Help with Retention and Deleting Old Backups

[jscooper22]

Hi,

I know this has been beaten into the ground here, and those of you who "get it" are fed up with those of us who don't, so apologies for being yet another one who doesn't, even after reading the documentation a few times. I did try the calculator(s) and even they seem more complicated than I needed or have time to decipher.

After days of reading (and not grasping) Veeam way of doing things, I think I'm finally getting it, not not entirely. My first problem was it took me a while to realize that "Forever Forward Incremental" is a method/result of other settings, not an actual checkbox for "what kind of backup would you like to run" that I just wasn't seeing.

I'm using Veeam to back up a couple VMWare servers, each with four VMs, for my company.

I'd like to run a full backup on Saturdays when the office is quiet. I'm fine with a Synthetic instead if that's the best way to do it; I just want to be sure that on Saturday, I get a snapshot of what there is NOW.

Then, on Wednesday, I'd like to run an incremental backup. Since this is during business hours want it to be faster and "doing less" with server and network resources.

Now, sometime Thursday or Friday, I can manually copy the backups (one full/synthetic with one incremental) to a drive I take off site. I'll have a snapshot of my VMs as of Wednesday. This seems simpler than the whole "rotating storage" method.

After the following Saturday's full (or Synth) backup runs, I'd like the previous full and incremental backup to be deleted.

This way, after Saturday's backups are done, I'll have a copy of everything as of Wednesday out of the building (that I took Thursday or Friday), and in the building I now have a copy of the new Saturday, ready for the next incremental the following Wednesday, and the cycle continues. So:

Sat: Full Backup
Wed: Incremental
Thu: Take off-site
Sat: Full Backup; delete old
Wed: Incremental
Thu: Take off-site
Sat: Full Backup; delete old
etc.

Is that something Veeam can do? How?

Thanks,

Jeff

[Mildur]

Hi Jeff

Theoretically it can be done.
You can configure the job schedule to only run on Saturday and Wednesday.
https://helpcenter.veeam.com/docs/backu ... ml?ver=120
Synthetic full must be configured to Saturday:
https://helpcenter.veeam.com/docs/backu ... ml?ver=120
And retention to 1 Restore Point. After the new full backup was written, veeam can delete the older chain.

But please allow me asking the question, why not keep it simple and do a backup each night when your employees most likely don‘t work? You could loose 2-3 days of work when you backup only twice per week. For most organizations loosing 3 days is really expensive.

Now, sometime Thursday or Friday, I can manually copy the backups (one full/synthetic with one incremental) to a drive I take off site. I'll have a snapshot of my VMs as of Wednesday. This seems simpler than the whole "rotating storage" method.

This could work, but it’s a lot of manual work. Using backup copy jobs sounds much easier because it‘s automatized and integrated in the veeam console.
Also you have to import backup files on the veeam server if you want to restore from them.

My suggestion, reconsider your backup design. Do daily incremental backups with a synthetic full on Saturday.
Use backup copy jobs to copy the backup files to rotated external usb disks (or object storage).

Best,
Fabian

Best,
Fabian

[jscooper22]

Thanks. I want it this way for a couple reasons:

My backups strategy (which has worked well for years) is a series of overlapping methods and platforms, one of which is physically taking drives home with me each week. Yes, I do backup offsite to a remote host nightly, but if our storage provider is down/unreachable/etc just when I most need them, I at least have the whole company through last week on a couples drives in fireproof cases in my closet. This being the case, the storage space is limited since I have file backups, email backups, databases, web server files, etc. on the drives. Because of this limited space, I want to keep the backup files as small/simple as possible.
The Veeam backups are for "a meteor hit the building or the servers just got taken down by ransomware and we need to rebuild everything." Our DC is really a password server since we're 90% Mac, only our other servers are generally "on the domain;" server configurations don't change much; and with fewer than 100 users I can fairly early put things back to right if all I have is 3-4 day old backups.
Many of the actual files, which are frankly more critical, (documents on the file server, email, databases, user documents from our RDS server, etc) are backed up nightly, and also included in this offsite disks. These are my "I accidentally deleted something and I need it back."

Also, I don't like the idea of a giant process wearing on my drives all night, nevermind that sometimes people work late/from home, and I don't want to slow down their disk access or the network if I don't have to, it can't be good to constantly grind on HDDs (SSDs are less of a concern thought the storage itself is not quite a reliable so there's that), so I spend the evenings/nights backing up the aforementioned actual files. I like it this way because I'm not dependent on any app/platform/service to get my files back. I just copy them from the backups to where I want them.

I first tried to use windows server backup (and a script as a chron job on my linux servers) to do this, but the windows stuff was not being reliable and I figured if I need Veeam to get a windows backup I can trust I may as well use it for everything.

It looks like a mess, but if any one "system" (cloud, drives in house, etc) fails for some reason I have the others. My theory is no ship is unsinkable; all I can do is reduce the chances of losing everything.

Thanks,
Jeff

[david.domask]

Hey @jscooper22,

Your idea is clear, a few comments.

- Synthetic Fulls likely are what you want if you want to avoid disrupting your users, but the workload shifts from stress on production (long snapshot time) to the repository. This can be mitigated if you are in a position to format as XFS/ReFS which gives you reflinks/block cloning, but if I get you right, you're using USB disks, and I don't think it's a good idea to run such file systems on USB. (IIRC, Microsoft doesn't support ReFS on removable drives, and there was a pretty spooky bug that made ReFS volumes on external drives show up as raw a bit back)

- If it's really USB disks, likely the Synthetic performance is going to be unbearable, and just from what I've seen in cases, the chassis hardware typically goes due to heavy random write IO, so your concern is warranted here. Regrettably this means Active Fulls which puts a wrench in the works for your need to avoid impacting users

It sounds like you've got a fairly modest backup footprint storage wise, so I'm curious if you've got the space available for a short term "proper" storage (linux or windows), and then maybe backup copy to your USB drive(s) to meet your off-siting needs. This gets you Active Full backups to the drive you off-site without stressing production, and the "proper" storage likely can handle XFS/ReFS.

Just some points to consider.