Discussions specific to the Microsoft Hyper-V hypervisor
Post Reply
pesos
Enthusiast
Posts: 82
Liked: 2 times
Joined: Nov 12, 2014 9:40 am
Full Name: John Johnson
Contact:

improving DR configuration

Post by pesos » Nov 05, 2017 7:47 pm

Hi all,

We have two hyper-v clusters separated by 50ms latency on a 225 mbit link (we can burst above this but one farm is on 95th percentile billing).

Production has 3 nodes, DR has 2 nodes. Both have a shared SAS production array and a second shared SAS array for veeam backups.

We have two veeam 9.5 servers each licensed for a single host. Currently both veeam servers are in the DR site.

We back up one Exchange DAG node in each site, and file servers and skype servers and web servers in the production site.
We replicate file servers to the DR site as well as a few other servers.
In the DR site we back up some utility servers as well as the replicas.

The last part is what I'm unsure of. Since we are backing up replicas, CBT cannot be used. I wish there was a way to have zero snapshots set on replicas (we have the lowest offered, which is one) so that avhdx files would not be used at all.

Perhaps there is a better way to approach this? I like having completely independent backups at each site vs. simply doing a copy job, but maybe that's how we should be reworking things, and not back up the replicas at all (rather, build the replicas from the local copy job)?

P.Tide
Product Manager
Posts: 5074
Liked: 439 times
Joined: May 19, 2015 1:46 pm
Contact:

Re: improving DR configuration

Post by P.Tide » Nov 05, 2017 8:28 pm

Since we are backing up replicas, CBT cannot be used.
That holds true for VMware. With Hyper-V it's OK to backup replica using CBT. Does that help? :)
I like having completely independent backups at each site vs. simply doing a copy job
May I ask you what's wrong with doing a copy job? It was specifically designed to transfer your backups in a safe and reliable manner.
rather, build the replicas from the local copy job)?
That should work too.

Thanks

pesos
Enthusiast
Posts: 82
Liked: 2 times
Joined: Nov 12, 2014 9:40 am
Full Name: John Johnson
Contact:

Re: improving DR configuration

Post by pesos » Nov 05, 2017 8:34 pm

CBT does not function on the replicas. We get the "cannot use CBT" warning and the backup takes hours as it runs through the full disk.

If we do a copy job, then we are dependent on the production backups vs having two fully independent backup chains.

P.Tide
Product Manager
Posts: 5074
Liked: 439 times
Joined: May 19, 2015 1:46 pm
Contact:

Re: improving DR configuration

Post by P.Tide » Nov 05, 2017 9:18 pm

We get the "cannot use CBT" warning and the backup takes hours as it runs through the full disk.
I suspect that is not related to the fact that you are taking backup of a replica. Please check this KB article and follow all the steps mentioned there to troubleshoot CBT.
then we are dependent on the production backups
What do you mean by "dependent"? Backup copy chains can be used separately from the main chain.

Thank you

pesos
Enthusiast
Posts: 82
Liked: 2 times
Joined: Nov 12, 2014 9:40 am
Full Name: John Johnson
Contact:

Re: improving DR configuration

Post by pesos » Nov 05, 2017 9:24 pm

Pretty sure it is related to backing up a replica. Had a prior support case on this.

P.Tide
Product Manager
Posts: 5074
Liked: 439 times
Joined: May 19, 2015 1:46 pm
Contact:

Re: improving DR configuration

Post by P.Tide » Nov 05, 2017 9:41 pm

Had a prior support case on this.
Could you post the case ID for reference please?

Thanks

pesos
Enthusiast
Posts: 82
Liked: 2 times
Joined: Nov 12, 2014 9:40 am
Full Name: John Johnson
Contact:

Re: improving DR configuration

Post by pesos » Nov 05, 2017 10:22 pm

I think it’s 02271491 but not positive

pesos
Enthusiast
Posts: 82
Liked: 2 times
Joined: Nov 12, 2014 9:40 am
Full Name: John Johnson
Contact:

Re: improving DR configuration

Post by pesos » Nov 14, 2017 8:02 am

Hi, any thoughts? We are still getting

Cannot use CBT: There is no change tracking data available associated with the specified change tracking identifier.. Failed to query changes for disk 'C:\ClusterStorage\Volume1\FILES1-REP\7B0AC05F93E246FEB7A0F82F2B9F3036(1)\SCSI0-3\FILES1-M_35D4D74A-9A77-4BB6-B130-4C312E646061.AVHDX'

for all our disks and out of 1.6 TB, 1.3 TB gets read and the backup takes 4.5 hours. With CBT this would be a lot faster, no?

P.Tide
Product Manager
Posts: 5074
Liked: 439 times
Joined: May 19, 2015 1:46 pm
Contact:

Re: improving DR configuration

Post by P.Tide » Nov 14, 2017 11:22 am

Hi,

Sorry for the delay. So, indeed, you're right - CBT does not work if you're backing up replicas. Then the only way to get things going faster would be to do it the other way around - start doing remote replicas from local backup copy job files so CBT can be utilized. Before we go any further please correct me if I'm wrong:

1. VBR-1 (on the DR site) manages backups that hit PROD local storage and also manages replicas going from PROD to DR.
1. VBR-2 (on the DR site) manages backup jobs happening on the DR, including those of replicas.

Is that correct?

Also, what do you mean by "dependent"? Backup copy chains can be used separately from the main chain. In case of doing backups from replicas you still have a "dependant" chain as your backups depend on the replica they are based on.

Thanks

Post Reply

Who is online

Users browsing this forum: No registered users and 8 guests