We use Veeam to replicate 16 VMs every hour from our production site to our DR site. This typically takes about 16 minutes
We use Veeam to backup about 40 VMs each night. This typically takes about 3 hours for both jobs to run
One job backs up these VMs from our production site to our DR site
Another job backs up these VMs from our production site to an offsite Exagrid
We are using SAN mode and this is our current physical proxy server:
Five year old IBM 3550 M4
Two Intel E5-2630 processors, 6 cores each, total of 12 cores, 2.30 GHz
32GB RAM
Four 10Gbe connections
In the next month or so we are purchasing new servers, here is our initial specs for the new physical proxy server
Either HP DL380 Gen10 or Dell Poweredge R730
Two Intel Gold level processors. A few choices here
Assuming this is only a proxy and nothing else, chances are any of those CPUs will be more than sufficient. Whatever the core count is, multiply that by 2 GB to get the amount of memory you should use. If you want to get a better idea of how to size the proxy this section of the Best Practices Guide has a few formulas that will help.
You will find if you look at the BP guide you will only need max 2 proxies to do this, two will be more than fine, use 1 for redundancy. What is your backup window? If 8 hours you should have no problem. Main question is do you replicate 24/7 or just during business hours so your jobs don’t overlap. Remember to factor in OS resources and 1 proxy core requires 2GB ram per vm disk (task) running. No need to buy more than you need, are you planning on all using direct storage or NBD mode, sometimes it’s worth having a vm or two running proxy services for hot add and fast recovery services. Just thoughts
I will be using SAN mode for replication and backup. I will have this physical proxy server and one virtual proxy. I have been using the virtual proxy as the guest interaction proxy. We are actually purchasing a total of 8 servers, 7 ESXi hosts and one physical proxy server. I am leaning towards the 6142 since it has a higher clock rate which will benefit my VDI setup and the fact we are running some single threaded applications like Autocad Inventor.
I only replicate from 6am to 10pm M-F and do the backups overnight so jobs do not overlap