Review: WAN Optimization Solutions

I recently had the opportunity to test out and implement some of the top WAN optimization products on the market today. I was amazed at the science behind these awesome devices and how they were able to “speed” up long distance data transfers.

WAN Optimization has been a bit of a buzz word for the past several years. The need for faster information, bigger pipes, and the need to reduce circuit costs has spurred a need for devices that will make all of that happen.

I had the pleasure of reviewing three different appliances that were designed for WAN Optimization.  Our test scenario was to move an entire data center from Sterling, Virginia, USA to Austin, TX.  The media we used was a 40Mbs TimeWarner metro ethernet circuit.  This was a straight layer two circuit, with no routing. This essentially enabled the two locations to act as one virtual location via the switched metro-ethernet.  We invited three vendors, Juniper, Cisco, and Riverbed, to bring in their appliances for an unofficial bake-off.  Each vendor was explained the architecture, were to provide two units, one for each end of the circuit.  All said their product could do the job the best. Lets see how they do…

Juniper

First to the gate was Juniper who brought in their WXC Wan Accelerator.  The Juniper Sales Engineers were top notch, came on site and helped us get the WXC installed and configured properly. For our scenario the appliance would need to be in-line with the traffic.  Juniper Engineers explained that they were more accustomed to being in a routed environment but assured us the WXC would accelerate only layer 2. Needless to say after weeks of tweaking, we saw little acceleration in our given layer 2 environment. The final suggested solution from Juniper was to create a layer 3 routed segment for the WAN.  This was not an option so it was on to Vendor #2.

Cisco

Cisco is amazing.  During our initial conference call with them, Cisco totally understood the test scenario, was extremely knowledgeable and immediately on the ball to determine our sizing requirements for testing the  Cisco Wide Area Application Services (WAAS) .

Riverbed

While waiting for Cisco to determine the “sizing requirements” we decided to give Riverbed a call.  From our experience it would take Cisco a few days to come up with a demo unit for us to test.  For Riverbed however, one phone call and we had the regional sales guy at our offices the very next day. We then had a demo unit, a Steelhead 6050,  in our racks 3 days later with another one shipped to the distant end in Virginia 3 days after that.  The speed in which Riverbed was able to get product into our hands was amazing but could they actually do the job?

This test would be another in-line test as our architecture was layer 2 only.  During the next change window we got both units in the traffic path and turned them both on. Immediately we saw a dramatic improvement. One of the tests we ran was to transfer a 1Gig file across the WAN without optimization. We used FTP as the file transport method. This I believe took approximately 40minutes.  With Optimization enabled – it took 8 minutes.  (I should have a disclaimer here – our tests were not scientific in the least – it was extremely late and these times are not exact – however the increase in perceived throughput was this dramatic) .

Over the next month we were able to completely transfer all of the data from the servers at the Virginia location to the Austin location. Hands down the Riverbed stole the show.

Oh, and Cisco?  Yeah, they never called us back. Finally after several attempts to contact them we were told that the WAAS would NOT function in a layer 2 only topology.

FREAK!!!