Another Hardened Repository Post

[nunciate]

It is that time again. My backup storage in my DR location is filling up, so I need to purchase a new server. The server out there is also super old, so it needs to be replaced.
Just wanted to throw this out there for comment to see how I could best handle this.

Current Setup:
VM hosting all Veeam B&R jobs running in DR (Controls both Prod and DR Proxies)
Prod DC: 1 HPE Apollo with about 300Tb of storage. Running Windows with FC backend connected to my SAN and my tape drives. This is my main proxy server using SAN mode jobs. I also have a VM Windows Proxy to handle replications.
DR DC: 1 HPE Apollo with about 200Tb of storage. Running Windows with FC backend connected to my SAN and my tape drives. This is my main proxy server using SAN mode jobs. I also have a VM Windows Proxy to handle replications.

Future State:
Keep the VM in DR running all my jobs.
Replace Prod DC Apollo with a new Apollo and the same 300Tb of storage. Build a new Apollo using Hardened Repo ISO. Purchase a secondary physical server to use as the physical proxy. I must have this to keep my FC connectivity to my SAN and tapes.
How would you all connect the hardened server to the physical proxy? I am guessing 2x10g connections from each should do the trick. I don't want all the extra traffic between the 2 servers on my switches, but I think it will be fine. Does anyone have any suggestions how how to configure the NICs and such to optimize this?

Once Prod is set up, I will rebuild the old Prod server and send it to DR to replace the one there, and set this up the same way with a new physical server as the proxy.

[HannesK]

Hello,
while 2x10Gbit sounds on the lower end for 300TB. That setup would be the bare minimum I would consider, yes.

I'm not sure what is meant with "I don't want the extra traffic on my switches", because that's where one would connect servers. Direct connections between infrastructure machines don't work because the backup server then cannot see the components anymore. And I would avoid having multipole networks connected to one server if not required.

Depending on the switch infrastructure, I would use LACP or active / backup links with the two network links.

Best regards
Hannes

[nunciate]

Yeah that wasn't very clear. By not on my switches, I am referring to the fact that my physical Windows storage server today uses FC backend, dedicated FC Switches, and SAN mode to pull data onto the storage. The new path will require networking switches so that the proxy server can communicate with the hardened repository. We have limited 10 Gbps here currently but plan to expand that in the near future, though.