Chris Kranky

Recent Posts

WebRTC: Does this fix lingering SIP issues?

Chris KoehnckeChris Koehncke

fixitIn communications, the more boxes something has to “flow” thru; the more likelihood that something is gonna get screwed up. I remembered years ago visiting a company called Dynamicsoft, who was then a leader in SIP based solutions.  They presented to me their “vision of the future.” The highlight chart was this “network of networks” which showed layers upon layers of session border controllers, back-to-back user agents (B2BUA) and all sorts of proxies in between. The senior management proudly explained they were leading the charge into this new market and going to profit wildly.

All I could think is my god it’s going to be a miracle that anything works. Unfortunately, looking at any “modern” VoIP network today, Dynamicsoft (may they rest in peace) was not far from the truth on the construct of VoIP networks.

WebRTC enters an reintroduces a concept that has long been dismissed by traditionalists. Namely, that end points should be smart and the network should be dumb.

A challenge with current VoIP is that it’s difficult to create a voice conversation from me directly to the party to whom I wish to speak (or video) without my “voice” packets transiting thru someway point. Peer-to-peer VoIP calls have always been nearly impossible to make happen reliability due to various local network devices and security constraints. No doubt telephone companies weren’t promoting this either usually while waving some “quality of service” banner.

The result of this, if I call my next door neighbor using a VoIP service, my voice packets may well travel thousands of miles back and forth to reach my neighbor 50 feet away. If you don’t see quality of service issues with that let me know.

The standards of WebRTC force the inclusion of a number of elements that dramatically increase the ease and ability for peer-to-peer connections using ICE, TURN, STUN and other capitalized letters. The creators of WebRTC didn’t see this as a challenge to fix, but more of a “oh course it needs to work this way” and simply set about to make it happen with limited fuss.

We’ve all experienced calls that don’t connect, one way audio, DTMF that doesn’t work (god I hate that), echo, static and other Walking Dead sorts of noises. The inevitable cause of all this is that somewhere along this “Dynamicsoft Network” some box failed to do it’s job properly.

Peer-to-peer communications will sharply increase quality and similarly reduce complexity. I think the industry is over looking the inherent power of this very simple concept.