Announcement

Collapse
No announcement yet.
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • HQ Server Failover

    Customer recently had a network outage on their HQ site (caused by a loopback’d patchlead). This has raised some interest in HQ Server Failover.

    I know the ShoreTel response is DoubleTake, but is there anything else..?

    Reason I ask is they would like their failover solution to be in a different site with a different subnet – and as far as I can tell DoubleTake will not accomplish this (as it reboots onto what was HQ IP address right..?).

    Any thoughts..?

  • #2
    We're providing a much simpler mode of failover for the HQ server - to where they can still answer calls but can't use call manager, workgroups etc. (This is usually the case as another server / Double Take etc is usually over most customers budgets)

    Every DID that points to a auto-attendant menu goes to a workgroup first. The workgroup is set to an always call handling mode to point to the AA menu. Workgroup has a backup extension of a huntgroup.

    This way, if HQ is down, all the calls to that menu will default to the huntgroup.


    I do remember some of the network guys in my advanced class talking about an image backup solution that allows you to restore to a virtual machine very quickly but the name of the software has escaped me.

    Comment


    • #3
      I would recommend that you not try to change sites. That is not going to go well no matter what.

      Simply have a second server ready-to-go but with all ShoreTel services and DCOM stopped. To make it active,
      1. Copy over the Shoreline Data folder
      2. Change the IP of the box to the same IP as the old box
      3. Start the ShoreTel services.

      Comment


      • #4
        Thanks for your comments guys - we do have a certain degree of failover in place using HG's, etc... The request for a failover solution on a different site comes from the customer. The broadcast storm they experienced on HQ site shook them up a bit! They are a large multi site Syntelect deployment by the way.

        They are also planning moving office in the not too distant future, and are considering logistics of their being no HQ Site for a couple of days.

        Comment


        • #5
          Shoretel introduced support for leveraging Doubletake across subnets in Shoretel 9.0. You actually specify in Director the IP address of the hot-standby Director server and Shoretel publishes this second IP to all relevant devices.

          We tested this for them back in 9.0 and it works like a charm.

          Comment


          • #6
            Thanks Chris - I have seen the option for a secondary host for the HQ Sites Application Server, will investigate.

            Comment


            • #7
              For what it is worth, I just watched a VMWorld 2008 presentation (available for free on VMWorld site) from Cisco that indicates they are successfully able to run the Cisco Unified Communication Server in VM without ANY issues. So, if you are willing to go this direction and you have or are going to have a VM farm, then you can have a real nice DR solution. Then you can have a single Shoretel server and you'll be protected against hardware failure as the Shoretel server will spin up on another ESX box automatically. PLUS!!! You get de-dup, snapshotting and off-site SAN vaulting which really helps take care of a catastrophic DR scenario.

              The key to it working for Cisco was to specify processor affinity for the Cisco server. That was with VMware 3.5. They indicated that VSphere 4 is more improved and they feel they can accomplish the same thing with simple reservations instead of processor affinities.

              Later this year, I'm going to try putting our Shoretel server on our ESX 4 environment and see if I can't get it to work. I tried it last year on VMWare Server (GSX) on a workstation grade desktop PC, but the the free/GSX edition just doesn't have the performance of the full blown ESX version on a real 8-way server, and it only held up for about 10-15 phones before we started having problems when all the users hit voice mail (this was a test environment thankfully.)

              Comment

              Working...
              X