Transport protocols and bulk file transfer

[stanislav shalunov <>]

                     Bulk Transport Protocol Call
                      Friday, February 11, 2005

Attendees:
        Stanislav Shalunov 
        Steve Senger
        Larry Dunn
        Injong Rhee
        Susan Evett (scribe)

Call started at noon, 12:00 US/Eastern.

Agenda

1. Review of previous action items
2. Transport group web presence
3. Joint Techs next week (Steven and Stanislav are going)
4. Design Space Survey comments
        Requirements
        Conclusions
        Appendix (1/p)
5. Internet2 WG status


1. Review of Previous Action Items 

        a. Stanislav to develop a charter and side materials to
           request Internet2 working group status.  [In progress]

        b. Stanislav: incorporate Lisong's TCP Westwood description.
           [Done]

        c. Stanislav: address Bartek's comments about MaxNet and XCP.
           [Done]

        d. Stanislav: draft conclusions section for design space
           survey.  [Done]

        e. All: look at the requirements section of the document to
           ensure it captures the intent of the group.  Send comments
           to the mailing list.  (A simple ``I think this is what we
           want to build'' is OK, too.)  [Done by Larry and Stas.]

        f. Steven: contact the authors of the SCTP RFC to discuss
           models of partial reliability other than timed reliability.
           [Pending.]

        g. Steven: write up API specification.  [In progress.  ETA of
           -00 version: 2005-02-18]

        h. Stanislav to modify the draft based on discussion items
           with Guy Almes.  [Done]

2. Transport group web presence

Stanislav created a temporary web space for the group at
http://www.internet2.edu/~shalunov/transport/ -- the idea is to move
that material to an official location once WG status has been
conferred upon the group.  Stanislav asked if folks had had a chance
to look at the web space.  Steve said that it looked like a good start
to him.  If someone wants to send an introductory paragraph, please
feel free to send it to Stanislav.  Injong asked if the name of the
group, Interent2 Transport WG, would be the one used -- felt might be
too broad.  Injong wondered if *bulk* transport was the only goal --
need to clarify that bulk transfer includes real-time, and other
qualifications.  They noted that all the deliverables the group
envisions would fall under ``bulk transport.''  Larry commented on the
historical process of calling WGs after the general space vs. the
particular problem (such as with the Routing WG).  Decision was
deferred until the advisory councils have had input.

3. Joint Techs next week (Steven and Stanislav are going)

Presentation slot on Wednesday 02/16/05 - Larry will also be present.
Stanislav wanted to inform the community about the effort so that they
can have input/get involved before the process goes too far.  Larry
suggested leading off with a couple of case studies of problems that
need a new bulk transport protocol.

4. Design Space Survey comments

Review of the most recent draft of the paper. Larry asked about the
inclusion of MaxNet -- Stanislav noted that this is just another
explicit option, like XCP.  Larry felt that this section did not
include a recommendation from the group on this option's suitability.
Stanislav said he spoke with Bartek, drafted the text, ran it past
Bartek, Lachlan Andrew, and Steven Low, and included the revised text
into the document.  Discussion about whether this is opening a can of
worms -- Stanislav noted that this is for explicit congestion control
and there are not that many options available.

Comment that there is a different approach between sections 2.1 and
2.2: while 2.1 explicitly lists a number of options, 2.2 does not;
Stanislav argues that this is because if the group explicitly listed
options in 2.2 it would make the document unwieldy and would be out of
scope for the group's efforts.  Section 2.1 is reasonably inclusive;
MaxNet was relevant to add.

Injong asked about the layout of the table of contents: by reading
2.1's header, you expect to see a discussion about the differences
between the two approaches.  Instead you see only explicit in 2.1 and
only implicit in 2.2.

Injong wanted to move the introductory para from section 2.1 up to 2,
then add references to what 2.1 and 2.2 will address (with some
names).  In section 2.2, title needs to be changed to reflect header
in section 2.1 and introductory paragraph changes (abstract out some
of the shared approaches of the various programs).

Larry noted that, in terms of balance, section 2 is much longer than
the other sections -- the bulk of the report.  Stanislav noted that
section 2 is close to completion but the other sections are not
complete so we should check the balance as we get closer to
completion.  Stanislav suggested, for example, removing the section
number from ``Future Work,'' which is a small paragraph of text.

Stanislav asked for input on the conclusions -- Larry said that he has
heard some pushback on this re: User Space (might really cost a lot
and might only allow one application to be running at a time).
Stanislav felt that we could include a description of the downsides;
he will attempt to quantify it.  Larry felt that quantification would
be heroic, but probably not necessary.

Stanislav asked Injong re: his agreement to being a co-author to the
paper - Injong just emailed Stanislav about it; Stanislav noted that
he will be credited in the next version of the document.

Stanislav asked about the appendix -- Larry had some concerns with the
applicability.  His concern is with the text connections between the
formulas is somewhat questionable.  ``Converge'' is not defined, for
example.  Stanislav believed that the derivation was OK (and was
willing to hear about any errors), but the applicability might have
been stretched and asked for examples of such stretching, if they
exist.  Larry asked if the derivation for this paper is self-contained
or taken from earlier works; Stanislav said that, although, there is
another version of this work, this is self-contained.  Larry offered
to take a closer look at the section and be more concrete with his
concerns by the time he sees Stanislav on 2005-02-16.

Stanislav asked about the new title.  Larry was also concerned that
the title for the document might be misleading about the status of the
work -- that it was already complete or nearly complete.  Injong
suggested including ``space'' again in the title to indicate that the
group is open to options; title of paper changed to ``Design Space for
a Bulk Transport Tool.''

Steve and Larry noted that the paper was printing with extra-large
margins; Stanislav will look into why this is so.

5. Internet2 WG status

Stanislav was recently told of the list of documents that need to be
completed and submitted.  He will draft the charter and submit to the
group for review and approval before submitting it.

ACTION ITEMS:

a. Stanislav to develop a charter and side materials to request
   Internet2 working group status.  [ETA for charter: 2005-02-18.]

b. Steven: contact the authors of the SCTP RFC to discuss models of
   partial reliability other than timed reliability.  [ETA:
   2005-02-18.]

c. Steven: write up API specification.  [ETA for -00: 2005-02-18.]

d. Larry to more carefully review the Appendix with an aim to having
   concrete suggestions for changes to this section by 2005-02-16.

e. Stanislav to change the title of the document to ``Design Space for
   a Bulk Transport Tool.''

f. Stanislav to change the .pdf in a future version to remove the
   extra-large margins.  [Low priority for now]

Next call is at the same time on 2005-02-18.
Call ended at 13:00 US/Eastern.

A copy of these minutes can also be found at
http://www.internet2.edu/~shalunov/transport/20050211-minutes.txt

-- 
Stanislav Shalunov              http://www.internet2.edu/~shalunov/

"I didn't attend the funeral, but I sent a nice letter saying that I
approved of it."                                -- Mark Twain