Summary of editors call

["Mark Day" <markday@cisco.com>]

On the call:

Lisa Amini <aminil@us.ibm.com>
Abbie Barbir <abbieb@nortelnetworks.com>
Brad Cain <bcain@mediaone.net>
Mark Day <markday@cisco.com>
Abhi Deshmukh <Abhi.Deshmukh@apogeenet.com>
Fred Douglis <douglis@research.att.com>
Alex French <afrench@vianetworks.com>
Don Gilletti <don@cacheflow.com>
Jay Guha <jayg@apogeenet.com>
Martin May <Martin.May@activia.net>
Phil Rzewski <philr@inktomi.com>
Paul Scanlon <Paul.Scanlon@qwest.com>
Oliver Spatscheck <spatsch@research.att.com>
Michael StJohns <stjohns@cacheflow.com>
Stephen Thomas <stephen.thomas@transnexus.com>
Gary Tomlinson <gary.tomlinson@cacheflow.com>

We started with a summary of the requirements work to date.

Brad Cain said that the request-routing requirements had come together well
and would be sent out to the list shortly after the call, but needed some
polishing before it became an internet-draft. Others on the call urged
immediately moving to internet-draft, remembering that i-ds are drafts.
[Brad has now submitted the request-routing document as an internet-draft].

Stephen Thomas said that comment on the distribution requirements, sent out
the earliest of the three documents, had died down somewhat.  He also looked
for guidance from the group about whether these were intended to be
requirements for a protocol or a system. In the course of examining that
question, some proposed that the distribution requirements should be more of
an abstract description of the problem to be solved.  While the existing
document has been a useful spur to discussion, there was some concern that
it might already be describing aspects of implementation as opposed to
requirements.  Stephen indicated that he was more comfortable with protocol
design than requirements, and that his schedule would make it difficult to
revise the document before the end of next week.  Lisa Amini volunteered to
edit a more abstract version of the requirements document.

Jay Guha proposed that the accounting requirements should focus on
identifying the items of work that people want accounted, because otherwise
it is hard to avoid simply constructing a generic accounting framework
little different from other generic frameworks.  He noted that making
progress on identifying those accounting elements depends on getting
feedback from service providers about what they believe they want accounting
systems to capture and deliver to each other in a peered arrangement.  There
was general agreement that it would be valuable to make the accounting
requirements somewhat more specific.

We then discussed what documents should actually be in the charter. Mark Day
proposed that we remove the protocols as deliverables and focus on
completing requirements and the other documents in the proposed charter
quickly, then use that success as a basis for rechartering to develop
protocols.  There was general agreement that this would be a good strategy.
Two important reasons are the current style in which the IETF Area Directors
are organizing working groups, and our currently weak understanding of how
many protocols there should be or how they should relate to each other.

Then we took up the question of organization of editor calls in the future.
This was a little more controversial. All parties acknowledged that we were
in a gray area where there is little official IETF guidance and widely
varying practices in different groups.  There was concern about whether the
proposed "alternating week teleconferences" might be setting up closed
groups to do the "real design" and present the results to the working group
as "take it or leave it." We agreed that was a style to avoid. Others felt
that we were already proceeding in a more open way than some working groups,
with all substantive design discussions happening via email on the group
mailing list. There was a concern that we shouldn't handicap our ability to
make progress on the documents unless we were sure there was a problem.  We
agreed that the editors should each have the ability to organize their
authors and tasks as seemed most effective -- so any organizational
proposals from the chair should only be suggestions. However, we also agreed
that it would be helpful to participants in the working group if all editors
follow the same basic structure, as opposed to having wildly different
approaches.

Our final item of business was summarizing the state of documents and
editors. Subject to modification in working group discussion, these are the
documents that will be called out as deliverables in the charter and the
editors for each:

"Old" merged & updated documents to be finished:
Vocabulary -- Mark Day
Scenarios -- Phil Rzewski
System Architecture -- Gary Tomlinson
Known Request-Routing Mechanisms -- Mark Green

"New" documents:
Requirements for Request-Routing -- Brad Cain
Requirements for Distribution -- Lisa Amini
Requirements for Accounting -- Jay Guha

I expect to send a corrected and updated proposed charter to the list no
later than Monday.

Comments and corrections to this summary are welcome, as usual.

Thanks to all participants in the call for another very useful discussion,
and special thanks to those on the list who have been actively engaged in
moving the requirements discussions and documents forward over the last
couple of weeks.  I think we're on a good path and I look forward to our
further progress.

--Mark

Mark Stuart Day
Senior Scientist
Cisco Systems
+1 (781) 663-8310
markday@cisco.com