[e2e] Transport Protocol Road Map(s)
braden at ISI.EDU
Fri Sep 19 13:50:54 PDT 2003
*> I quickly incorporated some of the suggestions that already arrived.
*> I'll be away for the next week, but I'll address comments as soon as I
*> get back.
I appreciate your continuing work on this.
*> Aside from the feedback points I requested previously, some other things
*> to consider:
*> - What is the appropriate status of this document? Is it a readme on
First of all, this is not a document yet, it is only initial input to
what could become a useful document.
*> the IETF website? An RFC itself? In what capacity does it become a
*> living document?
An RFC would be my eventual objective. I don't think that the core
specifications change very fast, so an archival checkpoint document is
appropriate (As an example, I claim that RFC 1122 is still very useful
and important, although it is obviously seriously out of date in many
respects.) In the event of significant updates, Internet Drafts can be
written, leading eventually to a new RFC. But TCP specs don't change
so awfully fast, really.
As far as process goes, after this large "committee" has produced
a useful-looking document, it might be introduced into the
Transport Area Working Group via the Transport Area directorate.
*> - Is there some housecleaning that needs to be done? I think 2581 and
*> 2988 should become standards. If someone has proposed a TCP option,
The important distinction is standards track vs non-standards track.
For all practical purposes today, a Proposed Standard or a Draft
Standard is (nearly) is a standard; S, D, and P are very nearly
equivalent. It seems very unlikely that the energy exists to advance
either of these documents.
*> where the syntax is universally accepted, does that make it a standard
*> (I have 1323 in mind)? Any EXPERIMENTALs that should become PROPOSED
That is a much bigger issue. Are there plausible candidates for this?
How do we know when the experiment is done?
*> Section 6.2 of RFC 2026 indicates that proposed standards may be
*> sunset after 24 months and reviewed annually thereafter. I think there
*> are a few listed below which deserve this treatment.
This part of RFC 2026 (which I helped frame) has never been enforced as
far as I can recall.
More information about the end2end-interest