WAN Replication - 15 hours to complete on 4Mbit WAN link

[andreapichler]

Hello.
I tried to use Veeam Backup & Restore to replicate some VM from one ESX cluster to another one in a remote location.
I started with a server with 45Gb of disk space. Veeam 4.1 is installed in the "source" site on a Physical server connected via FC to the Storage (I Used as Backup mode: SAN/NBD with changed block tracking).
The target is an ESX 4 Host on a site connected with a 4Mbit connection. The first replica had a duration of 24:22:58.
Now I started a second replica and the time remaining is around 15 hours (the job started 25 min ago). The first job was completed 2 days ago and the server has a very low activity (on the disk, except few logs nothing has changed).
Is something wrong in my job setup (VMWare Storage API/SAN with failover - automatic replication integrity checks enabled - Inline deduplication enabled - Compression: best - Use changed block tracking - rollback points to keep: 3)?
We planned to use Veeam to replicate also the fileserver but looking at this test job, it's impossible.
Thanks
Andrea

[Gostev]

Hello Andrea, remaining time does not show precise time if the job was started recently. Please wait for the second replication pass to complete, and looks up the produced VRB file size on the target. From that it should be clear what was the real amount of changes since 1st pass, and whether the issue is your link speed.

Giving your workload description I would expect on-site replication to complete under 3-5 minutes for your scenario, so only things I can suspect right now are WAN link speed or amount of changes needed to be replicated. 4Mbit is actually quite slow link, I have 7Mbit at home and I can say am not very happy with it

Please post an update when you have more information.

[andreapichler]

Hi Gostev.
I scheduled the job to run each hour.
Here the last results:
8 of 8 files processed

Total VM size: 45,00 GB
Processed size: 45,00 GB
Processing rate: 6 MB/s
Backup mode: SAN/NBD with changed block tracking
Start time: 27/12/2009 13.22.15
End time: 27/12/2009 15.21.46
Duration: 1:59:30

Checking retention policy... 1 rollback point(s) have been deleted.

Is the configuration of my job correct (see first post)?
What else can I check ?
Thanks
Andrea

[Gostev]

Andrea, yes your configuration is correct, and based on these stats change tracking works as expected (job's processing rate is 48 Mbit per second, whereas your declared link speed is only 4 Mbit, and in reality probably even lower). Results are not bad either for such a slow link, the bottleneck here is clearly your WAN link speed. If you look up VRB file size produced by this replication pass on target, and divide that by 2 hours, you can find out your effective WAN link speed.

One thing you can try to do to improve the replication performance is to move the swap file to a separate virtual disk on source VM, and exclude that disk from replication. Please lookup the existing threads on forum discussing this technique for more information.

Thank you.

[Gostev]

Wow, above is my post number 3000