[tsvwg] David Black's WGLC review of draft-ietf-tsvwg-ecn-encap-guidelines-12

["Black, David" <[e-mail]>]

First of all, this is a well-written and comprehensive draft on ECN encapsulation design guidelines.

I found no major issues, and all of the minor issues are truly minor - some border on editorial,
and all can be dealt with via relatively minor edits.

--- Minor ---

[A] Need an explicit statement of exactly what the update to RFC 3819 is,
preferably in a section whose name uses "RFC 3819" and appears in the
table of contents.  Do this before some AD tells you to ;-).  This is
also the only thing that idnits found that needs attention.

[B] 1. Introduction, first paragraph, last sentence

The phrase "egress at any layer" is too general, hence incorrect (e.g.,
layer 7 was not intended to be included) - "lower layer" needs to be
used for clarity.

OLD
   If an egress at any layer is not ECN-aware, or if the
   ultimate receiver or sender is not ECN-aware, congestion needs to be
   indicated by dropping a packet, not marking it.
NEW
   If a lower layer header that may contain ECN congestion indications is
   removed by an egress that is not ECN-aware, or if the ultimate receiver or
   sender is not ECN-aware, then lower layer congestion needs to be indicated
   by dropping a packet, not marking it.
END

[C] 1.1 Scope

Second paragraph and the two-bullet list imply that RFC 4774 states that use
of ECT(1) to signal alternative ECN semantics is a best current practice.
That's not correct, as RFC 4774 covers only the first bullet on use of DSCP,
not the second bullet on use of ECT(1), as that second bullet is based solely
on RFC 8311. Some rewriting is in order to disconnect the second bullet from
RFC 4774.

Based on this, all other places where RFC 4774 is mentioned or cited should
be checked to determine whether RFC 8311 also ought to be mentioned or cited.

[D] Page 5, first complete paragraph

   Alternative semantics for the ECN field can be defined to depend on
   the traffic class indicated by the DSCP.  Therefore correct
   propagation of congestion signals could depend on correct propagation
   of the DSCP between the layers.

This also depends on correct propagation across diffserv boundaries - that
and the potential damage wrought by bleaching DSCPs to zero at such boundaries
both deserve mention here.

[E] 1.1 Scope, end of section

   In the feed-backward mode, propagation of
   congestion signals for multicast and anycast packets is out-of-scope
   (because it would be so complicated that it is hoped no-one would
   attempt such an abomination).

Much as I may agree with it :-), the parenthetical remark is inappropriate
for an archival RFC, and hence needs to be removed.

[F] 2. Terminology

   ECN-PDU:  A PDU that is part of a feedback loop within which all the
      nodes that need to propagate explicit congestion notifications
      back to the Load Regulator are ECN-capable.

That first sentence is too broad as it includes both the forward and
reverse directions of the feedback loop, whereas only the forward
direction is intended (FWIW, the definition of PDU does not suffice to
narrow this to the forward direction only).  For example, a TCP packet
with the ECE flag set in the TCP header and Not-ECT in the ECN field in
the IP header would be an ECN-PDU under this definition, which is not
what was intended.  The immediately following definition of Not-ECN-PDU
has the same problem.

The narrowing concept that needs to be added is that if the PDU carries
a congestion indication, then that congestion indication participates in
such a feedback loop, and even that is subtle when there are multiple
possible congestion indication locations.  An example of the subtlety is
that the same PDU may be an ECN-PDU in feed-up-and-forward mode because
the layer 3 and 4 protocols support ECN but may be a Not-ECN-PDU for
feed-forward-and-up mode because the L2 decapsulator ignores and discards
L2 congestion indications.

[G] 2. Terminology

   Congestion Baseline:  The location of the function on the path that
      initialised the values of all congestion notification fields in a
      sequence of packets, before any are set to the congestion
      experienced (CE) codepoint if they experience congestion further
      downstream.  Typically the original data source at layer-4.

That's counter-intuitive - a baseline is a level, not a location. I've
tagged this as a minor issue as opposed to editorial because it causes
severe confusion in the only place that this term is used, namely item 3
in Section 4.3.  That item 3 is about a Monitoring Baseline which is
a level not a location.

The Congestion Baseline term should be removed, and item 3 in Section
4.3 revised accordingly.  The term "Congestion Baseline" is used twice
there, but only the latter of those two uses needs to be retained and
expanded after deletion of this definition.

[H] 3. Modes of Operation

      Feed-Up-and-Forward:  A lower layer switch feeds-up congestion
         notification directly into the ECN field in the higher layer
         (e.g.  IP) header, irrespective of whether the node is at the
         egress of a subnet.

Remove "the ECN field in" as that's too restrictive, even though that's
a common case, and move it into the example - "(e.g., ECN field in the
IP header)" where "header" moved inside the parentheses.

[I] 3.3.  Feed-Backward Mode

Please make it clear that the critique of the ATM ABR relative rate
control mechanism also applies to Ethernet QCN, as that's a more
current example.  This should be connected to the QCN discussion at
the end of Section 4.2.

[J] 4.3 Encapsulation Guidelines

Resolving minor issue [G] requires some edits to item 3 in this section:
- Remove "(the Congestion Baseline)" from the third paragraph.
- Rewrite start of fourth paragraph:

OLD
       Most information can be extracted if the Congestion Baseline is
       standardised at the node that is regulating the load (the Load
       Regulator--typically the data source).
NEW
       More information can be extracted if the protocol specification
       requires that congestion fields be initialized to indicate no
       congestion only by the node that is regulating the load (the
       Load Regulator--typically the data source).
END

[K] 12.2 Informative References

Please check the references to 802.1Qah and 802.1Qau with IEEE.  The
correct references to use now may be the latest version of 802.1Q with
specific functionality and/or clauses identified.


--- Editorial/Nits ---

Abstract, last sentence

   Following these guidelines should assure
   interworking between new lower layer congestion notification
   mechanisms, whether specified by the IETF or other standards bodies.

between new -> among IP layer and

1. Introduction, first paragraph

Suggest removing use of the word "buffer" - "lower layer" suffices in
all of the places where this word is used in this paragraph.

1. Introduction, second paragraph

   The purpose of this document is to guide the addition of congestion
   notification to any subnet technology or tunnelling protocol, so that
   lower layer equipment can signal congestion explicitly and it will
   propagate consistently into encapsulated (higher layer) headers,
   otherwise the signals will not reach their ultimate destination.

Suggest: "equipment" -> "functionality"

Page 5, first full paragraph:

   If a particular protocol design chooses to contradict a
   'SHOULD (NOT)' given in the advice below, it MUST include a sound
   justification.

chooses to contradict -> chooses not to follow

1.1. Scope, first paragraph

   This document only concerns wire protocol processing of explicit
   notification of congestion.

Remove "wire"

Section 3.1, second paragraph

   In these cases, ECN may best be supported by standardising explicit
   notification of congestion into the lower layer protocol that carries
   the data forwards.  It will then also be necessary to define how the
   egress of the lower layer subnet propagates this explicit signal into
   the forwarded upper layer (IP) header.  It can then continue forwards
   until it finally reaches the destination transport (at L4).  Then
   typically the destination will feed this congestion notification back
   to the source transport using an end-to-end protocol (e.g.  TCP).
   This is the arrangement that has already been used to add ECN to IP-
   in-IP tunnels [RFC6040], IP-in-MPLS and MPLS-in-MPLS [RFC5129].

The referents of "It" at the start of the second and third sentences are
unclear and different.  Suggested rephrasings:

It will then also be necessary to define ->
     This necessitates also specifying

It can then continue -> This signal continues

Section 3.1, first paragraph after Figure 1

Cite one or more of the 3GPP references for GTP tunnels, as this is the
first occurrence of that concept in this draft.

Section 3.1, second paragraph after Figure 1

Cite a Frame Relay reference for the FECN bit.  Moving the [Buck00]
citation up to immediately follow "Frame Relay" in the first line
suffices.

Section 3.2 first paragraph

   These are Ethernet switches that bury into the Ethernet payload ...

bury -> burrow
[Spelling checker is missing "read user's mind" feature :-)]

Section 3.4

The second paragraph could be improved by citing references for fat-tree
and Clos network topologies.

Section 4.1, second paragraph

   Therefore section 4 of
   [I-D.ietf-tsvwg-rfc6040update-shim] requires network operators to
   configure the ingress of a non-ECN tunnel so that it zeros the ECN
   field in the outer IP header.

non-ECN tunnel -> tunnel that does not support ECN

Section 4.1, third paragraph

   Even
   if the shim(s) encapsulate a L2 header, it is often possible to find
   an inner IP header within the L2 header

Latter instance of "L2 header" should be "L2 PDU"

Section 4.1, last paragraph

   Instead a companion specification
   [I-D.ietf-tsvwg-rfc6040update-shim] has been prepared that has
   sufficient standards track status to update standards track
   protocols.

sufficient -> the appropriate

Authors Addresses

Please check these - Pat Thaler's contact info probably needs to be updated.

Thanks, --David
----------------------------------------------------------------
David L. Black, Senior Distinguished Engineer
Dell EMC, 176 South St., Hopkinton, MA  01748
[phone] New    Mobile: [phone]
[e-mail]<mailto:[e-mail]>
----------------------------------------------------------------