VMware has a way to deal with this, just like Hyper-V does.  Unfortunately, I 
don't know what it's called for VMware. Hyper-V has the original name of 
Hyper-V Replica.

It consists of having uniqueness in a replicated VM. There is a VMGuid and a 
VMGeneration. VMWare has a way to do the same thing.

This allows you to replicate, failover, resynchronize, failback, etc. This is 
different than Hyper-V Clustering.

The CHALLENGE associated with using a "bit-for-bit" copy is associated with USN 
Rollback scenarios. USN is discussed in detail here:

https://technet.microsoft.com/en-us/library/virtual_active_directory_domain_controller_virtualization_hyperv(v=ws.10).aspx#usn_and_usn_rollback


-----Original Message-----
From: [email protected] [mailto:[email protected]] On 
Behalf Of Charles F Sullivan
Sent: Friday, February 5, 2016 2:47 PM
To: [email protected]
Subject: RE: [NTSysADM] Replicating AD VMs

I think your key words are "....and then the *same* VM is powered on at the DR 
site". As far as I can tell, it would be almost the same using VSphere 
Replication. Even in your case the VMs that are brought up at the other site 
are behind in AD changes with the one(s) still standing. I'll be testing this 
beforehand anyway. Unless the replicated VMs are somehow "marked" as different 
machines, I don't see where this could be an issue, but I’m still listening.

-----Original Message-----
From: [email protected] [mailto:[email protected]]
On Behalf Of Michael Leone
Sent: Friday, February 5, 2016 2:38 PM
To: [email protected]
Subject: Re: [NTSysADM] Replicating AD VMs

On Fri, Feb 5, 2016 at 2:18 PM, Brian Desmond <[email protected]>
wrote:
> Essentially you’re circumventing AD’s replication engine with 
> something that isn’t going to enforce consistency which has the 
> potential to turn out very poorly.

I'm not sure if this is truly the case here, tho. I will admit to being 
unfamiliar with VMware replication (as opposed to the way we do it, which is 
SAN replication via RecoverPoint). But in our situation, what is happening is 
that the LUN that holds the VM is being replicated (synchronously, in my
case) to another LUN at a different site. Basically the LUN at the DR site is a 
non-active copy. So if the LUN at the main production site becomes unavailable, 
then the LUN at the DR site is marked as active copy, and the VM on that LUN is 
started.

That doesn't interfere with AD replication, which happens via IP to other 
VMs/hosts. Effectively, the VM at the production site is shutdown, and then the 
*same* VM is powered on at the DR site.
Effectively, it's just like replication failed temporarily.

Same VM, same name, same IP, etc. I know that in my case, that is what happened 
- we had no replication breaks, corruption, etc. Granted, I do have other 
physical Dcs that never went offline, during this test.
But I never had to seize roles, etc. The roles that were on Prod-DC-1 are still 
on Prod-DC-1 - it's just that Prod-DC-1 was inaccessible for a few minutes. It 
wasn't like it was out of commission for so long, it tombstoned.
You'd get the same effect if you pulled the network cable out of Prod-DC-1 for 
15 minutes, wouldn't you?


Reply via email to