estimating storage size for backups

[dmitri-va]

Hello,

I'm testing veeam backup in our virtual environment and trying to estimate a required backup storage capacity, so we could procure it.

Here's what I did:
ran a veeamone report which shows me actual disk usage per VM
ran a test backup job for 5 linux and another job for 5 windows VMs (full backup with 'lan target' for dedup and 'optimal compression').
Here are the true data reduction ratios I currently observe:
Linux VMs: 135.91GB used space / 60.2GB backup size = 2.26
Win VMs : 367.46GB used space/253.7GB backup size = 1.45

I take our total used space by all 147 VMs - 11,913.15GB (50/50 mix between linux and windows) and divide by average ratio based on above of ~1.86 = 6,405GB

We need 30 days retention. We plan to do daily incrementals and weekly full backups, so 4x6,405GB + 24x640GB (assuming cbt based incrementals are ~10% of fulls) = 40,980GB required for the backup storage

Does this look right?

Also, does this pretty small data reduction ratio look OK/average?

[dellock6]

Our system engineer Timo Dewin wrote this great tool to estimate backup storage size consumption:
http://rps.dewin.me/
I don't agree with Timo usage of 10% as daily change, I think on average 5% is more common, but you can setup all the parameters to see the final consumption of any environment.

[dmitri-va]

Our system engineer Timo Dewin wrote this great tool to estimate backup storage size consumption:
http://rps.dewin.me/
I don't agree with Timo usage of 10% as daily change, I think on average 5% is more common, but you can setup all the parameters to see the final consumption of any environment.

god tool, and pretty much confirms my estimates. Do you think 50% reduction in there is a good number? (Close to the average of 1.86 I've observed). The 6TB of 'work space' is new to me though. Is this a percentage of the total data stored?

[dellock6]

Yes, 50% reduction is what we usually see on the field, and remember the best way to use the tool is as an "estimator", so even if numbers are higher than reality, it has the advantage of not giving you an under estimate and then sizing the repository too small.
Work space is the temporary free space needed for such operations like transforms, health checks. Also, plan to have at least the free space for an entire additional full, if you need to split a chain and create a manual full.

[dmitri-va]

thanks for the info!