All of a sudden, it feels like it is okay to talk about packet loss.
For years and years and IP years, packet loss was just an ugly truth and nothing could be done to fix it (without adding latency.) A conspiracy of silence suppressed even passing mention of packet loss in polite company. Way back at the back of the building those network geeks could see the packet loss: all network test and measurement systems include packet loss stats. The geeks never talked about it, though, except when it was clear that the loss could be traced to an errant backhoe operator.
Interestingly enough, there is another group that can see packet loss. That group is the users of videoconferencing systems. The management console of every videoconferencing system includes packet loss stats and with a bit of training you can make the connection between a halting, smeared, and artifacty videoconference call and the endemic phenomenon of packet loss. My theory is that Tandberg and Polycom got bold about it when they made the switch to IP. They couldn’t pretend that everything was just as good as when they were running over ISDN. They had to acknowledge the packet loss, if only to be able blame somebody else but to their credit, they have also tried to do something about it.
Intelligent Packet Loss Reduction (IPLR) is Tandberg’s name for their effort to fix it and Lost Packet Recovery (LPR) is Polycom’s current offering. The two big dogs of video conferencing have exposed the elephant insofar as their own technologies and markets go. Outside of that world, there still hasn’t been much public corporate talk about packet loss on IP networks but that may be about to change. The reason is Unified Communications (UC).
We are all hearing and reading a lot about UC. All of the big name software vendors and network gear makers have offerings in the category. I think my uncle might be working on a UC offering. His dog, too, but here’s the thing: UC by anyone’s definition includes support for rich media and videoconferencing functionality. And of course, UC has to run over IP networks or none of it makes any business dollar sense. But videoconferencing over IP—whether part of a UC suite or as a stand-alone—will always suffer the impact of packet loss because packet loss is endemic to IP networks. (I like that word. Endemic. It says exactly what I mean about packet loss and IP networks.) And, all real-time communications applications need protection from packet loss right across the board. In contrast, LPR and IPLR are both intended to mitigate packet loss damage to only the video part of a videoconference call and only a videoconference call between systems from the same vendor. Neither of these packet loss solutions are of any use outside their own restricted operating environments.
What is needed is a packet loss protection technology that operates at a lower layer of the communications stack and is vendor agnostic and fully supportive of any real time network application but supportive in an adaptive and progressive way. UC vendors know this and that is why it is okay to talk about packet loss. We’re talking to few of them and I believe those talks are going somewhere helpful.
It’s just like my uncle says: you can’t fix what you won’t acknowledge.
