Windows Backups, efficiently moving VHDs

[EffrafaxOfWug]

OK... so all of the windows machines in the house back up to their own little CIFS share on my file server (linux), and then get shunted off to airgapville.

This causes a problem with system images, the VHDs of which are frequently huge single files. Tools like rsync are meant to be able to do efficient copies over the network, only moving the blocks in the VHD that have changed, but because windows backup also renames the VHD every time it creates one, replication ends up taking an eternity even though very little in the VHD should have actually changed.

Is there a trick I'm missing to copy windows backup VHDs efficiently across the network under linux, i.e. a tool like vhdcopy or somesuch?

 

[cesmith9999]

There are tools to do this. Where I work we have a service that will seed the copies on the remote site to lessen the amount of data we really send over the wire. unfortunately, this is not a public tool.

Chris

 

[Naeblis]

hyper V replication does this for you

 

[Naeblis]

Sorry- could not figure out how get the pic in the previous post

 

[Lost-Benji]

www.arcserve.com

 

[EffrafaxOfWug]

Sorry chaps, not sure what you mean by these suggestions - I'm not using Hyper-V or ArcServe and don't have any plans to. I'm looking to move the VHDs from wndows client backups from server1 over the WAN to server2 in the most efficient way possible (i.e. only transferring the delta) and wondering if anyone knew of any tools that could accomplish this.

 

[Marsh]

Bvckup 2 | Simple fast backup

 

[Lost-Benji]

Put VHD on HDD, take from site one to site two, plug drive in and transfer.

P.S Arcserve can do over WAN transfer as well as backup.

 

[EffrafaxOfWug]

Mission accomplished - rsync was perfectly capable of transferring the VHDs efficiently but was just not doing so because of windows always placing the VHDs under a different path, so just need to make sure the source and destination directories always sync to a common name/path. Changed the initial sync jobs to ignore *.vhd and for the moment I'm manually hardlinking the VHDs (whose name never changes from the disc GUID but whose path does) to a static filename on src and dst and rsync that instead - sync time drops from days to a few hours. Just remains to be seen how reliably automated I can finagle a linking backup script.

Output from the manual run; this ~128GB VHD (a 256GB disc about half-full) synced with its previous (one month older) counterpart over a 1MB/s WAN (with a little bit of help form some compression) in under 90mins, only actually transferring about 3GB of data. Should have run it through `time` really but I was already halfway through my martini and getting impatient (and still not entirely convinced it would work).

root@bakstor:/storage# rsync -aAHv --partial --progress --compress --compress-level=9 --bwlimit=1024 effrabax@wug:/storage/backups/windows/heartofgold/ntbackup/WindowsImageBackup/latest.vhd_lnk /storage/priv/snafu/backups/windows/heartofgold/ntbackup/WindowsImageBackup/latest.vhd_lnk
receiving file list ...
1 file to consider
latest.vhd_lnk
135,129,908,224 100%   25.77MB/s    1:23:21 (xfr#1, to-chk=0/1)

sent 9,279,272 bytes  received 3,268,492,954 bytes  549,823.40 bytes/sec
total size is 135,129,908,224  speedup is 41.23