Hello!
A great insight into sizing Veeam proxies is to use the guide below:
Or the calculator for this purpose:
However, it is important to understand the basic concepts behind the calculator.
These are the assumptions for sizing with examples of a hypothetical environment:
- O = Total data source: 50 TB
- J = Backup window: 8 hours
- A = Daily change: 2%
Where:
- O = Source
- J = Window
- A = Change
- T = Throughput
Conversion formulas:
- Convert TB to KB = (TB * 1024 * 1024)
- Hours to seconds = (HRs * 3600)
- Throughput in Cores FULL = Throughput / 100
Right!
So our converted environment is:
- O – 52428800 MB
- J – 28800 SEG
- A – 0.02
The calculation for processing the FULL backup would be:
- O - 52428800 MB / J - 28800 SEG = T - 1.820 ~
- T - 1.820 / 100 = 18 ~ CORES
That's it! Our proxy needs 18 CPUs.
And the memory?
We can consider 2 GB of RAM for each 1 CPU.
So that would be 36 of RAM.
Our proxy needs to have 18 CPUs and 36 RAM.
For a more in-depth study, use:
https://bp.veeam.com/vbr/2_Design_Structures/D_Veeam_Components/D_backup_proxies/vmware_proxies.html
Consider:
✅Physical proxies: at least 2 servers for contingency;
Virtual proxies: consider several proxies with 8 vCPU to send CO-STOP issues;
Pay close attention to the chosen transport mode, depending on the Datacenter infrastructure, it can change drastically;
It is recommended to have the proxy server as close as possible to the source data with a high-bandwidth connection;
If you have backups and replicas, consider sizing the proxy to process the replicas as well;