[ppsp] IETF 89 meeting notes

[yunfei zhang <[e-mail]>]

Hi all,
   The following is the PPSP session notes in IETF 89. Thanks Haibin and
Roni for taking the notes. Please let Ning and me know if you have any
updates.
BR
Yunfei and Ning
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

--------------------------------------------------------------PPSP WG
meeting, IETF 89, March 4, Tuesday (16:15 -17:18)

Chairs: Yunfei Zhang, Ning Zong

Attendees: about 20 people

Notes taken by: Haibin Song and Roni Even



 Agenda bashing (Chairs, 5 minutes)

 No comments.

Yunfei states the status update： Almost finish peer protocol, now focus on
tracker protocol.

Show hands: many people read the tracker protocol draft.

RTCWeb is an example of new use case for this WG.

Survey is going for AD review.

Individual documents for next step.

Martin(AD): resubmit the peer protocol write-up.

Yunfei: yes.

Yunfei calls for implementations and usage guidance document.



Rui Cruz presents tracker protocol,
draft-ietf-ppsp-base-tracker-protocol-03<http://tools.ietf.org/html?draft=draft-ietf-ppsp-base-tracker-protocol-03>
.



Protocol overview:

Request Messages, Connect

Request Messages, STAT_REPORT,

Request Messages, FIND,

Changes in -03:

Examples

Next Steps

Yunfei: A separate draft for implementation using XML is a good way. This
version separates messages from encodings.

Johan: Nice to see this will move forward. Efficient design is needed.

Currently the message is larger than the actual data.

Dave Cottlehuber: Too much information in the draft.Compare the draft of
peer protocol, refine the draft on whether it is context for understanding
or needing to implement. As little code as possible for tracker protocol
(due to management of a large number of peers), not like peer protocol.

Rui(response): optional info (like chunk info) is not in the base protocol,
in the appendix.

Dave: need only "must to have" in the draft, leave that "may have" items to
implementers.

Rui(response): The formal description will be simpler. First two bullets in
the next step slide. Hope the next version will make it clearer.



Rachel Huang presents survey of WebRTC which can potentially use PPSP,
draft-huang-ppsp-p2p-webrtc-survey-00.

Martin (as individual): You asked the wrong WG. Socialize with WebRTC WG.

Rachel: OK. And there is a draft.

Johan: An nice example for it is PeerJS. Look at it.

Rachel: Will look at it.

Dave: Interesting. WEBRTC is mostly peer to peer and not for multiple peers
connecting to one peer. Fear from HTTP.

Michael: There is no HTTP. It is pretty efficient.

Alexander: Multiple SCTP/TCP? connections need congestion control.

Michael Tuexen: ....

Dave: if I have 100 sessions, do I have to keep 100 connections?

Michael: does not have to, only one single when need to talk to you.

Johan: Something about "Multiple party support".

Alexander: How to implement this device to device talk? API or peer to peer
talk?

Ning: This is in the initial stage.

Martin: WebRTC has many things to do.

Roni Even: People use the data channel of WebRTC to do some applications.
Just focus on the transport. Other things are not the focus.

Dave: Can we get PPSP support in browsers as HTTP?



Finished.