Veeam Design Questions

[dvdrbc]

Hey guys,

Just want to query with you all about some best practices for our environment.

This environment was setup and configured by our Managed Provider as part of our server hardware refresh.

Currently our Head Office sits on 10.0.5.0/24 network which is connected to our DR Site (which is still local/onsite just much further away, connected via fiber). We have backups going to our single B&R server and we replicate key VM's to our DR site with plans to also setup a Backup Copy Job to go to our DR site in the near future.

I am relatively happy with our backups, utilizing Direct SAN. Our replication uses Network Mode, which can be a little slow.

My question really is, in reference to the diagram attached, how can we better our backup infrastructure? Is it missing anything? Can we do things better?

I realize this is a very vague post but any input is appreciated.

[DaveWatkins]

Define better, do you want faster.. cheaper.. more redundant?

The thing that jumps out at me is, you have your backup repository in the same place you have your SAN. How bad would it be if that room burnt down?

Depending on how fast that fibre link is, you could still run an iSCSI connection to your SAN with your repository in the DR site, but you at least have some separation of your backups and primary storage. Of course this assumes you don't put anything to tape or have some other offsite solution

Adding some VM based proxies will probably help replication speed, at least on the DR hosts

[dvdrbc]

More redundant, and faster if possible.

That jumps out at me also, if that room burns down the 1 and only B&R server is gone....would I need to put another B&R server in the DR site to avoid this??

We send backups to tape, forgot to mention that.

I added 1 proxy to each DR host this morning, so now replication is using HotAdd instead of Network...but I've found it takes twice as long to replicate now, jobs that were taking 8 mins now take 16 mins

[DaveWatkins]

That is... odd, what do the bottleneck stats say?

If the fibre link is fast enough, I'd just move the whole server to the DR site honestly. You could add a second repo there and do backup copy jobs, but if it's a GB link I'm not sure what that buys you, restores will still take as long unless your iSCSI link is multiple 1Gb links, and if it really comes down to it, you could bring the whole box back if you had to do a lot of large restores but keeping a repository in the DR sites makes way more sense to me than keeping one in the core site if you only have one

[dvdrbc]

I run the replication hourly:
First Run (Network Mode) - 28/04/2016 9:05:59 AM :: Busy: Source 86% > Proxy 4% > Network 27% > Target 43%
Second Run (HotAdd Mode) - 28/04/2016 10:14:28 AM :: Busy: Source 99% > Proxy 4% > Network 0% > Target 31%

It's a 1GB fibre link. Here is a better diagram

[foggy]

Hotadd has some overhead required to hotadd disks to the proxy server, while network mode doesn't, which could make it faster in some environments.

[skrause]

I have found that Hot-add usually adds about 4-7 minutes to other methods for the process of reconfiguring the proxy VMs to mount/unmount the snapshots so your numbers look around what I would expect.

How often are you replicating?

I probably look at just putting your Veeam repository and tape server in the DR location to solve the "fire in the primary room" issue if they are on the same 'campus' and connected via a link the same speed as your network inside the datacenter. No reason to purchase additional hardware to store backup copies. You can use that money to invest in 10Gig switching to link the sites