Configuration advice please

[Chaswood]

Primary Site:
VMWare 6.5
VEEAM server (PRVMWARE)
VEEAM Proxy (PRPROXY)
StoreVirtual SV4530G2 (PRSAN)
StoreOnce SO5100

Secondary Site:
VMWare 6.5 (DRVMWARE)
VEEAM Proxy (DRPROXY)
StoreVirtual SV4500 (DRSAN)

There is a 100MB MPLS connection between sites.
I'm needing to backup VMs at the secondary site.

What goes where? And how do I configure it all? Do I setup a WAN accelerator on the Proxies? I'd like to take advantage of the StoreVirtual's integration with VEEAM and do backups from snapshots.

Just making sure what I'm thinking is the right thing to do.

Charles

[foggy]

Hi Charles, since you do not mention any backup storage for DR site, just the primary storage for virtual infrastructure, I assume you're talking about replication. You can either run it directly to the target host (using WAN acceleration, if required) or perform backups locally in production site first and then use replication from backup. In both cases, you can use integration with StoreVirtual storage snapshots.

[Chaswood]

No replication. Both sites have unique VMs running on them.
So, I'm not sure how exactly to backup the VMs at the DR site.

If I do the first (perform locally and replicate from backups), should I need the WAN acceleration?
If I do the directly to target host, should I make the proxy servers both WAN acceleration servers?

[foggy]

Whether you need WAN acceleration or not depends on the actual job performance (bottleneck stats will tell you - if the primary bottleneck is network, then WAN acceleration will allow you to reduce traffic by trading it for the target WAN accelerator disk I/O).

[BartP]

Before you really know how to set things up, i'd like to know a couple things:
- How many VMs are we talking about and off course what is the size of the Production Data to be backed up? (to calculate the amount of data you need to transport)
- How many VMs are you planning to have in backup and replica jobs to the 2nd site? (again.. how much data to transport)
- When you know this, you can calculate the needed throughput and if you need a WAN Acc (@100Mbps expect to use 15k in raid1 or ssd for cache) or not.

With a 100Mbps WAN you can (effectively) get about 850GB-1TB of data per day transported and a WAN acc can greatly reduce that amount, but calculating/comparing hashes takes time.
Starting backup copy jobs for machines with a rather bad amount of dedupe first and have the other jobs going over several WAN acc (to hash while the other jobs take all the bandwidth and then use less bandwidth at a later time)
Again, it all boils down to the amount of data you need to transport before you can design your setup/placement.

[Chaswood]

There are 27 VMs in the DRSite. Largest is 670GB Used space. For now, I'm planning on the whole everything: Approx. 8TB.
The 2nd site has production machines as well. So, I'm not having any replica jobs at all. All VMs will be backed up to the primary site's SO5100.

I did a test of a shutdown VM. The VM was 25GB Used space.
The VEEAM did snapshot the LUN.
The VEEAM did not use the DR Proxy, it used the PR Proxy.
Bottleneck was at the Source (Source 99%:Proxy 3%:Network 0%:Target 0%)
Processing Rate was a sad 11MB/s
I only let it run for about 10 minutes before I stopped the job.

[BartP]

Bottleneck was at the Source (Source 99%:Proxy 3%:Network 0%:Target 0%)
Processing Rate was a sad 11MB/s
I only let it run for about 10 minutes before I stopped the job.

I'm sorry to say, but i cannot say anything about the speed since i don't yet know a thing about the placement and components used in this job. (too little info provided, only the result)
This however does sounds like you make a backup from a VM while it's being processed by a proxy which is on the other side of the WAN connection (hence the 11MB).
You have an MPLS after all, so one big network. Veeam does not know this and random picks a proxy unless you have set one.

furthermore:
Considering 50% data left after reduction (default to calculate with); 4TB
An incremental would be about 10% change rate and i'd expect 400GB of data to be transferred daily to your offsite repository.

Place a proxy and repository at your DR site, this should be enough. Make sure that any backup for the DR site VMs have the proxy -set- (or at least have proxy affinity) on this DR proxy so Veeam won't pick a random proxy in your primary site.
I see not reason to use the WAN Acc at this point in time, but think about adding a server to be used as Guest Interaction Proxy.

[Chaswood]

So I have configured a job to us only the DRPROXY.
The job will backup a VM on the DRSAN to the SO5100 at the PR site.
The DRPROXY is only a proxy server and nothing else.
The DRPROXY (Windows 2012R2) has the iSCSI initiator running and the DRSAN IP is in the Discovery>Target Portals.
The firewall on DRPROXY is turned off.

I get and error: Cannot find a suitable proxy for retrieving data from storage snapshot...

I'm not sure what I need to check as the DRPROXY looks to be configured exactly as the PRPROXY.

[foggy]

Looks like this proxy is not configured to backup from storage snapshots.