Wish List - Primary Server Control

[RMullis]

This is our existing setup;

Site 1 servers being replicated to site 2

Site 2 servers being backed up and replicated to site 1

Site 3 servers being backed up to site 1 and replicated to site 2

The replication is required for quick recovery of datacenter failure.

Each datacenter backs up to local disk retaining 1-4 weeks, depending on type of data, for quick restore of single systems in the event of single server failure or simple file restore. Site 2 & 3 are backed up to site 1 to be archived to tape. The unfortunate part is the jobs keep bumping into each other causing issues. Since I need a backup server at each datacenter for quick recovery of either backed up or replicated data, we are constantly having to perform retries. If Veeam B&R had the ability to link the backup servers and the data was somehow replicated we could control all servers from either a primary server or a client app similar to the VMware client giving us the ability to control our backups & replications completely. I realize the data is store in an SQL database but if that could be replicated across the servers along with the metadata files then if the time came where we had to recover a datacenter by bringing our replicas online, all of the configurations would be available. We have a tri-pod datacenter setup so if any location goes down we will maintain business continuity, but our backups are almost a full time job. If there is already a solution for this kind of setup that i have missed then I do apologize, or if there is a better solution, I am definitely open to suggestions. My plan is to move to version 7 by week end.

Thank you for your time.

[dellock6]

Hi Rick,
your jobs are bumping into each other because the only way right now to cohordinate all the activities is to have a single central Veeam server, in order to manage proxies and repositories queues. In your situation, I will use a single veeam server at site1 (probably is already there?), and only leave proxies and repositories at site2 and site3.
First, replicas does not need any Veeam server to be activated.
Second, about backups, am I missing backups at site1? Those servers are only replicated and not backed up?

Anyway, I see at least two scenarios.
1. Site1 is down: if among the replicated servers of site1 there is also Veeam server, you can power it up at site2 and manage all restore operations from there.
2. All sites isolated: site2 can use again the replica version of Veeam server. Since network link is down, there is no problem of IP or machine conflict. Once the link is restore, the next replica job will simply overwrite the modifications you created on replicated Veeam server when you power it ip. Same for site3.

But even better, with at least Veeam VBR 6.5, you can simply replicate among different sites only the configuration backup. You keep an unconfigured Veeam server at each site, and if something happens to any site, you restore configuration into one of those Veeam servers and start restoring. Also for quick activities there is the command line extract.exe utility that can work without VBR console.

Luca.

[veremin]

Also, you might want to utilize Enterprise Manager. Even though it won’t make VB&R installations aware about each other, it will give you, at least, a necessary visibility across three servers, concentrating all control sticks in one point.

Thanks.

[RMullis]

Thank you v.Eremin but I have been using it for almost 2 years now and it has too many limitations.

Thank you Luca,
To fill in a couple of missing pieces, we do backup site 1 to local disk and archive to tape. And the backup servers at site 1 and site 2 are both physical machines. I really like what you are saying so if you would allow me to ask you a few questions with your train of thought I would be very appreciative.
If I move my backup server to virtual I would loose my direct San backup, correct?
Or should I keep it physical?
And if a data center we're to go down, during the reimplementation how would I get the backup server back to begin restoring my systems?
If I have access to the backup files all I really need is the backup server config file right?
Or do I need the SQL database too?
If I were only able to replicate my physical backup server or back it up with Veeam, it would greatly ease my DR plan.

Anyway, thank you for your thoughts. I think you are really on to a great setup.

[veremin]

If I move my backup server to virtual I would loose my direct San backup, correct?

Actually, you can move VB&R to VM machine, meanwhile, keep physical server as a Direct SAN proxy server.

And if a data center we're to go down, during the reimplementation how would I get the backup server back to begin restoring my systems?

You will have to install VB&R software and, then, import previously backed up configuration.

Or do I need the SQL database too?

Having configuration backup would be enough.

Thanks.

[dellock6]

Vladimir already gave you the answers

Basically, you only need to protect the Veeam server (and you can use the configration backups) and the repositories (obviously, they hold backups). Proxies are as I like to call them "cannon fodder", you can destroy and recreate them, install new ones, disable, as you need. They are totally controlled by the Veeam server and have no data in them.

Split the Veeam server from proxies, and protect the Veeam server.

Luca.

[RMullis]

I would like to thank you both tremendously for your ideas, suggestions and quality answers. I really like this design as I feel it would work perfect in our environment. Thank you, thank you, thank you!!!!

[RMullis]

There may be one problem with a virtualized backup server, our robotic tape library is connected to the site 1 physical host via fibre channel.

[veremin]

In this case, you can use this free 3rd party software in order to expose physically connected tape appliance via ISCSI protocol. Thanks.

[RMullis]

just an fyi from what i have read, when you use this kind of software you cant have the server vMotion to another host.

[veremin]

It seems that you’ve confused pass-thorough and tape redirector scenarios. The former certainly ties you up to a given host, depriving you of specific VMware features, such as vMotion, Storage vMotion , HA , DRS, etc. However, I believe it isn’t the case with the latter approach, since in this implementation the tape drive/library is connected via ISCSI initiator that isn’t locked to a specific host, nor is it affected by vMotion stuff.

Thanks.