WAN Accelerator resource utilization

[stephenc]

We're trying to copy a ~7TB data between two sites for retention, and deployed a pair of WAN accelerators hoping to speed things up a bit. Both the B&R core and wan accelerators (WA) are on individual VM's. Both source and target WA's have 4 CPUs on 500GB SSD.

My question is, the source WA is showing only about 20% CPU utilization with occasional 90% burst. Is this the normal behavior, or is there a way to make utilization higher?

The bottleneck breakdown also shows something strange with target WAN constantly at 0%.
Source 5%
Source WAN 92%
Network 98%
Target WAN 0%
Target 81%

Suggestion on how to optimize this is much appreciated.

Stephen

[foggy]

Hi Stephen, what kind of link is involved and what bandwidth mode is configured on both WAN accelerators?

[stephenc]

10G VPN tunnel. But our network admin did traffic shaping to bring it down to a percentage of that for backup traffic.

[PetrM]

Hello,

By the way, what is the processing rate shown in job statistics and how far it is from what you're trying to achieve?
The Network is the bottleneck ( 98 %) according to statistics provided so it makes sense to test bandwidth with iPerf and to compare
the results of test with the job processing rate.

The low-bandwidth mode is fine for links slower than 100 Mbps and high-bandwidth mode is recommended for WAN connections faster than 100 Mbps.
Please note that the direct data transfer without WAN accelerators will be still preferable for connections faster than 1 Gbps.

Thanks!

[stephenc]

Processing rate is 38 MB/s, and we are running high-bandwidth mode on the accelerators.

Don't have a target number in mind, just curious if we are gaining anything with the accelerators. I was hoping the WAN Accelerators would reduce the transfer size more than it is doing right now.

Thanks for the responses.

[foggy]

I'd check with your network admin regarding the exact allocated bandwidth and maybe try low bandwidth mode based on that.