Presentation at IIT VoIP, October 24, 2008
The IETF SIP specifications over the last decade have surpassed an astonishing 1,000 pages of text in well over a hundred documents. Within those documents are many optional protocol elements. From a computer science perspective the number of options results in a combinatorial explosion of option handling code, resulting in complex implementations, as well as a whole industry for “protocol repair.” If we have a situation of “too much stuff,” what can we do? Basic interoperability testing clearly is a start. This is where one can find out both if one’s implementation works, as well as learning from other’s experiences. However, further simplifications to the protocol, from a computer science and a practical perspective, can greatly help achieve the goal of better and more robust interoperability. This talk will discuss industry efforts in interoperability testing, approaches for profile and protocol simplification, and experiences with profiles, both good and bad, and how they can lead to better, or worse interoperability.