Since there was some interest on this topic and I'm still working with this gear in the lab I thought I'd dive a little deeper into what I've been finding and hopefully what you all might be interested in.
There definitely seems to be some value in measuring how these appliances affect network traffic and network application performance. However, there are several different ways to measure this. In my case, I'm monitoring the traffic from the ingress interface of the WAN accelerator and comparing it with the ingress interface of the adjacent WAN device. By monitoring it in this way, I'm able to clearly articulate the "before and after" effects on the traffic. Also, in some cases where I have two sites that are very similar in traffic amounts and patters, I'm able to compare the traffic of a device where we've deployed a WAN accelerator and sites that we haven't.
How would you optimally like to measure/monitor this traffic?
Josh
We use Riverbed Steelheads everywhere now. I look at the LAN versus the WAN on the same accelerator. If the Steelhead is accelerating to or traffic is being proxied on remote Steelheads the WAN should be pushing a lot less than the LAN is receiving on the same accelerator. Our biggest initial indicators were simply the Orion reports on WAN interfaces. We used to run steady at 60-70% utilization with 100ms latency and after the Steelheads were implemented we see 20-30% utiization with 70-80ms latency.