DR Failover capacity/performance modelling?

[yamy125]

Hi,

We are trying to establish a way to model a theoretical failover of replicated VMs at a DR site to see what impact they would have on DR infrastructure in the event of a disaster/failover. I thought the option was available with the 'what-if' deployment projects section of Veeam One Workspace, where I can select the DR host cluster and then add in VMs for the modelling scenario. The issue there is that I can only add in VMs that sit on the DR infrastructure (i.e. the *_replica VMs and a few live VMs that sit there on the DR cluster). This gives us a rough idea, but doesn't take into account the CPU statistics etc. of the actual VMs running because it is referencing the offline *_replica VMs and statistics.

What we really need to be able to do is select the DR cluster infrastructure, but then add in VMs from our production cluster so that statistics of the actual live running VMs are applied against the DR infrastructure. The modelling also doesn't appear to take into account that we already replicate all of these VMs to the DR cluster - and so it doubles up on the storage requirements for the modelling scenario. This isn't a big issue, but figured worth mentioning.

Is there any way to actually model the load on DR infrastructure for a failover scenario? This information is pretty much invaluable to ensure that, in the event of a disaster, the DR infrastructure is actually capable of running particular *_replica VMs required. I would have thought a fairly common question/scenario and hoping there is a relatively easy way to model and report on this that I've missed?

Cheers,

Tom

[Vitaliy S.]

Hi Tom,

The issue there is that I can only add in VMs that sit on the DR infrastructure (i.e. the *_replica VMs and a few live VMs that sit there on the DR cluster). This gives us a rough idea, but doesn't take into account the CPU statistics etc. of the actual VMs running because it is referencing the offline *_replica VMs and statistics.

It should be possible to add VMs from the entire VI scope, please contact our technical team to remove this limitation. You can let your support engineer that this case should be escalated to the dev team directly or give a reference to this topic.

What we really need to be able to do is select the DR cluster infrastructure, but then add in VMs from our production cluster so that statistics of the actual live running VMs are applied against the DR infrastructure. The modelling also doesn't appear to take into account that we already replicate all of these VMs to the DR cluster - and so it doubles up on the storage requirements for the modelling scenario. This isn't a big issue, but figured worth mentioning.

Yes, deployment projects are "not aware" that you replicate these VMs already. We will think how it could be addressed in our next updates, thanks for the heads up.

Is there any way to actually model the load on DR infrastructure for a failover scenario? This information is pretty much invaluable to ensure that, in the event of a disaster, the DR infrastructure is actually capable of running particular *_replica VMs required. I would have thought a fairly common question/scenario and hoping there is a relatively easy way to model and report on this that I've missed?

Yes, there is a way to do that now. You can try this trick, should help:

1. Select your source site as a scope for deployment projects report
2. Remove all source hosts, and VMs
3. Add new hosts (emulate them) from the DR site
4. Add VMs you want to do modelling for

Looks pretty complex, but it should give you the result you want.

Thanks!

[yamy125]

Thanks for the reply and suggestions Vitaliy.

Slight issue there is that there are some live-running VMs at the DR site that also need to be taken into account in terms of stats/load, so to semi-accurately model we really need to be able to add VMs (or hosts I guess also) from both clusters. Understand they could be emulated, but obviously not so accurate then. I'll have a play around with it anyway and have gone back to my support case (originally logged and they suggested I post here on the forums - case 00963344 for ref).

Cheers,

Tom

[Vitaliy S.]

Tom,

If you need to add VMs from any location, then our support team should be able to assist you with removing our default filtering that limits the set of available VMs.

Let me know if you need any further assistance.