Repository & Proxy Sizing with Veeam Replication

[jstajay]

Hi,

How do we size Veeam Repo when we perform Veeam Replication for VMware VMs?

Would like to know how many & how much Repo space required for 300 VMs ~ 250 TB Data.

Also # of proxies needed on both Main & DR site & Bandwidth needed.

[HannesK]

Hello,
and welcome to the forums.

As far as I know, nobody ever created a formula for repository space for Replication. The number is so low, that nobody seems to be bothered around that. If you need a number, then I would start with 500 GB.

Proxy resources are the same as for backup. You just need them on source and destination. For bandwidth, I would apply a generic 2:1 compression ratio. If you know the amount of data you need to transfer with each replication cycle, you can calculate the required bandwidth. If you have zero idea on change rate, then I would start with 5% per day.

Best regards,
Hannes

[jstajay]

Thanks, Hannes.

Also, I used the formula to calculate the number of Proxies needed available on this link https://bp.veeam.com/vbr/2_Design_Struc ... oxies.html

Using this one it says I need 10 Proxies. This means 10 Proxies on the source & 10 on the Destination. Are my calculations correct?

[HannesK]

having the same amount of proxy resources on source and target is correct, yes.

I don't know whether you calculations are correct, because I cannot see them

Overall, 10 proxies sounds too much for only 300 VMs / 250TB. 10 are possible, but I would probably go with less machines and more CPU power per proxy. It sounds like you plan to use virtual proxies. In that case, you can also start with 10 and then reduce them if you see they idle around.

For tasks per core: with modern CPUs you can go higher than 1:1. I don't know, when the best practices guide will adopt that. But in V12, the user interface will be "green" up to a 2:1 ratio

[HannesK]

Hello,
thanks for sending me the calculation.

- Source data: should be 250 TB, not 300 (it's 300 VMs as far as I see)
- there is a difference between first (full) replication and the other incremental replication cycles. If you want to transfer 300 TB in 8h, that needs almost a 100Gbit/s link... but you need much less, because after the first full, it's only incremental data (change rate)
- incremental replication (everything after the first full): if you have 5% change rate with 250TB data, that's 12,5 TB per day.
- bandwidth: apply data reduction. 12,5TB reduced to 50% is around 6 TB data per day

For virtual proxy servers, it is recommended to configure multiple proxies with a maximum of 8 vCPUs to avoid co-stop scheduling issues

well, it's 2023 and things have improved since that sentence was written. I disagree with it, but I cannot change the sentence. I would just avoid using more vCPUs than you have physical cores per CPU in the ESXi host.

Best regards,
Hannes

[jstajay]

Thanks, Henne, really appreciate your inputs.