Sizing Veeam to backup 20TB with retention 10 years

[asazami]

Hi everyone
I just have new project coming to implement Veeam backup for our customer with some requirement below:
1. Backup ~190 VMs with total 20TB data , around 5% data change rate
2. Backup to local physical server with 31 retention point
2. Backup copy job to Azure for off-site with retention 7 restore point and GFS with 12 monthly + 10 yearly

I do some calculator with https://rps.dewin.me and have some result below:
1. With Forever incremental The local disk will need 34TB
2. with Synthetic full every Sunday the local disk will need 86TB
3. With backup job copy to Azure with 7 retention point and GFS with 12 months, 10 years will need ~237TB on Azure

Here are my questions, please let me know your idea
1. Will a physical server with 1 CPU, 32GB can adapt these amount of backup ( we'll deploy another 5 off-host proxy to offload for backup server)
2. Forever incremental & Synthetic full every Sunday have a big storage gap (34TB vs 86TB). Do you have any experience on the restore performance between forever incremental with 31 restore point & synthetic full every Sunday, if it's not too much different, we'll stick to forever incremental for cost saving
3. Do we have any solution to reduce the 237 TB of backup job copy? Will windows 2016 deduplication help us alot on this? 237TB seem too much because Azure only support maximum 4TB per disk

Thanks for your help

[veremin]

Will a physical server with 1 CPU, 32GB can adapt these amount of backup ( we'll deploy another 5 off-host proxy to offload for backup server)

CPU and RAM settings do not affect backup server ability to store huge backups anyhow. So, you should be OK.

2. Forever incremental & Synthetic full every Sunday have a big storage gap (34TB vs 86TB). Do you have any experience on the restore performance between forever incremental with 31 restore point & synthetic full every Sunday, if it's not too much different, we'll stick to forever incremental for cost saving.

If your backup target can cope with regular transformation of the oldest backup file, feel free to stick to forward forever incremental mode. Might be worth checking it first by creating a backup job with short retention and letting it transform .vbk file.

3. Do we have any solution to reduce the 237 TB of backup job copy? Will windows 2016 deduplication help us alot on this? 237TB seem too much because Azure only support maximum 4TB per disk

Windows 2016 with REFS seems like the best solution.

Also, Azure backup repository might not necessarily be the perfect target for such huge data set due to limitations you mentioned.

Thanks.

[asazami]

Thanks for your advice, we'll do planning for physical server with ~40TB DAS and do the forever incremental ( in the past I have another project to backup incremental ~50TB but with 14 restore point only and the performance quite good).

I didn't have experience with backup job copy for retention 10 years with GFS 12 month, 10 year. Do we have any solution to reduce the 237TB space need for 10 year with the calculation? If we can do the archive with incremental I think that'll be great.

[ChrisGundry]

GFS full backups with ReFS take up very little space because of block cloning.

We have used 23.7TB on the disk, but we are actually storing 87.4TB of backup files currently, this saving is due to the way ReFS works with GFS.

[veremin]

Do we have any solution to reduce the 237TB space need for 10 year with the calculation?

Correct, REFS-based repository. You can read more information regarding Fast Clone and "spaceless" GFS here. Thanks.