Architecture: Which components best on ESXi mgmt net?

[pkelly_sts]

Sorry for a basic question but due to an office move I'm currently re-architecting our proxy setup & something is making me think of a message I see occasionally which points out that some element, I think the proxy VM, is "not on the same network as the management network of the host" or words to that effect.
Could someone please remind me what component that is, and why it might be beneficial (or bad).
I'm thinking beneficial for restores which, I think, always go in via the console interface, but then if doing so via a proxy it doesn't help anything anyway as the proxy will still talk to the repository over the network anyway.
Then again it could be the all-in-one B&R/Proxy/Repos (physical) box which is sensible to have on a "normal" network rather than same as the ESXi managment network anyway.
Unless the underlying mention is simply along the lines of "doing it this way means you're always routing to the ESXi interface which might be slightly lower performing than if it were on the same subnet".

Does my question even make sense?

[PTide]

Hi,

Could you please describe you infrastructure? What transport mode do you use?

Thank you.

[pkelly_sts]

Primary site as follows:

3 x ESXi Hosts
Physical B&R, Repos, Proxy, FC-SAN-connected Server with SAS tape library & dedicatated FC-SAN backup storage (30Tb usable) as well as some internal backup storage (12Tb useable)
3 x Proxy VMS, 1 per host
Proxy VMs on dedicated & separate "Veeam" subnet with dedicated NICs on the hosts

4 Backup jobs, 1.7Tb full, 1Tb full & a couple of small jobs

[Vitaliy S.]

Hi Paul,

Since you have an FC-SAN and would like to re-architect your confiration, then I would recommend to consider using physical proxy servers with direct SAN access for most efficient backups. If you configure direct SAN access, then you would be able to do full VM restores through this mode as well.

BTW, what is your current bottleneck statistics for the backup jobs?

Thanks!

[skrause]

I run my veeam servers on the same network as our ESXi management networks and our vCenter.

Those three types of servers are the only ones that live on those networks.

It works pretty well and cuts down on the amount of firewall work I need to do.

[pkelly_sts]

The rest of our story is that we'll be replicating this site to another (and the other similar site to this one) via MPLS and, if possible I'd rather avoid advertising the ESX managment subnet to the MPLS.

After testing I narrowed it down to the message showing during replication jobs only "No available proxies are running on ESX(i) management interface subnet. Using proxies from a different subnet, performance may be impacted" so once I found that again I was able to search on it & found another topic on here (http://forums.veeam.com/vmware-vsphere- ... tml#p67597) which basically states that as the subnets are on the same site then performance shouldn't really be impacted. I guess it's there as a bit of protection from accidentally (or I guess purposely) using a proxy from one site to talk directly to a host on another site which obviously wouldn't be ideal.

Thanks for the replies folks, I'll happily ignore this error & stick with a dedicated Veeam network for the big (non SAN-attached) operations ...

Paul

[Vitaliy S.]

I guess it's there as a bit of protection from accidentally (or I guess purposely) using a proxy from one site to talk directly to a host on another site which obviously wouldn't be ideal.

Yes, this message warns you that proxy server in the remote site might be used to retrieve source data. Other than that, you should be fine with this message.

[pkelly_sts]

Perfect, thanks.