First WAN Replication Job Running Long?

[dgerler]

Good Morning,
I'm still in the trial of Veeam B&R and want to make sure I'm doing this right.

I am putting together a DR plan with the foundation of it being replication between the two sites via the WAN. I have two VMware hosts currently running ESXi 4.1 at the production site and one at the DR site running ESXi 5. I have a separate physical windows server 2003 at the production site acting as the backup server. I have made separate physical servers at both the production and DR sites as backup proxies as well as backup repositories. so it looks like this:

Production site:
2 VM hosts run ESXi 4.1 using local storage for VMs
Backup Server
Backup Proxy
backup repository


DR Site:
1 VM host running ESXi 5 using SAN storage for the VMs
backup proxy
backup repository

As a precursor, I created a backup of the VM's and transfered them to the DR site to make the seed.
I configured the replication job for low bandwidth, separate virtual networks and different IP scheme. I set it to use the backup proxy at the DR site for both source and target thinking of the seed, but now I'm wondering if I made a mistake on that. I also configured seeding and mapped the replicas and enable image processing.

The job includes two VM's with combined size of 400GB.

I ran the job the first time yesterday, but after the seeding was done it slowed way down and I ended up cancelling the job because of production requirements. So tonight I restarted the job and It's been running for 8 hours and is only 9% complete.

Does this sound typical for the first run of a replication job?
Will it take this long every time it runs?
Should I change the source proxy to the production site?
I'm working on creating backup proxy VM's at each location. How much will that improve performance in a WAN repliaction job like this?

[foggy]

David, when replicating across sites, you should deploy at least one local proxy in each site and also it's better to install a backup repository server closer to the source (i.e. in the production site, this is explained in the UI) as it stores and processes replica metadata. Thanks.

[dgerler]

Thanks.
I setup several jobs, mostly backups so I had gotten a little familiar with it so did n't read it this time.
I have stopped the job, made the changes and restarted it. We'll see how that changes things. I know the realtime statistics was saying the bottleneck was the source so maybe it'll run faster now.

Would you say the first job runs much slower than sibsequent jobs?

[foggy]

Hard to say, depends on the amount of changes your VMs produce, time elapsed since the seed was created, periodicity of subsequent runs, etc.