Cold Recovery Site - Use Replication or Backup Copy Jobs

[aodhan]

Hi all,

I have a customer who wants to be able to achieve the following:

1) Backup all virtual machines (VMware only) and retain for 60 days - I have configured a 100TB Windows 2012 r2 Physical server with V8 B&R installed as a backup repository for this. This repository is not located in the customers DC. It is located away from the DC, but on the same network in a 10G connected location.

2) Replicate a subset (about 50%) of virtual machines to a DR site over a very small 50mb WAN link so they can be powered on in the event that they completely lose their production environment - The replication jobs will be seeded to 2 ESXi hosts which will temporarily be located on the production network. Once the initial replication is completed I will move the physical servers to the DR site and restart the replication.
- I have advised the customer that the WAN link is likely to be to small for the require rate of change - However I have time to test the replication before this goes into production, rather than relying on Veeam One reports or other estimates.
- Should I setup a WAN accelerated proxy first in production, place the DR ESXi servers in their final VLAN at Production then seed the WAN accelerators as well as the replica's? Or should I just do the initial replication seed then setup the WAN accelerators once the physical servers are in their final location?

2.1) They also want to retain that data from the above DR virtual machines for 5 years. - I was going to setup a backup job that runs against the replicated virtual machines at DR on a weekly basis and save this data to a de-duplication device like a Data domain. Fast recovery of this long term data is not important, however I believe with version 9 improvements I may be able to run a single Sure backup job on the data domain without it taking hours and hours.

Is this the best approach, or would it be recommended to run a backup copy job from Production to DR (to a second 100TB Windows 2012 R2 physical server) then run a restore of the backup copy jobs once they have finished? I would then need to run a second backup job on the replica's (with the Data Domain as the target repository) in order to keep them for the 5 years my customer requires.

Thanks for any feedback

[foggy]

- Should I setup a WAN accelerated proxy first in production, place the DR ESXi servers in their final VLAN at Production then seed the WAN accelerators as well as the replica's? Or should I just do the initial replication seed then setup the WAN accelerators once the physical servers are in their final location?

I'd do the initial replication seed, then configure WAN accelerators and pre-populate the target WAN accelerator cache, finally move the server in the target location.

2.1) They also want to retain that data from the above DR virtual machines for 5 years. - I was going to setup a backup job that runs against the replicated virtual machines at DR on a weekly basis and save this data to a de-duplication device like a Data domain. Fast recovery of this long term data is not important, however I believe with version 9 improvements I may be able to run a single Sure backup job on the data domain without it taking hours and hours.

Backup of replica VMs cannot use CBT, so I'd suggest using backup copy jobs for those selected VMs with required GFS retention instead.