I'm running Proxmox with currently only 1 guest: Win7Pro64bit Proxmox itself does use vzdump to create a backup of the VM. There is a patch to get it to create diffs when doing vzdump. So far all is good except that this machine currently has 30GB full and between 5-10GB diff backups. This is very weird as I definitely know there aren't that many files changed daily, its a file server, mostly xls and doc files and some apps but they don't write such bug files. I was looking to get the backups off-site i.e. Amazon S3 but with these file-sizes that is very impractical. So I have a few ideas but need some feedback, maybe someone else has solved this more elegantly? a) I could only upload 1 full backup every week to S3 and do a daily backup from inside the Windows guest to S3, i.e. with http://s3tools.org/s3cmd b) I could use this script that creates a vzdump, mounts it then runs duply/duplicity over it and basically duply backs up to S3 => https://github.com/maff/lvmsnapshot c) I could do a daily vzdump (full) and then run duply over it but I'm not sure if duply picks up differential changes in such a huge file. (Have duply store the backups on S3) Any other ideas? Basically I need to be able to restore the VM completely so I definitely need say 1 full VZ dump, say max. 1 week old and I need to secure the data inside the VM (preferably daily and store the diff backups on S3)
Btw. the machine has a RAID6 (hardware Raid) and I have been toying with the idea of deduplication but then I think installing OpenDedup inside the guest would just complicate matters and running another VM with dedup capabilities i.e. some sort of NAS with ZFS will only lead to complications too so I think "deduplicating" the backups and then uploading is the way to go... Talking about dedup, I found this backup tool but am unsure how and where to incorporate it into my backup process: http://zbackup.org/
