Slow Replication, any ideas?

[samrooney]

Hi,

We are replicating 18 VM's from a 3 host esx 4.0 cluster to an esx 4.0 host on our DR site accross the WAN with 20MB/s reserved bandwidth. All ESX hosts have update 2 and patch ESX400-200912401-BG. All replications are being done using VMWare vStoprage API Network Mode. The largest VM is 560GB the next largest is 215GB, the rest asre 30-50GB. We have turned off the NTFS file last accessed attribute on the largest VM and defragged the disks.

We are typically getting a throughput of 5-10MB/s total irrespective of how many vm's replicating at a time.

I have read through any of the previous posts about the same issue and tried some of the suggested solutions but we still seem to be having the same problem, anyone have any ideas on how to speed things up?

Thanks,
Sam

EDIT: also forgot to mention ,we are using vSphere 4.0 and Veeam Backup & Replication 4.1

[Vitaliy S.]

Hello Sam, could you tell us if you've tried using any WAN acceleration tools and what were the results? Also could you tell us if you have those numbers on the incremental job runs or initial run?

[samrooney]

Hi Vitaliy,
Thanks for your reply.
We were considering HyperIP but the price was quite off-putting, the fact that it works on a yearly subscription basis as opposed to a one off purchase of equipment seems a bit steep. Do you know if Cisco WAAS and Riverbed's licensing work on the same principle?
We did the initial run with the DR box here in the office and then shipped it out to the DR site for the incrementals, while it was on the LAN we were getting 200MBp/s to 5GBp/s.
I have contacted the local reseller for riverbed to see if i can get a quote. I'll do some research into Cisco WAAS later too, is there one you would reccommend?
Thanks,
Sam

[Gostev]

Riverbed means up to 98% traffic compression
VEEAM with Riverbed

There are some good feedback on Cisco WAAS and HyperIP in the same thread, too.