Design Question

[pamiller3]

Hey all,
I am in the middle of planning a veeam rebuild and would love some input from other admins.

Setup:
I currently have 2 B&R servers that serve as proxies as well one in each of my DCs. I am working at an ISP so there is a 10G connection between DCs. I am a vmware only shop and the servers themselves are vms. The transport mode is network on both servers as well.

Change #1:
I am currently seeing a lot of jobs failing to finish over night. I am running some updates and things and overall I believe these two server are not able to handle the load, so I am looking into installing just the proxy service on 2 more servers to help share the load. So this poses 2 questions:
1) Would extra proxies really help with this issue?
2) Would I have to deploy new VMs or is it really do-able to install this proxy on other VMs that are not used during the night?

Change #2:
I am looking into using vmware tags to map replication jobs to VMs, I think this might be easier to just apply a tag to a VM and get the same job settings applied, but I am worried about a few things. I currently have around 300 VMs and nearly each get their own job. I found this useful when I do need to use a replica or a single VM is messed up then ONLY that job is messed up as well. So the biggest question I have is What do you do for jobs and VM mappings that I am not doing, I am just trying to look for suggestions on this one.

Thanks all!

[Mildur]

Hi Andy

Welcome to the forum.

Change #1:
Difficult to say without seeing logs from your environment. I suggest opening a case with our customer support. Please don't forget to provide us with the case number. My personal recommendation would run a test with a Linux proxy on your vSphere Environment and use it as a hot-add proxy. If that works better, change your jobs.

Change #2:
Using 300 Jobs will blow Memory requirement for your VBR server. Remember, each concurrent job requires 500MB of RAM on your backup server. Maybe that's the issue in your environment. The support case will provide more details on it.
I suggest using only a few jobs instead of jobs per VMs. 300 VMs in a single job is not a problem.

Best,
Fabian

[dloseke]

Change #1:
I am currently seeing a lot of jobs failing to finish over night. I am running some updates and things and overall I believe these two server are not able to handle the load, so I am looking into installing just the proxy service on 2 more servers to help share the load. So this poses 2 questions:
1) Would extra proxies really help with this issue?
2) Would I have to deploy new VMs or is it really do-able to install this proxy on other VMs that are not used during the night?

I'm with Mildur, I'd recommend setting up a linux proxy server on each cluster for hotadd access. I don't really recommend using another VM in the cluster because you don't want to have the proxy service on a VM that you're also backing up, and I'm assuming that you'll want to backup pretty much all of your servers. What I'm confused about is why your VBR servers, which you stated are virtual, is why they are using Network mode. Do these VM's not have access to the datastores that your VM's that you're backing up reside on, or do they not hold the VMware proxy role?

I don't have any input on Change #2 as I'm not well versed enough in the use of tagging for replication -- something I need to get more into there.