Showing results for 
Search instead for 
Did you mean: 

Procedural suggestions for BE 2010 and Hyper-V

Level 2
I've got a fairly robust MS Hyper-v environment, with 3 non-clustered virtual hosts, hosting a total of about 30 virtual machines, and 1 CSV enabled cluster with 8 VMs.  Now that I'm on 2010, I'm ready to use the Hyper-V agent to back these machines up.  However, since the Hyper-V can only do a full backup, I'm left scratching my head as to the best method for backing up these machines efficiently and effectively.  Here are my thoughts on the various possibilities:(My backup media is a LTO4 tape library.
  • Back up all VMs with Full Hyper-V backup every day.  Pros:  Simple, and easy to control.  Cons:  Astronomical totals for data backed up.  Possible backup window problems, high cost associated with purchasing many more tapes. 
  • Back up VMs with Hyper-V backup weekly, and with normal Windows agent incremental daily.  Pros: Still not very complex, less data than full hyper-v every day.  Cons:  first incremental of the week will be a full backup unless the Hyper-V backup resets the archive bits on the individual files, which I don't think it does. 
  • Back up with Windows Agent weekly, incremental daily on weekdays, Hyper-V backup monthly.  Pros:   Weekly/daily proven to work.  Cons: Doesn't really utilize Hyper-V backup technology, restores could get messy, lots of jobs to keep track of.
  • Back up with Hyper-V agent weekly, and with Working Set daily on weekdays to back up that day's changed files. Pros: Least amount of redundant backup.   Cons: Restores could be messy, could miss data that was changed between Friday night and Monday on machines that are early in the Hyper-V backup process.  
  • Back up with Hyper-V agent weekly, and with Working Set (last 7 days) on weekdays to back up that week's changed files. Pros: Ensures that all data is backed up with minimal redundancy.   Cons: Restores could be very messy. 
So I come to the community asking you to weigh in on the above options, vote for your favorite, or if you see any missing pros/cons from the suggestions above, mention them.  


Level 6
 Dont forget Dedupe!  That will help you save a lot of space in regards to your backups to disk.

Level 2
I'm not sure how much Dedupe will help when I'm backing up hundreds of GBs of vhds every night.  The vhd files will be different, so therefore they won't be stopped with dedupe.

Level 6
 For Hyper-V Symantec recommends an agent in the guest for dedupe to work optimally.  When you purchase the hyper-v agent it includes the ability to install remote agents for windows anyways, so dedupe would help save space if doing it client side.

Level 2
I have a similiar setup.  Right now I am just backing up the VMs twice a month, and doing file level differential/full backups the rest of the time.  I know that Acronis is supposed to be able to do incremental VM backups, but I have never tested it.

Have you tried a restore of the VM yet?  Mine backup fine, but during the restore it failed to register the VM correctly to my cluster and I had to reconfigure the VM structure (the VHD file were fine though)

Level 2
So which scenario did you end up implementing?

I'm trying to backup a 3-host CSV cluster plus a few physical servers with BE2010, I have agents for Hyper-V, SQL, and Active Directory.  I have about 2TB of data, but if I'm to backup the VM plus the VHD files in CSV then it double to 4TB which takes ~40 hours even on an LTO4/Gbps environment.

If anyone can provide feedback on how to properly backup Hyper-V VM running SQL in a CSV that'd be greatly appreciated!

There's no clear instruction on how to properly backup Hyper-V from Symantec.  It'd be great if they can provide info for these.
1. Backup configuration for a general purpose server running on a Hyper-V guest in a CSV cluster.
2. Backup configuration for a SQL server running on a Hyper-V guest in a CSV cluster.
3. Backup configuration for a Active Directory server running on a Hyper-V guest in a CSV cluster.

Level 4
I'd like to know this as well.  2 clustered Hyper-V hosts, all machines on a CSV.  About 1.5tb of data