Overkilling Backup Strategy?

[braddock]

Hi, I would appreciate some feedback on our potential backup design.

We would like to achieve the following:

For our servers we plan to have one backup job that run an incremental every night with an Active Full on Saturdays. (so 7 incrementals and one full)

We will also be verifying the backups with a sure backup job each night (or maybe just once per week)

Now, because of the backup retention (1 year) that has been requested we will have a backup copy job that copies the main backup to a different repository each night, 7 incrementals, 4 weekly, 12 montly backups. This of means that we will have ~17 full backups over the course of a year. We will also be backing up to tape for offsite backup but management wants to be able to go back 1 year without requesting tapes. One full set of backups compressed are only around 1.8TB but with all the full backups combined that makes approx 30 TB required to store 1 years worth.

Are we overkilling it here (or errors in the above) and would an alternative backup strategy will less full backups make more sense?

[DaveWatkins]

For what you're proposing a dedup appliance seems like the perfect fit for your backup copy location, just make sure you push data to tape from your primary repository not the dedup device

[Shestakov]

Hi braddock,
The strategy looks good and stable.
You can save space on full backups switching primary job to forever forward method(disable active fulls). It should work fine with surebackup.
Also schedule backup copy GFS job to run say weekly on Sunday and monthly on the last Sunday, you will have less job runs and less backup files. Same backup will be marked as "weekly" and "monthly".

[nmdange]

If you don't have a huge amount of change data, you may be able to save space by keeping a months worth of incrementals and only keeping monthly full backups. I like the option to create synthetic fulls periodically and then transform previous backup chains into roll backs. This helps reduce restore time for the most recent backup when you have very long incremental chains.

Also instead of a dedup appliance, you may be able to use the native deduplication provided by Windows Server 2012 R2. Although it will be work much better with Windows Server 2016.

[braddock]

Ok thanks all, some good things to think about here, unfortunately we don't have a dedupe applicance.

i thought maybe we were doing something wrong given the high size of the combined backups, but it sounds like we are doing it right. Before we went with Veeam we had Datto recommending their appliance that only holds 10TB, and that was with a years retention. Is anyone familiar with how Datto works, could we replicate their 'backup strategy' with Veeam, or was the appliance massively undersized?

[Shestakov]

braddock,
could you give us more details of Datto`s backup recommendations?
What`s the size of VMs to backup? How many of them do you have?
Thanks!

[braddock]

Hi,
We have 15 VM's and the total size used is 3.5 TB. The reseller recommended we needed a 7TB storage appliance for that and the nearest model was 10TB. I'

I found out how Datto works via a reddit search:

The backup uses a "chain", meaning it takes a full backup and then "incrementals" after that. It then consolidates those changes daily, weekly and monthly. If links of the chain break you can run into issues that require remediation."

[braddock]

Hey ,I should add that happy with Veeam , just curious if there is a way we can optimise our Primary backup location storage to have less full backups since we will be using surebackup and backup copies to a second location (thanks to everyone above!)

[kte]

Wy not use a storeonce vsa vm on local disks on a hypervisor?
I think it starts at 4 TB

[Vitaliy S.]

If you need to optimize primary backup job location storage, then I would follow Nikita's advice on forever forward incremental backup usage, however running active full backups once in a while is also a good idea.

[nunciate]

What are you currently using for Production and Backup Storage today? We use an HP server with SAS attached disk arrays. We run Windows 2012 R2 and use the built in deduplication. 2012 Deduplication works great with Veeam. You can save a ton of space if you are doing weekly/monthly active fulls.

I agree with others though. Use Weekly/Monthly synthetic fulls with reverse incremental to save the most space.

[nunciate]

Just thought I would add something else. If storage is a premium for you perhaps storing longer term backup data on tape but keeping a copy onsite and offsite would be good.
It would take longer to restore but you would at least have the tapes local so you don't have to recall them from offsite.

[braddock]

Thanks for the help all, you've given us some great pointers.

[zoltank]

On a related note, if you have the time and resources, I'd run SureBackup after every backup. It's how I do it, and I sleep easier at night getting the SureBackup email telling me my backups were confirmed as viable.