Upload
rashad-campos
View
15
Download
0
Embed Size (px)
DESCRIPTION
The new bis. Why rewrite the specification?. IESG said so RFC2543 was never the model of clarity to begin with Bis got worse with micro-editing Symptoms Repitition of material in many places No overview of operations Structure not obvious - PowerPoint PPT Presentation
Citation preview
The new bis
www.dynamicsoft.com9th SiPiT
4 Dec 2001
Why rewrite the specification? IESG said so
RFC2543 was never the model of clarity to begin with
Bis got worse with micro-editing
Symptoms Repitition of material in many places No overview of operations Structure not obvious
Decision made at August IETF to move forward full steam with a rewrite
Goal: Preserve bis-04 normative content
www.dynamicsoft.com9th SiPiT
4 Dec 2001
How was it done? Recruited a bis rewrite team
Jonathan, Henning editors Added four co-authors to help write
Gonzalo Camarillo (Ericsson) Jon Peterson (Neustar) Alan Johnston (Worldcom) Robert Sparks (dynamicsoft)
Coaching from Dean Willis, Brian Rosen Project Management from Rakesh Shah Technical writing from Jean Mahoney
Jonathan prepared new outline and defines mapping of existing text to new outline (early Sep.)
Sections assigned to each writer
www.dynamicsoft.com9th SiPiT
4 Dec 2001
How was it done? Iterative approach utilizing bugzilla and cvs
First rewrite of each section done mid-September Several cycles of cross section review and revision Detailed verification of preservation of MUST/MAY/SHOULD
Brutally painful! Deviations captured in changes section
bis-05 complete and submitted 10/26/01
www.dynamicsoft.com9th SiPiT
4 Dec 2001
New Structure Semantically oriented
Present SIP as a layered protocol Message layer Transport layer Transaction layer Transaction users
Message layer Self explanatory – message formats
Transport layer Manages persistent connections Listens for requests and responses Via rules for sending responses, inserting received param
www.dynamicsoft.com9th SiPiT
4 Dec 2001
New Structure Transaction Layer
Reliability Request/Response Matching ACK generation for non-INVITE
Transaction Users Manage dialog/session semantics INVITE-200
www.dynamicsoft.com9th SiPiT
4 Dec 2001
Outline1. Intro
2. Overview of Functionality
3. Terminology
4. Overview of Operation
5. Structure of the Protocol
6. Definitions
7. SIP Messages
8. General UA Behavior
9. Canceling Requests
10. Registrations
11. Querying for Capabilities
12. Dialogs
13. Initiating a Session
14. Modifying a Session
15. Terminating a Session
16. Proxy Behavior
17. Transactions
18. Transport
19. Security
20. Common Message Components
21. Headers
22. Response Codes
23. SRV
24. Examples
25. BNF
www.dynamicsoft.com9th SiPiT
4 Dec 2001
Transport Layer Changes ACK-200 is officially a different transaction
ACK non-200 is part of the transaction
Same EXACT transaction machine for proxies and UA
Handling for INVITE 2xx response *NOT* part of the transaction layer!! UA state machinery retransmits 2xx and ACK Allows transaction machines to die instantly when 2xx received
Transitions based on timeouts, not # of retransmits, to unify machine between UDP, TCP More aggressive transaction timeouts defined for TCP
Proper RTT estimation defined
Actual diagrams for non-INVITE transactions included
www.dynamicsoft.com9th SiPiT
4 Dec 2001
INVITE Client FSM |INVITE from TU Timer A fires |INVITE sent
Reset A, V Timer B fires INVITE sent +-----------+ t.o. to TU +---------| |---------------+ | | Calling | | +-------->| |-------------->| +-----------+ 2xx | 300-699 | | 2xx to TU | ACK sent | |1xx | +---------------+ |1xx to TU | | | | | 1xx V Timer C fires | | 1xx to TU -----------+ t.o. to TU | | +---------| |-------------->| | | |Proceeding | | | +-------->| |-------------->| | +-----------+ 2xx | | 300-699 | 2xx to TU | | ACK sent, | | | resp. to TU| |
| | | NOTE: | 300-699 V |
| ACK sent +-----------+ | transitions | +---------| | | labeled with
| | | Completed | | the event | +-------->| | | over the action
| +-----------+ | to take | ^ | | | | | Timer D fires | +--------------+ | - | | | V | +-----------+ | | | | | Terminated|<--------------+
| | +-----------+
www.dynamicsoft.com9th SiPiT
4 Dec 2001
INVITE Server FSM |INVITE |pass to TU, send 100 INVITE V send response+-----------+ +--------| |--------+101-199 from TU | | Proceeding| |send response +------->| |<-------+ +-----------+ 300-699 from TU | |2xx from TU send response | |send response | +-------------------+ | | INVITE V Timer G fires | send response+-----------+ send response | +--------| |--------+ | | | Completed | | | +------->| |<-------+ | +-----------+ | | | | ACK | | | - | +------------------>+ | Timer H fires | V fail to TU | +-----------+ | | | | | Confirmed | | | | | +-----------+ | | | |Timer I fires | |- | | | V | +-----------+ | | | | | Terminated|<---------------+ | | +-----------+
www.dynamicsoft.com9th SiPiT
4 Dec 2001
Dialogs Equivalent to call-leg from bis-04
Call leg has been eradicated from the spec Generalization, presence
Dialog procedures are no longer INVITE specific Maintenance of CSeq, Route sets Construction of mid-dialog requests General construction guidelines
www.dynamicsoft.com9th SiPiT
4 Dec 2001
Other changes Collected BNF
BNF now uses explicit LWS
Responses no longer need to transmitted over TCP for server transactions! Does NOT include INV-2xx
CANCEL can’t be sent until 1xx received
BYE can’t be sent by UAS until ACK received
CR or LF alone deprecated
3xx to re-INVITE allowed and specified
Radical surgery on multicast No special treatment at ALL except
deciding where to send the messages Assumes only a single respondent If there are more than one, responses
look like retransmits Still needs more refinement in spec
Proxies no longer forward 6xx on receipt
CANCEL first, then 6xx after all responses
Merged requests detected only at endpoints
Serverfeatures integrated
100rel will be integrated
SDP extracted to a separate I-D
www.dynamicsoft.com9th SiPiT
4 Dec 2001
Big changes to –05 (so far) CANCEL and ACK contain the same Route header fields as
their associated request.
www.dynamicsoft.com9th SiPiT
4 Dec 2001
What’s not stable? SRV functionality will change
Under discussion with IESG Likely to be much simplified (no merging of transports)
Route/Record-Route Loose routing
Additional rigor and explanatory text needed Registration section Security section
www.dynamicsoft.com9th SiPiT
4 Dec 2001
Next steps Closing open issues
Explicitly called out in bis-05 where possible Currently all being tracked in Bugzilla 49 open issues, vast majority are minor
Major open issues: Loose source routing, proxy route processing Multicast operation Maximum messages sizes SRV
Target completion to send to IESG before 2002
Information Resource Robert Sparks+1.972.473.5467sip:[email protected]:[email protected]