Hi everyone,

I have two copies of DPM 2012 R2 and FireStreamer, and two sites - the main site in Detroit and a satellite in Houston. Almost all of the data is in Detroit, consisting of mainly documents, then Exchange, Lync, a little bit of RDS, and a little bit of SharePoint. Operating hours are 0730-2230 week days.

Currently I am backing up Detroit data to the Detroit DPM server, weekly spinning it off to one of (6) external 4 Gb backup drives, aka FireStreamer 'tape', and taking it to the bank. I'm backing up a 3 Tb SAN, and have 30 days of on-site storage.

I have recently upgraded our MPLS link from Detroit to Houston to a 100 Mb fiber circuit, and am re-evaluating my options. I would really like to leverage my fiber connection and second site for disaster recovery, with the idea of making it simple. Here are my options as I see them:

• Firing up a Houston DPM server, and backing up Detroit to Houston, and Houston to Detroit. That way I have backup and disaster recovery in one fell swoop; not sure about Detroit to Houston time- and bandwidth-wise.
• Backing up Detroit to Detroit, and backing up the Detroit DPM with a secondary Houston DPM server. (I tried this when I had a 6 Mb MPLS and never got it to sync.)
• Same as above, only de-dup from Detroit to Houston.
• Backup to Azure storage or Azure VM. (I have 100 Mb of internet in both locations.)

I know I could take countless hours to research this, but I would appreciate anyone out there who can shed some quick  'big picture' light on what my practical options are at this point, and point me in a direction. I don't mind being embarrassed.  :^) 

Rob

Hi Rob,

Looking into the information you have provided us with above I would suggest you to look into the needs of what your organization demands/needs:

Continuously backup if the primary DPM-server goes down?

Do you need a highly available and secure off-site storage?

Is there a need to be up and running almost instantly in case of a disaster in your primary data center?

I know that I'm only giving you more questions but I want you to first look into what you want/need, without focusing on any products, then we can have a look at the "big picture" :-)

Markus,

Thank you so much for your reply - it is much appreciated. The main answer (isn't it always?) is to do as much as you can with as little money as possible.

After thinking this through, and talking it through with a member of the IT staff, we have come up with a tentative and very workable solution.

We will use the Detroit DPM server to back up everything in Detroit, and Houston across the 100 Mb MPLS circuit, and then put a secondary DPM server in Houston.

We have a spare 1U server in the Detroit IT Room. We will put (2) 4 Tb drives in it, set it up as the secondary DPM server, let the two servers synchronize while both are still in Detroit, and then after the servers are synched, send the 1U server to Houston, where it will be the secondary DPM server at our other geographically separate location and will synch over the 100 Mb MPLS circuit.

(I actually tried to do this before the pipe was bumped up from 6 to 100 Mb, and never got the secondary to sync across the 6 Mb circuit.)

The more I think about it the more I like it. Simple, geographically separated, and no more driving external USB drives to and from the bank! No more manually swapping drives! No more manually mounting, and FireStreamer, and managing, and repairing, etc., the external USB drives! It's all automatic, and should run without intervention.

And as far as cost: we own the 100 Mb pipe, the server, the OS license, the DPM license, we already have a Samsung 840 SSD for the OS, all we need is two 4 Tb drives for Houston storage; done for less than $400.

If anyone has anything to add please do!

Rob

• Edited by rob.pettrey Thursday, February 12, 2015 1:01 PM

Hi Markus,

Another option which might be worth considering. I don't know if you have gone down the virtualisation route at all but Hyper-V has a facility called Hyper-V Replica from Server 2012 onwards. This allows near real time replication over a WAN of VMs for DR purposes. 

You could virtualise and replicate your DPM server. You could also replicate your production VMs giving you much faster recovery in the event of a problem at your primary site - no need to restore backups, your replicas are ready to go.

Tim

Tim,

A great idea, and frankly that was our first idea. We have a 2 node cluster in Detroit supporting 18 Hyper-V VM's, and a Hyper-V host in Houston supporting 3 VM's.

Unfortunately both cluster nodes in Detroit are still running 2008 R2 (every other server is running 2012 R2.) The cluster was hired out, the original architect has moved on and is not interested, I haven't yet found anyone local who has specific cluster and SAN experience, and it has to be a clean install. I don't want to jeopardize our production environment, as all our eggs are in one very critical cluster basket.

The attraction of the secondary DPM idea is that is doable right now, it is within our skill set, and it is cheap.

If I can get the cluster upgraded, I think that is the best idea of all! Thank you!

Rob