Chris Kranky

Recent Posts


Google is really in charge of WebRTC (call me surprised)

Chris KoehnckeChris Koehncke

elephant-in-the-roomIndustry pundits are yapping and various IETF types are writing draft specification documents surrounding WebRTC. It’s all good and it’s how the industry (mostly) works. WebRTC is an atypical communications application in that the client is in charge. Historically communications has been driven from the core outwards. Well times have changed. The clients are in charge and by that I mean ¬†Google’s Chrome and Mozilla’s Firefox browser.

Yap all we want, if Google or Mozilla don’t want to include something in the core browser code. Well, you just wasted your time. It ain’t gonna happen. Google is not trying to act like the elephant in the room, but clearly they are.

To wit is the discussion about SIP over Websockets for WebRTC. Google’s many Stanford types made the careful decision NOT to force a signaling standard into WebRTC. If you want to use SIP, more power to you, as the application developer you can do whatever you want, there are several decent SIP java libraries you can use. In fact, for those more inventive, you don’t even have to use WebRTC as your signaling path. You could resort back to good ole HTTP with a REST interface if that makes you happy.

The VoIP’ers argue that without a common signaling specification cross WebRTC applications will be it difficult to interoperate. My response – so what? Google is a bit more polite.

SIP started off on a rather simple mission to facilitate a communication channel. Unfortunately as the “be all end all” signalling standard, SIP started to get layer upon layer of complexity. Today SIP is far from a common standard. A common framework, yes, standard no. I reeled as one telecom service provider proudly told me recently that they’ve narrowed down the number of SIP profiles they will support to 80! I hesitated to ask how many they had started with. It’s a miracle any communications happens at all.

Much of the industry talk has been how do I take a WebRTC session and bring it back into my SIP infrastructure. This is backward thinking IMHO.

Clearly there will be a need to go from WebRTC to SIP (and reverse), clearly those comfortable with SIP might develop an WebRTC application that use SIP as the signaling engine. But we’re all missing the original goal. The goal is to allow standard HTML developers to quickly develop multimedia communication applications. Forcing them to learn SIP isn’t the way to do that.

Google has smart people, they’re on the right track, less arguing, more doing will propel this forward and open the world of new possibilities to us all.