MS SQL Replication larger than 2 TB

[unsichtbarre]

I am facing a situation, new to me, where I have to create a replica (for DR purposes) of a MS SQL Server which is 2.14 TB overall. I have two questions:

Question 1: When I create the job using Veeam Backup 4.1 & ESX 4 and provide no exclusions, then select a destination (datastore) which only has 1.64 TB available, "Check Space" validates my choice as having "Sufficient disk space available" when I reduce the number of rollbacks to 5. How can a complete replica VM which is 2.14 TB exist on a datastore which is 1.64 TB overall?

Question 2: I have two usable datastores on a single ESX 4 Server for DR (1.64 TB and 1 TB available respectively). The SQL Server has 4 Virtual Disks (0:0-30GB, 0:1-1.4TB, 0:2-700GB, 0:3-25GB). When I create two jobs, one for system disks and one for the data, I can no longer even get 5 rollbacks on the 1.64 TB datastore of only the 1.4 TB data disk. What would be an effective method of creating a crash-consistent replica?

Thx, -J

[Vitaliy S.]

Hello John,

1. You need at least the same storage space on your destination datastore to perform replication of 2.14 VM. Could you please clarify if you are choosing replication job and not the backup job option. While doing backup jobs you can use compression which is not used during replication, and you can fit in your 2.14 TB VM to a 1.64 TB datastore.

2. When you click check space button, it calculates estimations based on 10% changes for each incremental job run. But of course this value is just a prediction and depends on the VM itself. You may either reduce your retention policy vaule or continue in any way if you sure that you'll get less than 10% changes for each incremental job run.

Thank you!

[unsichtbarre]

The job is created as a replication. Quite simple, I select the 2.14 TB VM from production cluster, add it to its own replication job, select the 1.64 TB datastore and calculate space.
With the default 14 rollbacks, it confirms that there is not enough space. Whe number of rollbacks is reduced to 5, calculate space determines that space is adequate.
1. How is it possible to replicate 2.14 TB to 1.64 TB? Are the disks thinly provisioned?
2. What is the best way to split the VMs disks onto two datastores at the destination? Can I use two replication jobs and exclusions? We are trying to avoid extending the VMFS volumes across LUNs.

Thx, -j

[Vitaliy S.]

John,

1. It is not possible to replicate 2.14 TB to 1.64 TB, that seems to be a bug in calculation. It would be great if you could send a veeamshell.log to support team for investigation. As for me, I will try to reproduce it in my lab, for this could you tell me if you have thin or thick provisioned disk for that VM?
2. Yes, you can do that. And when the disaster happens you may add virtual disks to your VM configuration using vSphere Client.

Thanks!

[unsichtbarre]

The disks are thick provisioned. Since the job was never run (fortunately, consequential failure possibly leaving snapshots in place on source VM) would there be entries in a log file? Where would I find veeamshell.log?

I would gladly screenshot the process of setting up the job if it helps diagnose the problem

Thx, J

[Vitaliy S.]

John,

Yes, the screenshot would be great, you can find all the log files at Help->Support Information. I will update the thread tomorrow after I speak to a development team. But It does seems like the calculations are not accurate for some reason, so you should choose another destination datastore for your replica.

[unsichtbarre]

I suppose this about sums it up:

[Vitaliy S.]

John,

I've tried to reproduce it in my lab, but I never succeeded in fitting large VM into a smaller datastore. We use the following formula to calculate the estimated destination size, you can use it make your own predictions. But thank you for catching this weird bug, we'll continue investigation on that.