50 likes | 137 Views
Early Media Indirection (EMIND). Brian Stucker Nortel Systems/Standards Architect March 8 th , 2007. History/Mechanism. Draft proposes a thumbnail solution to early media for consideration/discussion by the working group.
E N D
Early Media Indirection (EMIND) Brian StuckerNortel Systems/Standards ArchitectMarch 8th, 2007
History/Mechanism • Draft proposes a thumbnail solution to early media for consideration/discussion by the working group. • Next step from draft-stucker-sipping-early-media-coping-03 which identified various problems that exist within SIP for early media today. • Mechanism uses secondary dialogs to specifically handle early media. • URL supplied via new header “Early-Media” in response message. • Discrimination of early/final media streams handled by using a unique dynamic RTP payload for early media endpoints. • Endpoint can switch over to next media stream when it detects a change in payload type to the preferred media stream, or additional “Early-Media” URL in SIP provisional. • Alternative to RFC-3959. • Minimal new protocol requirements.
User A supports emind, calls B. Receives Early-Media header in 180 w/ URL. Proxy inserts Early-Media header. Generates secondary dialog to Early-Media URL. Uses unique RTP Type for media to detect final media later. Early media flows to user. User B answers, begins speaking. User A detects RTP arriving with “final” RTP type. Switches to final media (no clipping). User A ends dialog with early media endpoint.
RFC-3959 (Early Session Disposition Type) • RFC uses overlapping SDP offer/answer exchanges in the same dialog. • Any implementations of this technique? • RFC mentions using a separate dialog. • RFC contains a list of advantages it has over the separate dialog approach: • It is simpler • Lack of synchronization problems, all early dialogs terminated upon session acceptance. • Does not require GRUUs. • Does not introduce service interaction issues related to services that may be wrongly applied to the new dialog. • Easier firewall management.
Comparison to RFC-3959 • 3959 is simpler? • Probably not simpler in many, if not most cases. • No BBUAs are necessary if client establishes separate dialog. • Possible synchronization problems? • Client knows to end early media dialog when RTP payload type switches in media stream to the “final” payload type. • Requirement to use GRUUs? • Usage of a GRUU for the early media dialog is entirely optional. • Many types of early media are not provided by an element associated with the final set of endpoints that may accept the session. • “Does not introduce service interaction issues related to services that may be wrongly applied to the new dialog?” • What? • Better Firewall management than RFC 3959 • Client can reuse the interface established for “final” media for early media dialogs. • No special SBC behavior, etc.