Son-Father-Grandfather rotation suggestion

[airzyk]

just an FYI I've been able to implement a GFS rotation in the post job activity using the scripts provided by Vladamir in this post: GFS rotation post job script

[bgutwein]

This may have already been discussed but what is the preferred way for setting up backup jobs for retention policies? For instance I have a job of 30 VMs that I want to keep 2 weeks of daily incrementals and then fulls every week for 4 weeks, then every month for 12 and then every year. Do I need to create 4 different jobs for this or is there some way to adjust the settings for just one job to accomplish this? Thanks!

[bbricker]

You can expect this feature sooner rather than later.

V7?

[airzyk]

This may have already been discussed but what is the preferred way for setting up backup jobs for retention policies? For instance I have a job of 30 VMs that I want to keep 2 weeks of daily incrementals and then fulls every week for 4 weeks, then every month for 12 and then every year. Do I need to create 4 different jobs for this or is there some way to adjust the settings for just one job to accomplish this? Thanks!

Brad, the best way to do that would be to utilize the post job activity powershell scripts that I linked to in the post above yours. Otherwise that would necessitate 3 jobs with 3 different retention policies and you would have duplicate full backups at times, not to mention trying to figure out the scheduling so the jobs don't run at the same time

[johnlong]

V7?

One can only hope!

[Gostev]

Chances are good that v7 will have this feature. Just not where most of you are expecting it to be implemented...

[dellock6]

My guess, disks-based backups will go on with usual retention points, tape backups will allow for GFS or like rotation schemes. It would make sense to me.

Luca.

[Gostev]

Your guess is not right, Luca

But regarding tape, my understanding is that long term tape rotation is normally handled manually due to the fact that tapes are typically stored offsite. E.g. bag with 10 tapes holding yearly backup is manually taken to the vault and kept there for required times (3 years). Then, it needs to be manually taken back and tapes inserted in the library before these tapes can be reused.

Our tape support will still have retention, but it is pretty basic, more of a fool-proof type retention (overwrite protection period on media sets).

[dellock6]

Well, I said it was a guess
Yes, for sure tape management will be something to be managed "manually", since once we have short term rotation on disk, tape will be used for long term, with sometimes tapes that will not be overwritten, but simply written once and then stored for ages.

Luca.

[johnlong]

Chances are good that v7 will have this feature. Just not where most of you are expecting it to be implemented...

I would just like a more intuitive scheduler and/or retention policies for disk based backups. However with the amount of dedup I am seeing with server 2012 dedup it's almost not an issue, I could probably take weekly fulls for a year and still have enough space!

[jp34]

Hi all,

I want to backup my VM every day (1 increment Mon to Sat) and 1 full on Sunday.
I want to keep 6 increments and 2 fulls, so I have to keep 8 restore points on disk. OK.

Now I would like to keep 1 full per month besides. How can I do ? Create a new monthly job ? other idea ?

Thanks in advance.

[veremin]

How can I do ? Create a new monthly job ? other idea ?

In fact, there is no GFS functionality present in the current version of Veeam Backup and Replication. So, you can either utilize this PS script that i've created previously or wait till the version 7 with backup copy job will be released.

The scenario with additional monthly job should also do the trick.

Hope this helps.
Thanks.

[jp34]

The scenario with additional monthly job should also do the trick.

I don't really know its algorithm but Change Block Tracking can work with 2 jobs on the same VM ?

[veremin]

Yep, without any issues. This is explained in the sticky FAQ topic. Thanks.

[mrstorey]

I've been asked about configuring a GFS retention strategy using Veeam, and understand this feature will be available in V7, albeit in the form of a backup copy job.

We do intend to use v7 accelerated backup copy jobs anyway, for offsiting our backup repos (their intended purpose!).....but I wondered if it's viable to restore entire VMs to the live site directly from a remote repo? We're talking about sites separated by 10mb MPLS links averaging 60% utilization.

I guess that remote repositories are intended for and best used in disaster scenarios, where you would restore entire VMs locally at the DR site - but in lieu of having GFS retention features at the live site, would it be viable to use them as general restore points at a live site?

I suppose I'm wondering if the WAN acceleration works in both directions - I understand its used to get backed up VMs offsite very efficiently, but can the technology be used in reverse, to restore VMs over WAN connections efficiently too? I read other post suggesting file level restores using this method would be viable, so I'm hoping VM restores are also!

Thanks for your help!

[foggy]

Alex, nothing prevents you from using backup copy jobs locally to implement GFS retention in your primary site on cheaper storage with lower cost per TB.

[Gostev]

No, restores from offsite are not currently accelerated. The value of applying WAN acceleration to restore was not seen as very big, considering that reference v7 architecture includes primary backup repository in every site containing the latest restore points. This makes recovery from the offsite is a rare occurrence (only require when production site backup storage is entirely lost). But in this scenario, even if restores were accelerated, you would not get actual benefits, because loss of primary backup repository also means there is no acceleration cache available.

Besides, such event means you have to transfer ALL backups back fast to be able to perform on-going restores? For this purpose, shipping a storage device is a much better option. And when it is not possible, the only other thing left to do is just setup WAN accelerated Backup Copy job in the reverse direction, and wait...

I guess I can see the value of accelerating restores when you only have a primary backup repository (short-term) in the main site, and your long-term repository is only offsite. In this case, you may want to restore older restore point from off-site, even when your primary backup repository is fine. But, this is questionable deployment overall - why keep most of your backup data offsite? Why not simply have GFS repository in the main site? May be there are few scenarios for that (there is a scenario for everything), but it is certainly uncommon, making it low value for the product overall.

But yes, single VM restores scenario from off-site repository are efficient (just as efficient as FLR over WAN), thanks to our architecture with data movers on the both sides of the wire.

[mrstorey]

Of course! I didn't consider using accelerated backup copy jobs to the same site in order to make use of GFS retention.

We have 3 branch offices in APAC - and one of these is a larger 'hub' where we've placed some key services like exchange.

I considered also using this site as a 'hub' for all APAC backups - receiving the backup copy jobs from the other two and committing them to tape. Meaning we'd only need tape hardware at one location - I understand a feature to commit backup copy jobs to tape *might* be coming in future point releases of v7 - which would be great!

Thanks for your replies - and it's comforting to know that restores from backup copy repos will be possible - even if this is an edge case!

Thanks,
Alex

[veremin]

Meaning we'd only need tape hardware at one location - I understand a feature to commit backup copy jobs to tape *might* be coming in future point releases of v7 - which would be great!

It will be possible even with the version 7.

You will have to deploy additional instance of VB&R server at “Hub” site, where the tape hardware is installed, and then use the “files to tape” copy job in order to regularly copy files produced by backup copy jobs to tape devices. Though, in this case you will lose tracking functionality that allows you to determine what VMs reside on given medias, what medias are required to restore certain VM, etc.

Thanks.

[ggillette]

I'm very excited to see this has been added to v7! We've been using a dirty workaround for quite some time.... We have scripts set up to manually copy VBK's to different storage, where it is stored for ~ days and deleted by another script.

[veremin]

Hi, Greg. Yep, the v7 release with the new backup copy job should answer your requirements, handling efficiently the process of copying backup data, as well as, implementing long-anticipated GFS rotation scheme. Thanks.

[joshandy]

Case # 00442758

I am hoping this post helps a lead to a new feature which allows grandfather , father, son backups in 1 job. Currently to achieve daily/weekly & monthly backups -one must create different jobs and thus duplicating data. When backing up across a WAN link, this is also double data transfer . However this feature is available in veeam copy I hear.
This feature is out the box with Acronis and some other backup solutions.
Why the Veeam developers have neglected to include this in backup job schedule when it is available in copy seems quite bizarre and the benefits of such a feature do not need much explaining.
Our current SAN costs £2K to add a £450 disk, and with a premium cost for storage, we need Veeam to be smarter and and allow more sophisticated scheduling via the GUI.
I had heard using Powershell cmdlets it might be possible in 7.0, but having just checked this with a senior engineer at veeam I now realize that the best I can hope for is a feature request, and the best way to get that done is to post on here.

I was told by Veeam engineer the best way to request this change was by forum, which I find slightly odd and it assumes we are loaded with spare time !