newtrk-1 newtrk new ietf standards track discussion bof chair: scott bradner agenda 1/ description...
TRANSCRIPT
newtrk-1
newtrkNew IETF Standards Track Discussion BOF
• Chair: Scott Bradner <[email protected]>• Agenda
1/ description of current IETF Standards Track
2/ observations from problem working group
3/ what other SDOs do
4/ proposals for alternate standards track processes
5/ what would define success in a revised IETF Standards track
6/ open discussion
newtrk-2
Current IETF Standards Track
• same basic track since at least 1988
• RFC 1083 - (IAB Proto Stds) December 1988– 1st stage “Proposed Protocol” (changed in 1990/1)
• RFCs 1100, 1130, 1140, 1200, & 1250
• RFC 1310 - (Stds Process) March 1992
• revision 2: RFC 1602 - March 1994
• revision 3: RFC 2026 - October 1996
• 3.? stages
newtrk-3
RFC 2026 Standards Track
• (Internet Draft)• Proposed Standard (PS)
– good idea, no known problems• Draft Standard (DS) (min 6-month wait)
– stable– multiple interoperable implementations– note IPR restriction
• Internet Standard (STD) (min 4-month wait)– wide use
newtrk-5
2.4 Three Stage Standards Hierarchy not properly Utilized
Input from the problem WG
draft-problem-issue-statement-05
Elwyn Davies (editor) [email protected]
newtrk-6
HEALTH WARNING
• The problem WG ….emphasize that
both the long list of problems and the root cause issues that we have derived from them are problems that are believed to exist by a significant constituency [in the IETF]
newtrk-7
Root Cause #4(in no particular order)
Three Stage Standards Hierarchy not properly Utilized
• Intention (RFC2026): Proposed(PS) Draft(DS) Full(FS)
• Situation today: Effectively… A single stage - Proposed– Relatively few standards progress beyond PS
newtrk-8
Subversion and Compression
IETF aims to produce effective standards:
• Demonstrated with running code
• With multiple, interoperable examples
• Matured by experience
Reduction to a single phase…
• Subverts the aims
• Compresses maturation process
newtrk-9
Perception of Higher Quality Bar • Perception: IESG has raised the bar for PS
– Need to specify a complete system rather than just an interface
• BUT.. Quality checked by thought experiment (mostly)– PS does not require multiple running and
interoperating code instances
newtrk-10
Howlround in Standards
• A positive feedback loop exists– Increasing pressure on time to market is speeding
up the cycle time – Vendors deploy specifications at PS as if they are
fully matured– To avoid damage to our reputation, we have
responded by trying to make PS specifications ready for prime time
• Results in baked-in problems
newtrk-11
Cautionary Note
Elsewhere in problem issues….
• Need to be aware of the market deadline
• Need to know what the engineering trade-offs are for a piece of work
• Need to avoid perfectionism
newtrk-12
Lack of Aftercare for Standards
• Nobody is responsible for maintenance
• There is no formal bug reporting and tracking system
• Periodic reviews not being carried out
newtrk-13
Inevitable Consequences
The 3-stage standards maturity level process is PERCEIVED
(by some IETF participants) as excessive
newtrk-15
What Other SDOs Do• W3C
– Working Draft (WD)• published for review by the community
– Candidate Recommendation (CR)• a document that W3C believes has been widely reviewed and satisfies the Working
Group's technical requirements
– Proposed Recommendation (PR)• a mature technical report that, after wide review for technical soundness and
implementability, W3C has sent to the W3C Advisory Committee for final endorsement.
– W3C Recommendation (REC)• a specification or set of guidelines that, after extensive consensus-building, has
received the endorsement of W3C Members and the Director. W3C recommends the wide deployment of its Recommendations.
newtrk-17
New Tracks
• draft-bradner-ietf-stds-trk-00.txt
• draft-dawkins-pstmt-twostage-01.txt
• draft-iesg-hardie-outline-00.txt
• draft-hardie-category-descriptions-00.txt
• draft-loughney-what-standards-00.txt
newtrk-18
An Idea for an Alternate IETF Standards Track
• IESG practice raised bar for PS over time– close to old requirement for DS– not let nits go
• vendors implement from IDs– but IDs change and disappear
• little difference between DS & S
newtrk-19
Alternate Standards Track
• (Internet Draft)• Stable Snapshot
– like old PS w/o much IESG review
– immature, pre-standard specifications
– note any omissions from requirement
• Proposed Standard– new PS - IESG cross area review etc
– some implementation experience
• Internet Standard– old DS + S
newtrk-20
Other Ideas
• draft-dawkins-pstmt-twostage-01.txt
• draft-iesg-hardie-outline-00.txt
• draft-hardie-category-descriptions-00.txt
• draft-loughney-what-standards-00.txt
newtrk-22
Two-Stages and a Label <http://brandenburg.com/presentations/tw
ostage-minne.ppt>
Problems
– Onerous barriers
– Unused stages
– Unused process and false advertising
– Uncoordinated use of drafts
– Cruft in Archive
newtrk-23
Proposal
Proposed Standard
– Go build productGo build product
– Completed specification, same as today
– 1 implementation
– 36 month timeout
Internet Standard
– Successful part of InternetSuccessful part of Internet
– Community adoption and use
Working Group Snapshot
– Formal, working groupworking group “synchronization”, eg., go test go test the specthe spec
– Working group consensus on version of Internet-DraftInternet-Draft
– No IETF-level status
– IESG opportunity to comment
– 6 month timeout
newtrk-26
Standards, What Standards?
• The IETF has produced a good body of work.– 3655 RFCs at last count– 63 STDs
• People seem to be using our standards.• But, there are problems.• http://www.rfc-editor.org/errata.html
– Published RFCs never change. Although every published RFC has been submitted to careful proof reading by the RFC Editor and the author(s), errors do sometimes go undetected. As a service to the readers of RFCs, this page contains a list of technical and editorial errors that have been reported to the RFC Editor and verified by the authors or the IESG.
newtrk-27
More Problems
• Relatively few specifications are now progressed beyond Proposed Standard (PS)
• There is no formal bug reporting or tracking system in place for IETF specifications.
• Periodic review of protocols are not being carried out.
• No individual or body is given the task of 'maintaining' a specification.
newtrk-28
Solutions?
• Improved errata pages with hyperlinks?
• Maintanence teams?
• Early assignment of STD numbers?
• Enhanced STD numbers?
newtrk-29
Next Steps
• Incorporate more discussions of solutions.
• Solicit input for which possible solutions seem reasonable.
• Apply test cases to the above.
newtrk-30
Appendix
• http://www.rfc-editor.org/cgi-bin/rfcsearchTest.pl
• Based on your search of [Transmission Control Protocol] in the All Fields field 61 matches were found
• Based on your search of [tcp] in the All Fields field 119 matches were found
newtrk-33
Defining Success
• what would define success in a revised IETF Standards track– more advancement (assuming N>1-stage)– fewer ID-based products– better WG/participant understanding– less press stories saying “IETF standard” when
referring to IDs– other?
newtrk-35
Discussion
• is change needed?
• designated ID stage?
• stage requiring multiple implementations?
• N=? (N-stage)
• maintaining standards
• IPR hook
• other?
newtrk-36
Conclusions
• Scott to say if he thinks there is consensus on specific things– is change needed?– designated ID stage?– stage requiring multiple implementations?– N=? (N-stage)– maintaining standards– IPR hook– other