Clustering SBS

On June 26, 2006, in Backup, by

Invariably in newsgroups and listserves the question comes up “Can I cluster SBS?” and even more so now with Virtual Server and Virtualization the question comes up again…but it seems to me that everyone is looking for a technology issue to what I see is a people issue.

I don’t think people are comfortable with the Disaster Recovery story of SBS because they haven’t even tested out the recovery process of their own SBS box.

I’ll bet you that most of you reading this blog haven’t even tested out the process documented here in this SBS whitepaper on backing up and restoring.

But here’s the thing that came out of Jeff Middleton’s talk on disaster recovery of a small business at TechEd… sometimes a bit of understanding and placing in standardized spare RAID cards to ensure that you can slide from one piece of hardware to another is wise… sometimes understanding that the small business firm typically can handle some amount of downtime and as long as your DR strategy includes backdoors to the Internet and what not… it’s about understanding that you have options and are not limited.

Want more options?  Then look into Acronis for imaging just the single domain controller (and don’t add a secondary DC) …or add a secondary DC .. or script system state backups and park them offsite… or look into Symantec Live State…

..but get comfortable with the basic backup and restore solution for SBS…because right now I get the feeling that too many folks aren’t comfortable at all with their DR story that they are ‘selling’ to their clients.


6 Responses to Clustering SBS

  1. I still feel that Microsoft should have the restoring of their server software and our data made easier and less complicated to do and understand. IMHO there are people out there who install and support SBS 2003 etc, and only ever think about the install, and general runnign of the network, but they really should be thinking along the lines of “If this all dies how do I restore it”.

    The white paper on SBS2003 is fine, and as you mentioned, how many people actually read and then try this out…. probably very few!

    The Acronis approach is something I am currently looking at after experiencing a disaster of my own, and the white paper really did not work for me because of the time involved…

    I firmly believe MS should focus more on data restoration and spend less time on fancy new OS’s that will still need patches every month.

    To me SBS2003 is still an os with features that impresses me, but the backup and restore side “IS” still weak no matter what you say.

    Anyone else?

  2. Cluster? They must be kidding. If you want to diminish your downtime in case of total server failure (which is highly unlikely) then add another DC to the domain. It’ll be far more cost effective and will actual add functionality to the small business.

  3. Jim Barr says:

    Data Recovery is a subset of disaster recovery and is skill set that none of us practice often enough and too many not at all.

    If you want to understand data recovery in detail use ntbackup to backup a SBS server (dual processor, raid…) to a USB disk. Then recover the system to a laptop. You may have to drop some data due to disk space, you may need to add a PCMCIA NIC if the server has 2 NICs.

    It can be done and if you work through it you will understand the recovery process. Then tell me again why you need to cluster SBS.

  4. DonMurphy says:

    Great topic as usual! Clustering SBS isn’t that like an oxymoron.

    I *have* been through the SBS2003 restore process recently. It went very smoothly (As long as you don’t try to take shortcuts!).

    I also try to take a manual ASR backup every 30 days at every one of my sites. Why? Because:

    1) I try to do this just before patch Tuesday to get another snap of the server before changes.
    2) If the SBS2003 documented restore process fails you have at least system state and the integral files to get you back on your feet again.
    3) You get that all important asr floppy to help you be more effeiceint when recreateing the server if need be.

    One thing I currently don’t understand is when you are attempting an ASR restore process with the typical SBS setup (one DC) the concept of authorative and non-authorative restore blurs. Would you simply use the authortavie restore in this case and never have a need for non-authoratative restore? How does the concept of tombstone play into this? OK enough questions from me 😉

  5. Tony says:

    So someone wanted to cluster SBS in case of a hurricane? Clustering is not my thing, so I’ll state that up front, but wouldn’t that destroy two servers instead of one? Aren’t they usually clustered together in the same building?

    I agree with Amy, why not just another DC located offsite somewhere. I thought a DC located in a Virtual Server setting would be just the thing since the restore of that is on standardized equipment by the nature of the Virtual Server environment. This way, you can perform a Swing Migration as part of the Disaster Recovery effort to new hardware (assuming there’s good backups of the data, Sharepoint, any SQL databases and Exchange store).

  6. Tony says:

    One more thing…most people do practice their disaster recovery…at 2 AM after hours of finger nail biting, cursing, praying, pleading to be a better sys admin if the Server Gods will just let this machine come back online.

    It usually turns out badly.

    Like alot of posters above have stated, you need to ask yourself this question: Do I want to practice restoring the server on MY terms? Or under incredible pressure in the middle of the night? If you chose the second answer, then may I suggest you download all the white papers and KB’s you need BEFORE you lose internet connectivity.