[tcpm] Another requirement that L4S/TCPPrague depends on

Bob Briscoe <research@bobbriscoe.net> Fri, 22 July 2016 09:20 UTC

To: marcelo bagnulo braun <marcelo@it.uc3m.es>
From: Bob Briscoe <research@bobbriscoe.net>
Message-ID: <5791E566.7010201@bobbriscoe.net>
Date: Fri, 22 Jul 2016 10:20:38 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.8.0
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"; format="flowed"
Content-Transfer-Encoding: 7bit
Archived-At: <https://mailarchive.ietf.org/arch/msg/tcpm/-S4ctZMa_jxeJ91Xi25OruHlZRk>
Cc: tcpm IETF list <tcpm@ietf.org>, TCP Prague List <tcpPrague@ietf.org>
Subject: [tcpm] Another requirement that L4S/TCPPrague depends on
Precedence: list

Marcelo,

We need to add this I-D to the list of requirements that TCP Prague will 
have to depend on - in the L4S Problem Statement (assuming it could turn 
into an L4S architecture draft).

A Mechanism for ECN Path Probing and Fallback
https://tools.ietf.org/html/draft-kuehlewind-tcpm-ecn-fallback-01

Just asked one of the authors (Mirja). She says it has some wrong stuff 
in it, so it will need updating quite a bit anyway.
And it will need to be updated wrt L4S changes.

The probing and fall-back logic relevant to AccECN is already in 
draft-ietf-tcpm-accurate-ecn, which will have to be a prerequisite for 
L4S in TCP.

Nonetheless, this draft is broader than AccECN, and broader than L4S:
* It ought to cover any transport, not just TCP.
* And even in the case of TCP, it covers Classic ECN
* and even in the case of TCP Prague, it would cover the fall-back 
behaviour of the sending handler (at either end of a TCP connection), 
because that was ruled out of scope for AccECN.



Bob

-- 
________________________________________________________________
Bob Briscoe                               http://bobbriscoe.net/

[tcpm] Another requirement that L4S/TCPPrague dep… Bob Briscoe