failing over to replica without vcenter

[bdufour]

good veeam folks,

is there a way to failover to a replica without vcenter - say through the host the replicas are stored on? although the jobs are configured through vcenter.

veeam is at our DR site - vcenter is on the production side. i moved veeam when i started, as both, veeam and vcenter, were on production when i started with the company. i have replicated vcenter to the dr site, but after speaking with vmware on the matter - they arent positive the restored replica will actually work because of the ssl cert/ip tied to the vcenter server appliance. which is an entirely different project that i have some ideas for.

- ok so, back to square one, can i failover to a replica via the veeam console without vcenter? say through the host the replicas are stored on (BTW vcenter was used for the jobs).
- if i cant failover via the veeam console, would the next step be to manually power on the replicas and consolidate the disk?
- at which point, i would create new jobs in veeam to replicate the now powered on and consolidated replicas back to production using the production vms are targets? im pretty sure veeam will have to recalculate the disk which is fine, as long as i dont have to replicate the entire vms and can use cbt.

thanks!

[DGrinev]

Hi Blake,

1. You're right, you cannot failover to a replica via Veeam console without vCenter used in the jobs.
2. You can manually fire up replicas and here is the KB with the testing approach.
3. In case you target production VM for Replica Mapping the digest calculation is expected to determine differences between the VMs.

Please review the existing thread for additional information. Thanks!

[bdufour]

thanks Dmitry,

im testing this as we speak. i understand i can manually fire up replicas for testing, as i use that often. this is basically, if we lose the production side (fire, flood, catastrophic failure, ect). i think it's pivotal in our business continuity plan to understand how to not only bring up replicas manually but also replicate data back to production using CBT after manual failover - once production is back up. as i cant rely on vcenter at the dr site working upon start up - as vmware said the appliance will not work on a diff network (dr site) bc of the ssl cert and the ip it was created with - this is unchangeable according to vmware unless you regenerate the cert. while everything is down, jumping through hoops to get vcenter up and regenerating certs ect before you can failover to replicas is not the path i want to go down. a better plan, would be to manually fire up the replicas without vcenter.

test plan of action in a catastrophic event at the primary site:
1. replicate test vm to dr site with a few restore points
2. power down vm on prod site
3. power up replica vm at dr site
4. consolidate replica vm disk
5. 'once prod comes back up' - replicate vm back to prod using prod vm as target (seed)
6. once replicated back to prod - fire up vm
7. replicate prod vm to dr site using replica at dr site as seed

i understand that disk recalculation will indeed need to be performed and thats fine, bc after recalculation cbt will be used on the first pass. which is what im looking for.

other idea*
in an attempt to get vcenter up with original ip for ssl cert functionality straight away - create production network vlan at dr site (prod is completely down) and advertise out via bgp. it would make dns easier, and no need to re ip.

ill post my results here, as id imagine most places dont have veeam/vcenter at their dr site. if i would have designed this network from scratch i would have placed both veeam and vcenter at the dr site. this is the best design in my opinion.

[foggy]

If you point the job to an ESXi host directly, you will be able to failover via Veeam B&R without vCenter. You can either do this for all VMs or have a separate job replicating just vCenter Server VM pointed to a host, quickly failover vCenter itself manually in case of DR, and then all other VMs via Veeam B&R using connection to a vCenter replica.

[bdufour]

thanks foggy