Host-based backup of VMware vSphere VMs.
Post Reply
CraigWatson
Lurker
Posts: 1
Liked: never
Joined: Feb 23, 2015 1:43 pm
Full Name: Craig Watson

Sanity-Check: Retention Policies

Post by CraigWatson »

Hi all,

I'm currently at the point where the world goes blurry and my eyes go crossed, so I'd appreciate a sanity-check on my understanding of Veeam's retention settings. My apologies if any of the below isn't clear - I'm attempting to brain-dump!

Some setup detail:
  • Veeam B&R v7
  • Backup Mode: Incremental with synthetic fulls (every Friday)
  • Current Retention Policy: 7 restore points on disk
My understanding is this:
  • One new VBK file will be created every Friday via synthetic full, and one new VIB file will be created every other day (i.e. Saturday to Thursday inclusive).
  • After every backup, retention policy is parsed. This is where things start getting hazy.
  • Because we have a 7-point retention policy and synthetic fulls are weekly, Veeam will wait until there's at least one full set of files (i.e. 1 x VBK + 6 x VIB) other than the last full before removing old files.
  • Million-dollar question: How many full backups (i.e. VBK files) will be on-disk at any one time?
My ideal behaviour: After the full, we will have the full that has just been created, plus the previous full and 6 x incrementals

Any help appreciated!
Craig
alanbolte
Veteran
Posts: 635
Liked: 174 times
Joined: Jun 18, 2012 8:58 pm
Full Name: Alan Bolte
Contact:

Re: Sanity-Check: Retention Policies

Post by alanbolte »

After the full, we will have the full that has just been created, plus the previous full and 6 x incrementals
Consider these two approaches, focusing on retention as executed immediately after a full:
8 restore points: This would be 2 full points plus 6 incremental points. The upside of this approach is that you guarantee you never have less than the points you want. The downside is that if you miss an increment you could end up with more points than you want, because the oldest chain would not be deleted until after the next incremental session. You must absolutely plan for 3 fulls with this approach.
2 restore points: This would be the most recent point, plus the previous chain. The upside of this approach is that you guarantee the oldest chain is removed before the full backup starts, and at the end of the full backup session you will still have the previous chain. You can safely plan for only two fulls with this approach.

Of course, we have to consider retention after subsequent sessions as well - with the 2-point method, the previous chain is removed after the next incremental, leaving you with just a VBK And VIB (2 points). With intermediate values like 7 points, hypothetically speaking you will only ever have 2 fulls, but there's always the possibility that a job will fail for one reason or another, leaving you with fewer incremental points than planned during the week, and then the old chain doesn't get deleted before the job starts a new full backup session. There may also be occasions where you want to run an active full backup in the middle of the week, such as when adding disks containing significant data to one of the VMs in that job, or if a VM had its MoRef changed and has to be re-added to the job.
veremin
Product Manager
Posts: 20270
Liked: 2252 times
Joined: Oct 26, 2012 3:28 pm
Full Name: Vladimir Eremin
Contact:

Re: Sanity-Check: Retention Policies

Post by veremin »

Additionally, you might refer to these visual aids for more understanding regarding different backup modes and how retention works in each case. Thanks.
Post Reply

Who is online

Users browsing this forum: No registered users and 62 guests