Help reducing backup size

[AndyR]

Hey Guys,

Need a bit of help here with reducing my backup sizes so just looking for ideas.

At the moment I'm backing up about about 2.5TB of VMs each night which are going to an iSCSI NTFS lun attached to my Veeam VM, backing everything up in Virtual Appliance mode. Speed is great and I can get everything backed up each night in about 4hrs. At the moment the LUN they backup to is 800GB in size (no space to expand) and because of the amount of VMs I can only keep 4 restore points for each job in a reversed incremental setup, as each morning I have to dump the backups to an LTO4 tape as well which takes 5hrs. Problem is both the LUN and the tapes are getting full, and my free space on both hovers around 70Gb free all the time. This isn't too bad, but if something big changes on a VM one day the free space can jump down to 40Gb free etc so its getting tight.

My daily reverse incremental jobs each day (5 jobs total) come in at a combined size of anywhere between 30Gb - 50Gb depending on how much changes, hence I only keep 4 restore points, as well as the forever full backup for each job.

The majority of the VMs are RHEL but I have about 20 Windows Servers too, so on these I have the pagefiles all set to a seperate VMDK file on SCSI3:0 which is excluded on the jobs to save space.

I'm a one man IT guy here, supporting 35 users and around 60VMs, 15physical servers so try to do everything I can to keep my back covered from a recovery point of view. Basically, I'm just looking for suggestions on how to get the backup sizes down for both storage and tape while still giving me the best scope for recovery.

Thanks all,

Andy

[tsightler]

OK, I know this isn't exactly what you asked, but why not just buy a small NAS? They're cheap, and then the issue goes away. You can buy a 2TB NAS for approximately the cost of pack of LTO4 tapes. Certainly if you can afford all the hardware to run all of this, and tape libraries, and VMware licenses, etc, etc, you can find a few hundred bucks for a decent NAS to eliminate any concerns here.

Just about the only thing you could do otherwise is to switch to WAN target, this will lower the block size for change blocks which will likely lower the size of your rollbacks at the cost of some performance (hard to predict how much). You'd have to rerun your full backups though. You could also consolidate the jobs which should increase the dedupe slightly. If your RHEL systems are configured with separate OS and application partitions, and are relatively identical, you could choose not to backup the OS portions.

[J1mbo]

Agree on the NAS, for instance an old(er) server could be re-purposed with some 2TB sata drives, preferably in RAID-10.

But that aside, delete Windows update roll backs, old logs, the usual stuff, then clean free space using dd on linux or sdelete on Windows, then the entire free space area will take no space on the backup. A couple of notes of caution; it may be possible to run out of space in the interim on your backup LUN doing this en-mass, because of the extent of the resulting deltas that would follow; and secondly this will expand any thin-provisioned VMDKs to their full size, which can then be condensed again by storage vMotion (or cold migration) to another datastore.

[Vitaliy S.]

Andy, you may also want to try best compression, that would surely reduce the size of the backup files, but will consume much more CPU resources on the backup server and may affect overall jobs' performance rates.

[AndyR]

Thanks for the input guys, looks like I have a few options. Might try bumping the compression up first and see how much of a hit it gives me as thats the quickest solution.

Buying a NAS would be my preferred option, but I know they will never go for that as costs are tight as it is now, but I have a ton of spare desktops lying around all fairly high spec so I'm thinking it might be worth me re-purposing one of these as a Veeam server and putting in a 2TB Sata drive or something, and running the backups to this.

One thing I have just thought of which someone might be able to clarify. As I take forever incremental backups there is always a full backup file each day plus the incremental. Technically, because I backup the Veeam backup lun to tape each day, could I just backup the full backup file, and not each of the incremental as if I need to do a recovery from a certain day, I could just restore that full backup file for that day and recover from that? That would solve my problem of lack of tape space until I can get approval for either an autoloader or replacing the drive with LTO5.

Thanks!

[Gostev]

That is correct, with the reversed incremental backup mode, VBK file is all you need to restore.

[frankenherder]

Wow, did a search in the forum and I am having the exact same issues!
Do we have to manually delete back ups or is Veeam supposed to do it? I have a 4 day retention period but they do not get deleted automatically after 4 days.
I want to get this going where every 4 days a full back up is done, then have 3 days of incremental backups. When the next full backup is done, then I want all the previous backups deleted automatically. Is this possible?

Matt

[Gostev]

No, this is not possible, as retention policy will always keep the set amount of restore points.

With incremental backup mode, this also means that previous full backup chain will not be deleted until the latest incremental backup falls out of scope of retention policy, in which case the whole chain will get deleted. Effectively, this means that in some cases there will be more backup files on storage than defined by retention. By the way, reversed incremental backup mode does not have this "issue".

One thing you could do is delete all previous restore points manually (or with script) as soon as new full backup is taken, however I am not sure why would you want to do that - what if you need to restore from restore point other than the latest, because it had captured some corruption?

[foggy]

Matthew, please consider this topic for more detailed explanation of how retention policy works for incremental mode. Thanks.

[tsightler]

Wow, did a search in the forum and I am having the exact same issues!
Do we have to manually delete back ups or is Veeam supposed to do it? I have a 4 day retention period but they do not get deleted automatically after 4 days.
I want to get this going where every 4 days a full back up is done, then have 3 days of incremental backups. When the next full backup is done, then I want all the previous backups deleted automatically. Is this possible?

Actually, just set your retention policy to 1 restore point. Then, whenever a full backup job is done the previous backups will be deleted. When the incrementals are run it won't delete the older restore points because they wouldn't be any good without the full backup, but when the next full is run the previous fulls and all incrementals will be deleted starting the chain again.