Using Veeam replication for long distance P2V?

[MattG]

I need to migrate 10 physical servers from SiteA to SiteB with a 100Mb WAN connection between them. I was thinking of:
- Perform P2V to local ESXi server in SiteA
- Use Veeam replication job to replicate VM from SiteA to SiteB. I would use removable storage option to seed the initial replication.
- Copy the seed replica to the destination ESXi host in SiteB.
- Connect to Veeam console in SiteA and restart replication job to catch up on changes
- Perform a VM failover from SiteA to SiteB using the Veeam console in SiteA.
- Leave VM in SiteB as production and kill the replication job

Does this make sense? Any gothcas? Will the SiteB replicated VM be assembled as a normal VMDK file and ready for production or do I need to perform another task?

-MattG

[Vitaliy S.]

Hello Matt,

Your plan makes perfect sense for me. All your replicated VMs will be ready to go for production.

Though if you want to delete all replication jobs as soon as you perform a failover, then you may want start VMs on Site B using vSphere Client. If you do that with Veeam console, please don't forget to remove snapshots that would be automatically created on replicated VMs (Site B) during the failover.

Thank you!

[MattG]

Vitaliy,

Thanks for the quick response.

So I am assuming that during replication that the CBT files are sent to SiteB and then a VMware job is submitted to reconstitute the VMDK files with the change blocks or does the replica VM have the initial VMDK with each replication cycle's CBT file?

So what would be the steps to get the replica up as the primary and bring down the source VM?

I am thinking something like this:
- Power off VM in SiteA
- Force replication cycle after power off
- Using VI Client power on VM in SiteB
- Change IP if needed
- Do I need to do anything else with the disks or snapshots, or are snapshots only an issue with using Veeam to do the failover?

Also, will I be able to determine when the last replica was sent to the remote site and it's size or do I need to guess based on the replication job's schedule?

Thanks again,
-MattG

[Vitaliy S.]

Matt,

At each incremental job run only changed blocks will be transmitted to Site B, those changes will reconstruct initial VMDK file. Yes, your steps are absolutely fine. There is no need to do anything with disks or VMs provided that you start replicated VMs manually.

And you don't need to guess, just check out Session tab in the backup console, there you'll see all the replication sessions that you had recently.

Thanks!

[MattG]

Vitaly,

So I:
- Create replication job to run manually
- Replicate to removable storage
- Copy from removable storage to remote site's VMFS
- Restart the replication job (should only copy change blocks between initial seed replication and now)
- Power off source VM - to guarantee that I don't miss any data
- Complete one more replication cycle (should be short)
- Rename source VM to _old
- Rename replica to SRV name (remove _replica)
- sVMotion the SRV to a new datastore (this will move it out of the Veeambackup directory and remove vrb and vbk files)

Did I miss any steps? Does this seem right for the task at hand?

-Matt

[Vitaliy S.]

Hi Matt,

Sorry for the delay. Yes, your approach looks absolutely right.

Thank you!

[st0623]

Matt,
If you want to replicate your initial seed or just speed up your incrementals, take a look at HyperIP to optimize replication. http://www.netex.com. We are a Veeam Technology Partner and would be glad to let you try the product free for 30 days. We could saturate your 100Mbs WAN, or rate limit it, then do some compression of the data stream. My email address is steve.thompson@netex.com. Reach out and we can talk about it.

[MattG]

Is it possible to replicate SQL server without VSS and without losing transactions?

I want to:
- leave SQL VM running
- begin the replication job without VSS
- After large intial replication job finishes, shutdown SQL VM and run replication job again
- Register the replica on the destination as the primary VM and remove the original source.

Since I am shutting down the VM with the final replication, would this ensure that all blocks and transactions would have survived this entire procedure?

Thanks,
-MattG

[Vitaliy S.]

Is it possible to replicate SQL server without VSS and without losing transactions?

Could you please elaborate why don't you want to use VSS for SQL server? What applications does this SQL server host?

Since I am shutting down the VM with the final replication, would this ensure that all blocks and transactions would have survived this entire procedure?

Oh...I see where you're pointing to, I assume the replicated VM should be in consistent state.

[Gostev]

Yes, the replica image will be consistent if last incremental pass is done on properly shutdown VM.