locked
Support DNS round-robin for Exchange 2013 clients RRS feed

  • Question

    1. Connections clients OWA, Outlook (Anywhere), ActiveSync to load balanced by DNS round-robin Client Access Servers is supported solution for Exchange 2013 by Microsoft when contacting support service?
    2. If used round-robin the clients specified above will automatically switch to the other CAS after connection timeout or failed?
    • Edited by rizor Tuesday, December 10, 2013 8:16 PM
    Tuesday, December 10, 2013 11:59 AM

Answers

  • I know Zbyněk,

    1. but you still need to delete that DNS entry manually, as round robin will still provide that IP to the server that is down. SO, some client will still try to connect to that cas server!

    HW LB will provide that service for you.

    2. Outlook will connect to another CAS server after a short timeout period, so you dont even have to make lower DNS TTL. 

    DNS round robin will work fine IF you can live with the manual adminstration AND some timeouts

    /anker

    • Marked as answer by cara chen Thursday, December 19, 2013 6:44 AM
    Thursday, December 12, 2013 9:54 AM
  • Microsoft also recommends the solution as you read at the document I linked. The supportability always depends the nature business/infrastructure you have

    Regards from ExchangeOnline.in|Windows Administrator Area | Skype:manuphilip@outlook.com

    • Proposed as answer by cara chen Wednesday, December 11, 2013 5:49 AM
    • Marked as answer by cara chen Thursday, December 19, 2013 6:46 AM
    Tuesday, December 10, 2013 4:25 PM

All replies

  • Hi,

    DNS Round-Robin is the recommended technology for CAS Load balancing on Exchange 2013. In Exchange 2013 technology change, client access servers are 'stateless' and this supports the DNS Round-Robin. See more details here: http://technet.microsoft.com/en-us/library/dd298114(v=exchg.150).aspx


    Regards from ExchangeOnline.in|Windows Administrator Area | Skype:manuphilip@outlook.com

    Tuesday, December 10, 2013 1:53 PM
  • Thanks for your reply but I need to confirm that the solution officially supported by Microsoft when contacting support service.
    Tuesday, December 10, 2013 4:20 PM
  • Microsoft also recommends the solution as you read at the document I linked. The supportability always depends the nature business/infrastructure you have

    Regards from ExchangeOnline.in|Windows Administrator Area | Skype:manuphilip@outlook.com

    • Proposed as answer by cara chen Wednesday, December 11, 2013 5:49 AM
    • Marked as answer by cara chen Thursday, December 19, 2013 6:46 AM
    Tuesday, December 10, 2013 4:25 PM
  • Thank you.
    If used round-robin the clients specified above will automatically switch to the other CAS after connection timeout or failed?
    Tuesday, December 10, 2013 8:15 PM
  • Hello, It is but don´t forget to decrease TTL value for DNS records to acceptable time. For example 1 minute.

    Regards


    Zbyněk

    • Proposed as answer by cara chen Wednesday, December 11, 2013 5:49 AM
    Tuesday, December 10, 2013 9:11 PM
  • I posed the question a bit incorrectly. The round-robin is optional in my case.

    Let's me explain:
    The DNS server has several A record for same name. The client uses this name to connect to the CAS. Query by that name returns multiple IP. Whether the clients will switch to the next IP if occur connection timeout or failed?

    On an example of a farm of terminal servers, this process is shown here http://www.brianmadden.com/content/images/SBLBgif.gif

    My question is more about resiliency than load balancing.

    Wednesday, December 11, 2013 4:54 PM
  • Its not like for example lync client, that uses DNS loadbalancing on the application side.

    if your server fails, outlook will still try to connect to that ip address (it has already a session).

    /anker

    Wednesday, December 11, 2013 5:25 PM
  • Hi,

    Exchange 2013 is not session based anymore (since 2013). Your client can connect to any available CAS server (frontEnd) which is stateless proxy and request is redirected to  right Mailbox server / proxied to another CAS server. Resiliency in this case is handled by DNS record TTL. For example if you have TTL for 1 minute, if server goes down, DNS record for that particular server ages out and you will be connected to next available IP address for the load balanced name. So total loss for your client is 1 minute in case of server failure. 1 minute of a year is  0,00019% which is totally ok for 99,999 availability. Hopefully I have counted it right.

    Regards


    Zbyněk

    Wednesday, December 11, 2013 5:39 PM
  • I know Zbyněk,

    1. but you still need to delete that DNS entry manually, as round robin will still provide that IP to the server that is down. SO, some client will still try to connect to that cas server!

    HW LB will provide that service for you.

    2. Outlook will connect to another CAS server after a short timeout period, so you dont even have to make lower DNS TTL. 

    DNS round robin will work fine IF you can live with the manual adminstration AND some timeouts

    /anker

    • Marked as answer by cara chen Thursday, December 19, 2013 6:44 AM
    Thursday, December 12, 2013 9:54 AM
  • Hi Anker,

    you are absolutely right! My fault. I apologize.

    Regard


    Zbyněk

    Thursday, December 12, 2013 10:28 AM