What’s Happening with SIP and IPv6?
At the recent Netnod Spring Meeting Olle E. Johansson from Edvina was invited to talk about his work with dual stacks in SIP. The talk was very personal and covered the progress, or lack of progress, in the IETF and the work done in the SIP Forum IPv6 working group.
The work started when Olle detected strange issues when testing the IPv6 implementation in Asterisk. It wasn’t the code, it was something else. After testing IPv6 at the SIPit events, work started in a working group of the SIP forum that later resulted in documents contributed to the IETF – one issue was adopted by the IETF SIPcore working group but is not seeing any interest so it doesn’t move forward. In the IETF, strange resistance was met and not much support for doing any changes or additions to the protocols in order to fix proven problems. In fact, quite a surprising opposition was met.
How do you fix issues in an old protocol?
Should it really be this hard to fix issues in a published protocol? Are SIP and HTTP the only protocols that have issues with dual stacks or is it just the two protocols that have been tested in detail?
Further tests at SIPit has not only confirmed the documented issues, but also revealed issues in many implementations that has been supporting IPv6 for a long time. Source address selection was missing. There’s a lot of work to do before we can get SIP implementations ready for dual stack use, both in development and in the standards.
The presentation is available on slideshare. We apologise that it is made for live presentation and doesn’t cover all the details on the actual slides, but it may still give some interesting facts.