Planned Failover using Failover Plan [Feature Request]

[haslund]

Hi all,

Loving the new failover plans in v8!
I understand the Failover Plan currently is to be used in an emergency situation.

However what if I need to perform a planned failover?
Why would I want to have to select VMs again, specificy failover order again etc. if I already specified all this in the Failover Plan?
A use case I even see Veeam specific in the documentation is recurring failover testing (weekly fx?) - http://helpcenter.veeam.com/backup/80/v ... eason.html

Could we please have an option of performing a Planned Failover using a Failover Plan?
If yes, could we also have the option of scheduling this so it could run every X weeks or every X months?

Thank you for your kind consideration.

[SBeel]

That sounds like a great idea, I would like this feature too please.

[veremin]

If yes, could we also have the option of scheduling this so it could run every X weeks or every X months?

Can you elaborate on the use case of having automatic (regular) schedule for planned failover plan? Will it be used mostly for testing purpose or something? Thanks.

[Gostev]

Rasmus - since cold failovers takes time (at least a few minutes for each VM, more depending on the amount of incremental changes since last replicatioN), we will not be able to meet the delay timeouts as specified in Failover Plan. Is that OK you think? I am thinking this may cause distributed applications to time out.

[haslund]

Can you elaborate on the use case of having automatic (regular) schedule for planned failover plan? Will it be used mostly for testing purpose or something? Thanks.

Having a secondary location to failover to is great, but there can be many pitfalls.
One such example could be when using re-ip. Can other machines still reach these failed over VMs.

Having the failover tested regularly you have the confidence that it will work when the day comes you really need it.
Same situation as why basically we are performing SureReplica testing but in this instance we also want to test connectivity to the machines from other machines.

[haslund]

Rasmus - since cold failovers takes time (at least a few minutes for each VM, more depending on the amount of incremental changes since last replicatioN), we will not be able to meet the delay timeouts as specified in Failover Plan. Is that OK you think? I am thinking this may cause distributed applications to time out.

Maybe I am misunderstanding something, but I thought failovers using a Failover Plan is always cold (production VM is already powered off before executing failover plan)?

[veremin]

I believe Anton was talking about planned/cold failover. During such failover source VMs are shutdown, then, backup server transfers changes that have occurred since the last replication cycle, and finally replicas get powered up. The step of transferring incremental changes might take some time, and, thus, isn't currently used by Failover Plans.

Thanks.

[haslund]

Well, as one customer told me:

"It does not make any sense to have to plan for doomsday and then not being able to test your plan.
Obviously I can select all the VMs manually, but why must I manually do something that I already told the machine how I want done?"

I guess it would make sense to have an option added to schedule a replication to run in advance of the planned failover?

A different approach is to simply add a scheduling option to the Failover Plan so it is not a "Planned Failover" where last changes are replicated.

[Gostev]

A different approach is to simply add a scheduling option to the Failover Plan so it is not a "Planned Failover" where last changes are replicated.

This is exactly how Failover Plans work already, if you just go ahead and initiate failover using a Failover Plan. Last changes will not be replicated...

[haslund]

This is exactly how Failover Plans work already, if you just go ahead and initiate failover using a Failover Plan. Last changes will not be replicated...

But there is no scheduling option?

[veremin]

If that's what your customer really wants, then, let him schedule the following script via Windows Scheduler:

Code: Select all

Start-VBRFailoverPlan -FailoverPlan (Get-VBRFailoverPlan -Name "Name of your Failover Plan")

Thanks.

[dellock6]

After playing with failovers in these days, I'm with Rasmus that scheduling options would be nice.
And not just for failover plans, but also for the "planned failover", right now it can just be executed manually or scheduled via powershell, a scheduler option in the wizard would be nice. Say I want to run a planned failover at 4am, I simply prepare it and schedule it, but I don't have to stay awake to run it.

[lobo519]

Just to be clear - If I setup a Failover Plan and right click and start it - it will NOT touch my production VM's in any way correct?

Basically I want to test my plan, prior to Failover Plan I would just do a Restore > Failover to replica and I was able to test all my replicas without impacting my production VMs.

[foggy]

Correct, failover plan does not touch production VMs.

[jveerd1]

We are with Rasmus as well. Very often we have customers asking why a Planned Failover Plan is not possible, this is a general misunderstanding. There clearly is a need to logically group a set a replica's and perform a planned failover on them. Of course you could select a group of replica's manually (easy for 10 replica's, much harder with 100s), but that is not what customers want/expect.