Announcement

Collapse
No announcement yet.

Planning RD strategy

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

  • Planning RD strategy

    Hello,
    I´m running some test for DR methods that could apply to my company´s needs and limitations,
    we have a single forest, single domain, single W2K DC running Exchange 2K, we plan to add a sencond box for failover, but we do not have room for extra boxes to act as additional DCs.

    So, following teiger´s suggestion I set up a virtual machine to act as a tempDC and ran some drills as if I was doing a swing migration, so far so good, but I am also looking for a plan in case of a disaster situation...

    Now, what I am really interested is in recovering the network in the event of SRV1 (the main DC) going totally down. So what I did was to pull the plug and take SRV1 off the lan, just like that.
    Then I seized roles in the tempDC, cleaned the metadata, etc. after that I set up SRV2 with the same name/IP, installed Exchange with /disasterRecovery and restored a ntbackup.

    I ended up with a working domain, the users restored and exchange running fine, the time it took was pretty much the same as in my swing migration simulation, but the big difference was that I kept the original SRV1 without demoting it or anthing like that...

    Doesn´t this mean that, if anything would have gone wrong I could just take SRV2 and tempDC out and re-plug SRV1 having the lan working fine under the original master DC??
    If this is correct then why isn´t this approach better than demoting SRV1 as in a swing migration?
    Or am I missing something and some data is not making it to the rebuilt DC/Exchange?


    As related questions,
    * how long does it take for DCs to get in sync? (so I can proote/demote them).
    * Is the "Active DIrectory Replication Monitor" tool the best way to confirm everything is in sync so I can continue the migration?

    Thanks a lot in advance.
    Siko.

  • #2
    Re: Planning RD strategy

    In reality, if SRV1 will totally crash, you should act as the following:
    1. Seize all the roles that located on SRV1 to SRV2.
    2. Restore the Exchange server with the setup /disasterrecovery
    3. Clean the SRV1 from AD (metadata cleanup).

    If the SRV1 will not totally crash and it is accessible - then whey demote it by Dcpromo and not fix it?

    Remember that if you created additional DC SRV2 - you can not setup /disasterrecovery the Exchange to this machine, because you must keep the same Exchange name.

    * In your environment, when you have only 2 DCs (if network is OK) - it should not take too long, about 15 minutes, but you should check with replmon and event viewer, those tools are enough
    CNE 5, CCA, MCSE NT4.0-2003, MCSE 2003 messaging, Exchange Server MVP.
    Tzahi Kolber - IT Supervisor
    Polycom Israel.

    Comment


    • #3
      Re: Planning RD strategy

      In Swing Migration you do NOT run DCPromo, but leave it running and run SRV2 off-line and seize the roles - meaning you can leave your production LAN running. In case of DR you have a running SRV2 which you can then swing back to SRV1. But why can't you just run DSRM and restore Active Directory from your latest backup?
      TIA

      Steven Teiger [SBS-MVP(2003-2009)]
      http://www.wintra.co.il/
      sigpic
      I’m honoured to have been selected for the SMB 150 list for 2013. This is the third time in succession (no logo available for 2011) that I have been honoured with this award.

      We don’t stop playing because we grow old, we grow old because we stop playing.

      Comment


      • #4
        Re: Planning RD strategy

        Hi, thanks fir the replies!
        Originally posted by tkolber View Post
        In reality, if SRV1 will totally crash, you should act as the following:
        1. Seize all the roles that located on SRV1 to SRV2.
        2. Restore the Exchange server with the setup /disasterrecovery
        3. Clean the SRV1 from AD (metadata cleanup).
        But then I either loose the AD by not making SRV2 a synced DC or prevent myself from restoring an exchange backup made with ntbackup since I cant restore to a machine with a different name. Having a third (virtual) DC alloews me to keep the lan running and restoring Exchange in a different box with the same name.

        Originally posted by tkolber View Post
        If the SRV1 will not totally crash and it is accessible - then whey demote it by Dcpromo and not fix it?
        hmm I´m inclined to think that when the time comes it will be faster to swap the machines rather than fix the DC. For this company it is more important to have the mails not bouncing and maybe accesing them through OWA than anything else.

        Originally posted by tkolber View Post
        * In your environment, when you have only 2 DCs (if network is OK) - it should not take too long, about 15 minutes, but you should check with replmon and event viewer, those tools are enough
        Thanks!

        Originally posted by teiger
        In Swing Migration you do NOT run DCPromo, but leave it running and run SRV2 off-line and seize the roles - meaning you can leave your production LAN running. In case of DR you have a running SRV2 which you can then swing back to SRV1.
        I´m sorry, you lost me there ...
        I thought I DCPROMO in order to cleanly remove SRV1 from AD and have SRV2 with the same name/IP than SRV1.
        I also thought I avoid seizing the roles by transferring them instead using a third DC as a temporary FSMO master.
        Maybe that´s not a swing migration but a half-swing, or maybe I am totally off the mark....
        Originally posted by teiger
        But why can't you just run DSRM and restore Active Directory from your latest backup?
        I am trying to have a plan for when that is not enough. Again, maybe I am totally of the mark....

        Thanks

        Comment

        Working...
        X