Announcement

Collapse
No announcement yet.

Clustered Exchange remote disaster recovery site options.

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Clustered Exchange remote disaster recovery site options.

    Hi

    I am looking for advice and ideas.

    We have an Exchange 2003 server operating in a two node cluster located one city. I need to establish a disaster recovery site which would be located in a different city / same AD. Both sites are well connected with high bandwidth connections.

    The preface is the first site is destroyed but we would still have our current active directory available which is common to both sites.

    The primary objective is to establish email flow in and out as quickly as possible for around 1800 users with restoration of historical mail being a secondary issue.

    I am familiar with the great setup/disasterecovery available with a stand alone system but thatís not available with a clustered system. Iíve created a full set of ASR backups of the current system but that requires the same hardware be available to restore which is not going to be possible at the alternative site.

    Iíve read about using standby nodes in 2003 and that looks like an option but itís not recommended to do that on different subnets which would be the case at the remote site.

    Perhaps the best option is to rebuild the cluster and exchange server at the remote site but that would require clearing the old cluster info from AD and I am not sure what would then happen re users / mailbox reconnections.

    I notice Exchange 2007 offers the ability to do a setup/disasterecovery type dial tone recovery to a clustered system. That looks pretty good but would require justifying a lot of new machines and I had planned to wait till they release service pack 1 for 2007.

    I do have one advantage in that I have a copy of Ontrack Power controls so I can restore the old databases and extract historical mail and dump it back to the server without having to mount the original systems databases.

    Thanks

    Mike

  • #2
    Re: Clustered Exchange remote disaster recovery site options.

    How much disk space is Exchange using at your original site?
    VCDX3 #34, VCDX4, VCDX5, VCAP4-DCA #14, VCAP4-DCD #35, VCAP5-DCD, VCPx4, vEXPERTx4, MCSEx3, MCSAx2, MCP, CCAx2, A+
    boche.net - VMware Virtualization Evangelist
    My advice has no warranties. Follow at your own risk.

    Comment


    • #3
      Re: Clustered Exchange remote disaster recovery site options.

      It is possible to do this with separate subnets, and we have this setup at my place of employ. Being an Exchange person and not a full-on network person, I may not be much use on that side of things. But it is possible.

      I think that to get a project of this scale up and running, you might need a Technology Partner - or at least we did on the networking side.

      It's supposed to be easier if you have it all on one subnet, and if you do, then the failover can be automatic. So if you lose your main site, then an automatic failover can take place to your DR site.

      We'd already gone some way towards building our new networks, all triangulated and that, and all based around different subnets.

      So for Exchange, we had to look at doing the geographically dispersed thing with different subnets. Downside? In a Disaster, my networking team need to be on hand to adjust the switches to use a different "VLAN". But then it's all hands on deck in a DR situ, so they'll be there! It's a few lines of Cisco-speak they type into a terminal type connection. Once they do what they do, our Outlook clients are none the wiser, and don't need reconfiguring.

      To get all this working required two SAN's, one at each site, and a replication solution (Continuous Access) so that the Exchange databases replicate between the SANs. We chose HP hardware to achieve this.

      This is the expensive option. The key thing here is that it is possible to recover from a Disaster without losing any data.

      Required:
      Your Network capable of creating VLANs across your network
      SAN at each end
      CA equipment at each end
      High bandwidth connectivity between your two sites.

      Advantage:
      Failover and recovery within typically 10 mins with virtually no loss of data
      When your network guys make the VLAN switch, your Outlook clients won't need reconfiguring

      Disadvantage:
      Harder to set up
      Costs of extra SAN and disks
      Costs of CA equipment
      Costs of getting a Technology Partner in

      Another option\idea I've had that you might consider is offsite backup using an online service.

      With this, you might have your Exchange servers at one site. Have them backed up remotely by another company to an offsite store. Have warm Exchange Servers configured up at your remote site. They'd be the same as your live servers, but perhaps in a separate VLAN on a separate network so that the rest of the main network does not see them - get this working by taking in a copy of a DC into the separate network.

      In a disaster at the main site, you can request that the restore is directed to your DR site. Because your DR site is warm and identical, you could be up and working within a couple of hours.

      Required:
      Your Network capable of creating VLANs like I've made up above!
      An off-site backup company
      High bandwidth connectivity between the offsite company and your two sites.

      Advantage:
      Easier to set up
      Easier and quicker to recover from (compared with rebuilding servers)
      When your network guys make the VLAN switch, your Outlook clients won't need reconfiguring (might need restarting)

      Disadvantage:
      Costs of storing your backups offsite (but this can be comparable to the tapes you might use)
      In a site-destroyed disaster, you revert back to the previous night's data (but that is perhaps what would be expected)


      Your Option:
      ========
      "Perhaps the best option is to rebuild the cluster and exchange server at the remote site but that would require clearing the old cluster info from AD and I am not sure what would then happen re users / mailbox reconnections."

      This is definitely doable, and without any fancy VLAN stuff.

      Have ready your cold Exchange Server, built but not joined to the network. You could go further and build it in a test lab, with a copy of a local DC in the test lab.

      Assuming you build your lab at the remote site, you might bring a copy of your Dc from the main site. Once you've built it in the test lab, it will be as your Live Server.

      You should leave it in your test lab, and patch it whenever you patch your Live server.

      When Disaster Stirkes, and assuming you have an offsite policy for your backup media, you would just recall the tape and restore to your server in the Test Lab.

      Once that's done, move the server from the Test Lab to the Live network in your Secondary Site. This will no doubt be on a separate subnet, so you'll need to change the IP addresses on the restored server.

      Because you keep the server name as the same, you won't need to do anything with AD - as far as AD is concerned, it is the same server.

      You'll then have to update your DNS servers so that EMAIL001 server is pointing to your new IP.

      Now then, the PCs that are still up and running, perhaps in the secondary site, might still be confused. They'll have cached DNS entries that would still see Outlook stubbornly trying to access the deceased server at the destroyed IP address. You might get away with a \flushdns etc. You might need to restart the PC. This is the downside with this solution - you'll almost certainly end up needing to visit each PC!


      Hope this ramble helps.

      I should probably go to bed now.

      Comment


      • #4
        Re: Clustered Exchange remote disaster recovery site options.

        Originally posted by jasonboche View Post
        How much disk space is Exchange using at your original site?
        About 260 gb at present. Thats on a SAN the DR site would be local storage.

        Comment


        • #5
          Re: Clustered Exchange remote disaster recovery site options.

          Thanks for that Robbyb

          I am going to have a play in the lab next week. I kind of like the idea of the cluster recovery available in 2007. If that proves to be as good as 2000 / 2003 stand alone recovery option I might push for an upgrade on the basis of DR safety.

          Comment

          Working...
          X