more robust uptime, you'd need to look at an Exchange-aware load balancer, such as one of F5 or Barracuda's products.

-Matt
0
 
gtfijiCommented:
Matt is right, although it should be noted that DNS round robin is a supported solution.  MAPI clients (e.g. Outlook), ActiveSync clients (e.g. iPhones), and OWA clients (browsers) are all configured to attempt to contact all CAS servers to which they are referred by DNS.  If a CAS server dies and its A record persists in the DNS zone, clients who are directed to that CAS server first will simply experience a delay before they connect to one of the other servers.

There's also the matter of what happens to the users who are connected to a CAS server when it dies.  If you are using a Hardware Load Balancer or Network Load Balancing, OWA clients will seamlessly start being serviced by a different server in the event of an outage (as long as forms-based authentication is in use).  The browser would have to be closed and re-opened if DNS round robin is used.  MAPI clients will have to be closed and re-opened to connect to another server regardless of which load balancing solution you use.
0
 
Glen KnightCommented:
This question has been classified as abandoned and is being closed as part of the Cleanup Program. See my comment at the end of the question for more details.
0

Featured Post

Creating Active Directory Users from a Text File

If your organization has a need to mass-create AD user accounts, watch this video to see how its done without the need for scripting or other unnecessary complexities.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now