Two sites - Two VEEAM servers - which should run a job?

[mysticalice]

Hi

Two sites - each running production VMs. Vcenter on each end and a VEEAM server on each side as well.
Currently each side, nightly backs up all of the VMs at its site to the other site (opposite server is a backup repository for each veeam machine).
This works, as I know even if the one side goes down, worst case we could always extract the backups from the other side manually

Looking into doing veeam replication to the other site, but have a question:

Should we set up the replication job on the veeam server at the other site? For instance, usually if I wanted to replicate siteA servers to siteB, I would set up the job on VEEAM at siteA, and point the jobs at vcenter at siteB. In thinking about it though, if I ever lose siteA, chances are I would lose veeam along with it. Now I know in the case of a replica I COULD just manually power on the replicas at siteB, but in the event we want to fail back, or recover an older replica, I wouldn't be able to until I got siteA VEEAM back online (to my knowledge).

So i'm considering setting the job up at the other end VEEAM server. Ex: To replicate siteA servers to siteB, I would set the replication job up on siteB VEEAM. That way if something ever happened to siteA, I would still have all the replication information on VEEAM at siteB, and could easily recover.

My concern is whether that is a) best practice, and b) what happens in terms of data dedup, compression, etc.

Gigabit connection between the two sites.

[veremin]

a) best practice

Yep, it's a recommend practice, indeed.

b) what happens in terms of data dedup, compression, etc.

Nothing, as long as you have proxy servers setup at both ends (site A, site B). So, if you want for replication traffic to cross 1gbit line in deduplicated, compressed stated, be aware to specify both source and target proxy servers. Say, you create a replication job on site B willing to replicate residing on site A. In this case, using backup server B, you should assign proxy role to some machine at site A (will be used as source proxy) and do the same for a machine at site B (target proxy).

Thanks.

[Cragdoo]

Do the 2 sites use different IP subnets? Are the VMs Windows OS? Do you want to re-ip the VMs?

If you have windows VMs and want to re-ip then I would setup the replication job at site B, and 'pull' the VMs from site A. In the event of a site failure at site A , you can use site B VBR console to re-ip your windows servers.

[veremin]

Yep, as mentioned above, the so-called "pull" strategy is recommended for replication scenarios. If something happens with the production site, a user will be able to execute DR procedures, such as failover, failback etc. from the DR site without any issues. Thanks.

[michaelryancook]

I think my question is answered here: http://forums.veeam.com/post163373.html ... er#p163373 but I wanted to ensure my slightly different scenario would not impact the solution. I have two Veeam B&R servers. One in our prod site and one in our DR site. I have backup jobs, backup copy jobs, surebackup jobs and replication jobs all running in the prod site right now. Backup jobs target is in prod. Backup copy jobs and replication targets are in DR. I am working towards creating a failover plan and in reading believe I should have the replication jobs in DR not prod. My concern is that we use the same IP address space in prod and DR so that we do not need to do re-ip. We use NAT to provide connectivity between the two sites. We only have a single vCenter that manages hosts/clusters in both sites. That vCenter is located in the prod site and is replicated.

Can anyone see an issue with the DR Veeam server running the replication jobs? Do my steps below make sense if we need to bring up the DR site if the prod site is down? Is there an easier way of doing this?

1. Our external connectivity will flip to the DR site based on route metrics.
2. Connect directly to ESXi host to power on domain controller VMs.
3. Update DNS so that hosts in DR site are accessed with real IP rather than NAT IP.
4. Connect directly to ESXi host to power on vCenter VM and SQL VM.
5. After vCenter is running use DR Veeam to execute failover plan.
6. Based on type of disaster, decide on steps beyond concerning permanent failover, failback, etc.

[foggy]

Michael, provided you switch to pull replication scenario, your plan looks pretty solid.

[michaelryancook]

Thanks Alexander.

[leegregory]

We have been using Veeam for a couple of years now and we are going to be replacing SRM with Veeam Replication.

I have set-up a proxy at the remote end that links back to our main backup server and have set-up replication jobs.

All the replication jobs are working great and the planned fail over also works fine.

My question is, if the main site is down and this includes the backup server as well then how do you initiate a fail-over?

I appreciate that you could just power on the replicas at the remote site and that would work but if you need to fail back form these then you need Veeam to be aware that the fail-over has occurred in the first place.

Also, if you are in DR then the remote site will also need to backup the now live replicas. I assume this just needs a 2nd backup server at the remote site to facilitate this, in our case thsi would be a virtual machine running the same version of Veeam as the production site?

Regards

Lee

[veremin]

My question is, if the main site is down and this includes the backup server as well then how do you initiate a fail-over?

You'd better install a backup server in DR site and orchestrate both backup and replication jobs from it. Thanks.

[leegregory]

Thanks Vladimir i thought that would be the case.

Regards

Lee

[veremin]

You're welcome. Should other help be needed, don't hesitate to reach us. Thanks.

[rosenstand]

Hi all

I'm hoping some of you can give me some guidance onto this problem.

I need to find the best solution for having an DR site and at the same time locally and long term backup.

Here's my situation today.

At my production site (Site A from now on (warm site)), I got 7 hosts, running 140VM on an Nimble storage unit.
At my DR site (Site B from now on (cold site)), I got 5 hosts.

At site A I also have the following at the moment.
1 Virtual Backup Server (Master server)
1 Physical Backup Proxy Server

Between the two sites, I have an redundant 1Gbit connection.

I'm planning to buy an Nimble unit to place on Site B, and also two new physical servers to store my backup on (One for Site A, and one for site B). Both with 12x 10TB (Raid6) disc.

The various solutions I've thought of so far.

1. - Nimble to Nimble
Setup volume replication on site A to site B, to being able to failover to site B.
Set the among of snapshots on site A low (for local backup) and long on site B (for long term backup).

This seems smart, because if out of the box directly from Nimble, but it is not that flexible to find an single file to restore, either locally or remotely.
It will also give me some administrative work, to make the failover, because I need to manually clone the volumes on site B, and present these to VMware, to be able to start all of my VM again.

2. - VMware SRM with Nimble to VMware SRM with Nimble
Install VMware SRM and integrate this with Nimble, to perform the replication from site A to site B.
This also gives me the chance to test if everything works as planned, in an failover case.
Less administrative work.

But it leaves me without any backup at either sites.

3. - Veeam B&R to Veeam B&R
Setup two physical servers, on at each site, and install these as two locally backup servers.
At site A, it will make backup locally and replicate some files for long term to the server at Site B.
At site A, setup replication of VM to Veeam server at Site B, and then have this to upload it onto the Nimble unit again?! (Is this the right way of doing this?)

This will give me locally backup and long term backup at site B, and makes it a lot easier to find files / VM back in time, to restore from.
But will the replication work as easy as it does directly on the Nimble unit?!

4. - A hole new way, that I haven't t come by here
If you know about another way of doing what I want, and also smarter, then please let me know.

I'm hoping that someone can help out with this problem, and give me some guidance in what approach would be the best.
If I where to go with two physical Veeam servers, would I then make them both as proxy's for my VM Backup server, or make clean installation on both of them, to have them to work as two separately backup servers?!

Because if I go with an Veeam solution, I need to make sure, that everything backup at site A, whatever it is short term or long term, can be found and restored from the Veeam server at site B (not restored short term backup, but that I knows about that files are in the long term repository, and also about all of the replications jobs).

Thanks

Rosenstand

[PTide]

Hi,

You can also

4. Create replica from backup on site B, though this will slighlty increase your RPO.

5. You can utilize existing Veeam backup and replica from storage snapshot capabilities.

Thanks