90 likes | 206 Vues
Victor Pascual Avila, John Elwell. IETF 76 DISPATCH WG ad hoc meeting on SIP Identity. History. Various drafts published in 2008 timeframe, proposing a number of solutions Also drafts relating to the E.164 problem and to identity handling at a UA Presentation by Jon Peterson at IETF#73
E N D
Victor Pascual Avila, John Elwell IETF 76 DISPATCH WGad hoc meeting on SIP Identity
History • Various drafts published in 2008 timeframe, proposing a number of solutions • Also drafts relating to the E.164 problem and to identity handling at a UA • Presentation by Jon Peterson at IETF#73 • Informal “design team” held conference calls between IETF#73 and IETF#74 and developed draft-elwell-sip-e2e-identity-important
History (continued) • Discussion at IETF#74, based on draft-elwell-sip-e2e-identity-important-03 • Mixed feedback • Desire to see requirements stated more clearly • Mixed opinions on whether to pursue in the way it was at that time being pursued • But consensus that discussion should continue – need a different way of pursuing it • Draft-elwell-dispatch-identity-reqs-00 published before IETF#75, but no discussion there • Draft-elwell-dispatch-identity-reqs-01 published in October, reflecting comments received
What are we trying to achieve? • From IETF#74 presentation: • The ability for a participant in a communication to know with a high degree of certainty who the other party (caller or callee) is and that information (e.g., voice, text, video) is being sent to / received from that other party • “e2e authenticated identification” • to work in a large number of practical deployments
What is the problem that is preventing this? • From IETF#74 presentation: • Intermediate B2BUAs modify signed parts of request • Modify SDP for media steering • Modify call-ID, contact, etc for topology hiding • Intermediate B2BUAs modify E.164-based SIP URIs • Canonicalization • Both practices break e2e authenticated identification as currently defined
Current status • Draft-elwell-dispatch-identity-reqs-01 expresses requirements • A number of expired drafts propose solutions, which can be roughly categorized as follows: • reducing the amount of signed information compared to RFC 4474 (to avoid the need for breaking the signature when changing information en route) • signing a copy of certain information, so that if the original information is changed en route, the signature is still valid and the copy of the original information is still available) • performing a return routability check of some form
Current status (continued) • It is recognised that there are difficulties with E.164 numbers, in the sense of what a domain can assert about an E.164 number. However, since E.164-based SIP URIs are by far the most common (for voice at least), we should try to do something with these. • There seems to be a consensus we can't do much about PSTN interworking. The most we can do for a call from PSTN is to provide some authentication of the gateway that asserts the calling number (based on the assertion received from PSTN).
Where to go from here? • Are requirements heading in the right direction? • Should we now try to map different known solutions to these requirements to assess what is covered/missing? • From there, try to assess whether a generic solution is possible or whether we need a small number of solutions for different use cases.
Questions for the group • Do we agree we have a problem? • Is this stated clearly in any of the prior drafts (e.g., draft-elwell-sip-e2e-identity-important)? • Or do we need to state the problem in some other way? • If we have a problem, what should be the next step? • Map known solutions to requirements as suggested on previous slide? • Who is willing to be involved in ongoing work (e.g., drafting, reviewing, proposing solutions…)? • Is it appropriate to keep this as a DISPATCH topic and try to take more positive steps at IETF#77?