[Diffserv] Hard questions (was: Diffserv PIB approved as Informational RFC)

[Dan Grossman <[e-mail]>]

I believe that with this, the working group has completed all the items on its charter.
Congratulations to all, especially the chairs and authors.

Now, at risk of creating controversy,  some hard questions:
We have completed all the items we were supposed to.  But is that enough?

In particular, we agreed, at much cost in pain, to create these PDBs, which were intended to be the
technical part of service definitions (changing the name to confuse the innocent).  Yet no PDB was
ever a working group draft, never mind being published as informational.   VW, which  (arguably) is
of the greatest commercial interest,  ran into some hard questions in working group review, and
also got hung up when its authors diverged from working group consensus surrounding RFC3246 and
RFC3248.  The draft has expired and the authors have not chosen to revise it and nobody has chosen
to pick it up.  So we don't have an end-to-end, interoperable definition of how to make a BA
achieve hard delay jitter bounds.    The relative assurances ("Olympic") PDB draft had some
disagreements and I believe has also expired.   Ditto the bulk transport PDB.   Has the group
completely lost interest in PDBs, preferring (to coin a metaphor) a bin full of disassembled parts
to a usable widget?  Or are the problems they expose too hard or too controversial?

Yoram's Diffserv framework draft raised a large number of system issues.  Few of them were ever
adequately addressed.  The view was that these would be resolved by experimental deployment.  Yet
we've seen no report on how successfully experimental deployment addressed any of these issues,
much less specific proposals resulting from experimental deployments.   In particular, there has
been an assumption that there would be a bandwidth broker that would ensure adequate capacity along
all hops in a DS-domain, even in the face of route flaps.    We've all seen journal and conference
papers on various ideas for how such a thing might work.  Yet except (arguably) for the PIB, the
whole subject was declared out-of-scope for the WG, and there has been no progress (at least that
I'm aware of) on another IETF working group activity that would address bandwidth reservations
(although one could argue that RFC3270 might indirectly be a response to this problem).

Finally, it was intended that RFC 2474 and RFC 2475 be updated after a decent interval and taken to
draft.  They were published in December, 1998, or three-and-one-half years ago.   Is it the view of
the group that there has not been enough deployment experience, that deployment experience has been
inconclusive, or that we've all lost interest?

In short, have we tacitly agreed to (in the words of the late Sen. Aiken)  "declare victory and go
home", leaving a partially complete system specification, riddled with holes and built on a
foundation of dubious soundness?  Or are we taking a breather before finishing the job?

Brian E Carpenter wrote:

> -------- Original Message --------
> Subject: Document Action: Framework Policy Information Base toInformational
> Date: Tue, 11 Jun 2002 13:00:48 -0400
> From: The IESG <[e-mail]>
> To: IETF-Announce:;@zurich.ibm.com
> CC: RFC Editor <[e-mail]>, Internet Architecture Board <[e-mail]>,[e-mail]
>
>
> The IESG has approved publication of the following two Internet-Drafts
> as Informational RFCs:
>
>  o Framework Policy Information Base
>         <draft-ietf-rap-frameworkpib-09.txt>
>  o Differentiated Services Quality of Service Policy Information Base
>       <draft-ietf-diffserv-pib-08.txt>
>
> These documents are the product of the Resource Access Protocol (RAP)
> and Differentiated Services (DIFFSERV) Working Groups respectively.
>
> The IESG contact persons are Bert Wijnen and Randy Bush.
>
> The Working Groups had requested for standards track, but there is not
> IESG consensus enough to support publication of these documents on the
> standards track at this time thus we will publish them as Informational
> RFCs. The IAB just held a workshop on network management and the report
> on that workshop is being developed. If the results of the workshop and
> subsequent discussion changes the consensus in the IESG we will then
> issue a last-call to republish the documents as standards track RFCs.
>
> _______________________________________________
> diffserv mailing list
> [e-mail]
> https://www1.ietf.org/mailman/listinfo/diffserv
> Archive: http://www.ietf.org/mail-archive/working-groups/diffserv/current/maillist.html


_______________________________________________
diffserv mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/diffserv
Archive: http://www.ietf.org/mail-archive/working-groups/diffserv/current/maillist.html