backup strategy and disaster recovery RRS feed

  • Question

  • Hi,

    if i got the picture right backup is done differentially. For the first backup all data clusters are unsaved and therefor a "full backup" has to be done. For following backups only clusters changed since last backup are saved. In this case restore or viewing of a snapshot would become slower and slower when many backups exist because i would have to start with the first backup each time and go thru all subsequent backups. Imagine you have daily backups for 2 years there are about 700 snapshots. Is my assumption right? No, stop: i remember backup maintainence where you configure which (automatically saved) backups to keep; for the long term only 3 backup sets would be kept (except manual backups).


    Disaster recovery: essentially important for me would be a way to have reliable backups in the case of some kind of home server damage. There should be support for some users who need reliable backups; eventually by means of replication to a second home server or some other techniques (offline backup/dump to external disk, mirroring, ...). I understand that most of the SOHO users would not need such a reliable system.





    Tuesday, April 17, 2007 3:23 PM

All replies

  • Umm, sort of. The very first backup, of the very first computer backed up to WHS, is a backup of every used cluster on that computer. Every backup for every computer after that is really more of a delta: only clusters that don't exist in the current set of clusters on WHS will be backed up. That's somewhat like a traditional differential backup, except at a lower level. If all the backups that contain a particular cluster expire, the cluster itself expires during the next backup maintenance window (Sundays) and the space it occupied is recovered.

    As for "slowing down" I haven't noticed any problems accessing 2 month old backups and extracting files. I can think of several ways to achieve this, the most likely being that every backup is stored as a list of the clusters it contains. I know the clusters are stored separately; there are a large number of 4 GB files full of them on the average WHS, I suspect.

    The disaster of WHS itself dying isn't as serious as you imagine. Unless whatever happens physically destroys the entire computer, you still have the data that was on any hard drive that wasn't affected by the problem. If you lose the stored backups, so what? You've still got the PCs those backups were from. Rebuild the WHS, or buy a new one, and reconnect the clients. (Remember, this is Windows Home Server.) Backups run, everyone is happy. The backups on WHS are obviously not intended as a solution to Fat Finger Syndrome (though they can help with it); they are truly for disaster recovery, as in "My PC crashed! Daddy, please help me!"
    Tuesday, April 17, 2007 4:19 PM
  • The short answer is...it doesn't work that way.  WHS uses its own backup technique that presents the complete volume as of a specific date when you restore.


    The backup of backups issue has been discussed in the Suggestions section.  

    Tuesday, April 17, 2007 4:26 PM
  • ...they are truly for disaster recovery, as in "My PC crashed! Daddy, please help me!"


    As you know, true disasters include house fires and flooding among other things. WHS does very little to help in these cases, which is why offsite backups make very good sense as a companion/extension to the software.

    Tuesday, April 17, 2007 8:03 PM