Do WAN Accelerators Help Voice or Video?Do WAN Accelerators Help Voice or Video?

The WAN acceleration tools have a number of tricks. Will this help my voice and video? Not directly, but indirectly.

Clients often ask how WAN accelerators like Cisco WAAS or Riverbed Steelhead can help support voice or video on their enterprise networks. They have often recently listened to a sales pitch from one of the WAN acceleration vendors. These devices often do a great job of reducing WAN utilization and increasing application performance at the same time. But can they help voice and video performance?Well not directly, but indirectly yes. Here is how it works.

The WAN acceleration tools have a number of tricks that they use to accomplish their results. The first is a form of compression. The WAN accelerators work hard to minimize the number of times the same data is sent over the WAN link. This is both similar to and different from standard compression.

When you compress a file using WinZip, the program goes through the file and finds common strings of information. Say, for instance, that one of my files has multiple references to my recent trip to "VoiceCon 2008" in San Francisco. The first time WinZip sees the phrase "VoiceCon 2008" it notes its location in the file. The second time it sees the same phrase, instead of inserting the phrase it inserts a short reference that says "look back in the file by xx bytes and pick up the phrase there that is 13-bytes long." This can probably be done in 2 bytes or 3 bytes, thus saving about 10-bytes in the file.

The WAN acceleration vendors do a similar thing with a library of references. The key to their technology is to find as long a string as possible, and to then store references to that string in libraries on each end of the WAN link. A long string might be a phrase, but it might be a whole Word document or other file.

Say, for instance, that a corporate memo has been sent to everyone in the organization. If I am working with my 19 colleagues in the Acton office and I am the first person to open my email, the full file will be transferred across the WAN link. As it is sent, the WAN accelerators on each end will enter that file into their libraries as a phrase. But when my colleague opens his email, the sending WAN accelerator notices that the exact same file is attached, and just sends a reference. The receiving end looks up that reference in its matching library, extracts the original file and forwards it to my colleague. So for my colleague the file is delivered as expected but response is much faster because the file did not need to be transferred across the WAN.

Note that two key values are achieved in the above scenario. First, the file is not transferred across the WAN the second time, so bandwidth usage on the network is reduced. Secondly, the file is delivered faster, thus the user experiences a faster application. This is a win-win situation.

Note that in my simplified example I talk about a complete file. Most WAN accelerators actually work on chunks of data smaller than a file, and thus they can recognize portions of a file that is almost the same as another file and accelerate it in the same way. Enterprises are often pleasantly surprised at how well this technology works.

But our original question was: Will this help my voice and video? I said not directly, but indirectly.

Voice and video are already highly compressed, and they do not duplicate information. If I say "um" a lot in my phone conversation, you might think that compression could cut those out, but my tone is a little different and the background noise of the dog barking is different, so this type of compression is not yet possible. Voice codecs do sometimes stop transmitting if the speaker is not speaking, but otherwise the information must be sent. The same is true of video. So it is usually not possible to reduce the volume of data sent by a voice or video conferencing device.

However, the fact that WAN accelerators reduce the bandwidth used by data traffic often opens up the link and makes bandwidth available for voice and video. The big bursts of data traffic that are needed to support data application loads are managed by the WAN accelerators as well, making the data traffic better behaved. So the indirect effect is that WAN accelerators often make room for voice or video and tame the data so the two traffic types can coexist more peacefully on a bandwidth limited WAN link.The WAN acceleration tools have a number of tricks. Will this help my voice and video? Not directly, but indirectly.