locked
Lync 2013 - Stretched pool RRS feed

  • Question

  • Hi.

    In my organization we have deployed a stretched pool over two sites(2 FE's,1 Edge and one SQL-server(mirror) in each site).  It's my understanding that this is not a supported deployment?

    We have 3-way redundancy on our network lines between our sites.

    I've read more about this stretched-pool solution and I'm concerned what will happen if all network lines between our sites goes down. 

    It's my understanding that both sites will operate with the assumption that the other site is down, and Lync will contiune to serve users. And when the network line is re-established we have incosistent data between the servers.

    If this happens, what will we have to do? Is restore the only option? Perhaps we should redesign our Lync-deployment?

    What are your thoughts on this?

    Friday, December 12, 2014 10:31 AM

Answers

  • You would configure 2 Lync Pools.  Pool A=SiteA Pool B=SiteB.  You would then configure pool pairing.  If site A fails there will be no impact to site B.

    Please check the below link for more details about pool pairing:

    http://technet.microsoft.com/en-us/library/jj205293.aspx


    Tek-Nerd

    • Proposed as answer by Ben Donaldson Friday, December 12, 2014 9:03 PM
    • Marked as answer by Eric_YangK Sunday, December 21, 2014 6:13 AM
    Friday, December 12, 2014 2:41 PM
  • Agree with Tek,

    A Pool pair will not only move you into a supported topology, but also provide you with site level disaster recovery and further scheduled maintenance options.

    Providing your not already running your pools at capacity (in terms of users), you will be able to transition to a new topology without great impact to your production environment through a well thought out user migration process without the need to introduce additional Front Ends.

    Pool Pairing: http://www.gecko-studio.co.uk/dont-play-with-fire-play-with-pool-pairing-configuring-testing/

    Kind regards
    Ben


    Note: If you find a post informative, please mark it so using the arrow to the left. If it answers a question you've asked, please mark the thread as answered to aid others when they're looking for solutions to similar problems or queries.

    For Fun: Gecko-Studio | For Work: Nexus Open Systems

    • Edited by Ben Donaldson Friday, December 12, 2014 9:08 PM
    • Marked as answer by Eric_YangK Sunday, December 21, 2014 6:13 AM
    Friday, December 12, 2014 9:08 PM
  • Hi Christer C,

    Based on my experience, it should work when one site was down.

    With four servers in a pool, at least two servers should be running for pool to be functional.

    I suggest you deploy SQL mirroring with witness, I recommend using a witness because it enables failover of the Back End Server to be automatic. Otherwise, you must manually invoke failover.

    If you don’t deploy the witness server, you can refer to the following article to failover and failback your database.

    http://technet.microsoft.com/en-us/library/jj204991.aspx

    In addition, the best practice is to follow Official documentation.

    Best regards,

    Eric

    • Marked as answer by Eric_YangK Sunday, December 21, 2014 6:13 AM
    Monday, December 15, 2014 9:34 AM

All replies

  • You would configure 2 Lync Pools.  Pool A=SiteA Pool B=SiteB.  You would then configure pool pairing.  If site A fails there will be no impact to site B.

    Please check the below link for more details about pool pairing:

    http://technet.microsoft.com/en-us/library/jj205293.aspx


    Tek-Nerd

    • Proposed as answer by Ben Donaldson Friday, December 12, 2014 9:03 PM
    • Marked as answer by Eric_YangK Sunday, December 21, 2014 6:13 AM
    Friday, December 12, 2014 2:41 PM
  • Agree with Tek,

    A Pool pair will not only move you into a supported topology, but also provide you with site level disaster recovery and further scheduled maintenance options.

    Providing your not already running your pools at capacity (in terms of users), you will be able to transition to a new topology without great impact to your production environment through a well thought out user migration process without the need to introduce additional Front Ends.

    Pool Pairing: http://www.gecko-studio.co.uk/dont-play-with-fire-play-with-pool-pairing-configuring-testing/

    Kind regards
    Ben


    Note: If you find a post informative, please mark it so using the arrow to the left. If it answers a question you've asked, please mark the thread as answered to aid others when they're looking for solutions to similar problems or queries.

    For Fun: Gecko-Studio | For Work: Nexus Open Systems

    • Edited by Ben Donaldson Friday, December 12, 2014 9:08 PM
    • Marked as answer by Eric_YangK Sunday, December 21, 2014 6:13 AM
    Friday, December 12, 2014 9:08 PM
  • Thank you for quick answer.

    I understand that re-designing is the better solution here.

    But I'm curious what would happen in an event as I described in the OP. Anyone have any info on that?

    Monday, December 15, 2014 9:19 AM
  • Hi Christer C,

    Based on my experience, it should work when one site was down.

    With four servers in a pool, at least two servers should be running for pool to be functional.

    I suggest you deploy SQL mirroring with witness, I recommend using a witness because it enables failover of the Back End Server to be automatic. Otherwise, you must manually invoke failover.

    If you don’t deploy the witness server, you can refer to the following article to failover and failback your database.

    http://technet.microsoft.com/en-us/library/jj204991.aspx

    In addition, the best practice is to follow Official documentation.

    Best regards,

    Eric

    • Marked as answer by Eric_YangK Sunday, December 21, 2014 6:13 AM
    Monday, December 15, 2014 9:34 AM