Thoughts / experience around a GFS backup solution

[sunnydays]

Hi Guys,

Just wanted thoughts around a GFS backup solution. I am sure this gets asked a lot, but the more and more I use Veeam, I find caveats on how things should have been done! and making changes mid way through jobs tends to complicate thing.

A little bit of background, we are currently running a Forward Incremental - Forever and a 28 day retention across 12 jobs. This was recommended as it was the most storage efficient method.
We have a 2 x NAS devices (4TB 7.2k SATA RAID5 per volume, 2 volumes per NAS) running Windows 2012 with Deduplication enabled. This was running fine until deduplication kicked in and merges and restores took a very long time.

I am planning to implement a GFS backup solution the old way and introduce a dedeplication device and this is what I have planned. The deduplication device will most likely be a Quantum DXi4701.

Daily:
•Repository: NAS
•Retention: 7
•Backup Mode: Forward Incremental – Forever Mode
•Create Active full backups periodically: Monthly On Second Friday
•Schedule: Daily at 9:00pm Everyday

Weekly Copy:
•Copy Every: 7 Days at 9:00AM
•Virtual Machines: Daily Job above
•Repository: Quantum
•Restore Points: 7
•Keep the following restore points for archival purposes: Enabled
•Weekly Backup: 4 (or 5 if that is recommended??)
•Schedule: Weekly Backup: Saturday 8:00AM (will create the full backup from Friday nights backup)
•Schedule: 9:00am – 5:00pm

Tape:
•Backup Files: Backup Copy Job above – Latest file in chain
•Media Pool: Tape Drive
•Schedule: See question below.
•Process incremental backup files: Disabled (I wanted just the VBK rather than the chain).
•Schedule: Monthly 9:00am Fourth Friday

I have a couple of question or concerns,

- I understand that when the daily job runs and the monthly active full kicks off, there will be an overlap of 14 restore points for that 2 week point?
- What do I configure the schedule to be for the tape job? With the Forward Incremental – Forever Forward, I would select Monthly on the Fourth Friday to ensure that it created a virtual full backup on the fourth Friday. But considering I am trying to move the Weeky 4 VBK from the copy job, what do I configure?
- I am planning to distribute Active full backup jobs across Friday and Saturday, as I am concerned with full backups taking too long and incrementals or copy jobs are missed (6 x Fulls and 6 x Incrementals will run on Friday vice versa for Saturday).
- I have concerns with the backup copy jobs not finishing in time before the next backup? How will this affect things if it was a active full or an incremental backup job?
- When the backup copy job merges the 1x full and 6 x incrementals into a weekly full, what happens the following week? Does the weekly full become the new full and 6 x incrementals will chain with it?

Thanks for the assistance!

Cheers,
Sunny

[foggy]

Hello, Sunny. Your intended move to using backup copy for backups archival is a recommended way, indeed, coming along with our reference architecture.

- I understand that when the daily job runs and the monthly active full kicks off, there will be an overlap of 14 restore points for that 2 week point?

The fact is you cannot set up periodic active fulls for incremental forever backup mode. Once you configure periodic fulls, you automatically switch to regular forward incremental. And with regular forward incrementals, you will have a whole month of restore points prior to creating an active full. After reaching the number of 14 restore points after this active full, the entire previous chain will be deleted from disk.

- What do I configure the schedule to be for the tape job? With the Forward Incremental – Forever Forward, I would select Monthly on the Fourth Friday to ensure that it created a virtual full backup on the fourth Friday. But considering I am trying to move the Weeky 4 VBK from the copy job, what do I configure?

You can schedule it the same way and provided virtual full is scheduled on the same day as weekly restore point, you should get only GFS restore points on tape (however I'd check this to be sure).

- I am planning to distribute Active full backup jobs across Friday and Saturday, as I am concerned with full backups taking too long and incrementals or copy jobs are missed (6 x Fulls and 6 x Incrementals will run on Friday vice versa for Saturday).

That's a good point in case you're tight on processing resources and not all of the jobs are executed in parallel.

- I have concerns with the backup copy jobs not finishing in time before the next backup? How will this affect things if it was a active full or an incremental backup job?

Even in case of active full backup copy job will still copy only changes occurred since last restore point. If the backup copy job isn't able to copy all the required data, it will resume data transfer during the next sync interval, copying all the blocks required to build the new restore point corresponding to the VM latest state. Previous restore point (the one the connection drop occurred on) though will be left in incomplete state.

- When the backup copy job merges the 1x full and 6 x incrementals into a weekly full, what happens the following week? Does the weekly full become the new full and 6 x incrementals will chain with it?

Backup copy does not merge anything to get weekly full, GFS works in a bit different way.

[sunnydays]

The fact is you cannot set up periodic active fulls for incremental forever backup mode. Once you configure periodic fulls, you automatically switch to regular forward incremental. And with regular forward incrementals, you will have a whole month of restore points prior to creating an active full. After reaching the number of 14 restore points after this active full, the entire previous chain will be deleted from disk.

Hi Foggy, thanks for the reply! Sorry, Yes still getting used to the terminology! I guess the problem I was trying to avoid was, we had a Forward Incremental - Forever with a 28 day retention we found out that 2 x VM's were corrupted in the job but 28 days had already passed. I only realised it when I tried to push it to tape and it couldn't create the virtual full. So as recommended by support I ended up copying all the backup files to tape. So I thought introducing this, would ensure that the full backup file was "reset". Is there any other recommended ways around this? Would prefer to not have a long chain.

[foggy]

A recommended approach is to use SureBackup for backups recovery verification.