Veeam Users Please advise

[petben]

1. I am finding my Veeam B&R incremental backups are very large.

 

e.g.

VM File server = 2000GB

daily change in Windows disk size = 1GB approx

Veeam daily Incremental backup size = 25GB

 

VM Ubuntu = 11GB

Veeam some of the daily incremental backups = 2 or 3GB

nothing much changes on this server, a few logs..

 

Veeam support say this is 'expected' but why?

 

2. If I drastically change the schedules to multiple shorter term jobs, am I expecting a month incremental file to be 30 x the current daily incremental backup size?

 

 

 

3. Should I take some full backups too? On the 2000GB file server this obviously has a massive disk cost. Veeam support say forever forward incremental is not recommended.

 

---

 

can you please advise?

[Davit2005]

If a full backup is needed how far will you have to go back? What is your backup storage plan i.e. Backups stored to cloud only, off site, seperate building etc? Edited March 18, 2019 by Davit2005

[localzuk]

Your daily change isn't just "size", it is files editing/changed. So, 25GB of files have changed, even if only the disk increases only 1GB, there will be a lot more files that have changed.

I would say that in today's "large disk" world, 25GB a day change is not a lot, and 2TB is also not a lot of stuff. A NAS with 24TB of RAID 6 storage would be £2300.

 

To give some comparative numbers - our H drive VM has 2.5TB of data on it, and the incrementals are roughly 38GB a night (uncompressed). For our entire setup, on 7.4TB of data leads to 174.6GB of uncompressed change, compressed down to 42.4GB each night.

Edited March 18, 2019 by localzuk

[jtotheb]

Your daily change isn't just "size", it is files editing/changed. So, 25GB of files have changed, even if only the disk increases only 1GB, there will be a lot more files that have changed.

 

This. 15-20 Gig daily change is about the average amount on a server of a slightly smaller size.

 

 

3. Should I take some full backups too? On the 2000GB file server this obviously has a massive disk cost. Veeam support say forever forward incremental is not recommended.

 

It's difficult to comment without knowing your exact setup, but I would. Or look at reverse incremental in Veeam. Your most recent backup is a "full" backup and it creates incremental points to go back to. Fairly modest disk space requirements as you're storing one full and X number of points. Downside is it's very disk intensive and not really suitable if you're uploading to online backup storage as you'd be uploading a full backup everyday. Great for local (i.e. onsite rather than local to the server!) though even if it will chew up a few disks.

Edited March 18, 2019 by jtotheb

[petben]

Thanks for the replies, so to conclude I was thinking of the following setup. As you cant have multiple schedules on the same job I will need to setup multiple jobs (which is crazy).

 

DAILY

Everyday

7 restore points

1 full initial backup, no further full or synthetic backups

 

 

MONTHLY

First of the month

12 restore points

1 full initial backup, no further full or synthetic backups

 

 

Then as the uplink to the backup server from the virtual network is 10G and its a high spec server i was thinking of the reverse incremental backup (as recommended above)

 

BTW we already have 30 days worth of network folder backups via DPM, so this would be for VM restore (using the daily backup) or further back in time user file restore (using the month backup).

 

I am totally up for changing the plan if there's any ideas?

[jtotheb]

MONTHLY

First of the month

12 restore points

1 full initial backup, no further full or synthetic backups

 

That might help you visualise how the reverse incremental works. The disk churn on a monthly reverse incremental is going to be pretty gigantic. It would probably work from a purely technical perspective, but you might want to raise doing it at such intervals with Veeam themselves to sanity check it.

https://www.veeam.com/kb1933

[KevinB]

The disk churn on a monthly reverse incremental is going to be pretty gigantic.

 

It's not so bad. Our reverse incremental full is about 7TB in size and each month about 1.5TB of that is changed and so shows up in the diff. The job starts at midnight and is complete by 05:30 so it doesn't take an inordinately long time to run.

 

Weekly diffs are about 350GB and take about an hour, dailies average about 120GB and take about an hour as well.