Re: [tsvwg] Comments on draft-ietf-tsvwg-ecn-encap-guidelines-13

[Bob Briscoe <[e-mail]>]

David,

On 10/03/2020 20:14, Black, David wrote:
>
> Writing as draft shepherd …
>
> > Section 4.4, point 1, first starred subpoint, there is something odd 
> about
>
> >     "the packet MAY be forwarded, but it SHOULD be dropped".
>
> > Any better (I've added some of the context for the list)?:
>
> >          If the congestion marking is the
> >          most severe possible, the packet MUST be dropped.  However, if
> >          congestion can be marked with multiple levels of severity and
> >          the packet's marking is not the most severe, this requirement
> >          can be relaxed to: the packet SHOULD be dropped, but it MAY be
> >          forwarded.
>
> That’s not much of an improvement.  Here’s the full -13 text to 
> provide the context:
>
> 1.  If the arriving inner header is a Not-ECN-PDU it implies the L4
>
>        transport will not understand explicit congestion markings.
>
>        Then:
>
>        *  If the outer header carries an explicit congestion marking,
>
>           drop is the only indication of congestion that the L4
>
>           transport will understand.  If the congestion marking is the
>
>           most severe possible, the packet MUST be dropped.  However, if
>
>           congestion can be marked with multiple levels severity and the
>
>           packet's marking is not the most severe, the packet MAY be
>
>           forwarded, but it SHOULD be dropped.
>
> Assuming that we don’t tinker with the technical content of the last 
> sentence, the better thing to
>
> do with a statement of a “SHOULD” requirement is to explain what may 
> happen if something else
>
> is done.   I think it’s implicit that a packet that is not dropped 
> gets forwarded instead, so here’s
>
> a suggestion for better text:
>
>            If the congestion marking is the
>            most severe possible, the packet MUST be dropped.  However, if
>            congestion can be marked with multiple levels of severity and
>            the packet's marking is not the most severe, this requirement
>            can be relaxed to: the packet SHOULD be dropped, as forwarding
>            all such packets may conceal the network congestion that they
>            from the L4 transport that does not understand explicit congestion
>            markings.
I see your point now (and I now see Donald's).
Nonetheless, I'd rather just end the sentence at "...SHOULD be dropped".

The possibility the L4 transport won't understand is given as the 
opening reason for the whole paragraph anyway. Repeating it here, makes 
it look like it's the reason for demoting this low severity case from 
MUST to SHOULD, which is not the intention.

Think SCE - it might not matter whether the L4 transport understands the 
lower severity marking or not.
(I had similar schemes in mind when I added in this text about severity 
levels in 2016).



Bob



> Thanks, --David
>
> *From:*tsvwg <[e-mail]> *On Behalf Of *Bob Briscoe
> *Sent:* Tuesday, March 10, 2020 1:29 PM
> *To:* Donald Eastlake; [e-mail]
> *Cc:* John Kaippallimalil
> *Subject:* Re: [tsvwg] Comments on 
> draft-ietf-tsvwg-ecn-encap-guidelines-13
>
> [EXTERNAL EMAIL]
>
> Donald,
>
> Thank you for taking the time to review this (rather long) draft.
> Apologies for not getting to your review until now.
>
> On 06/02/2020 23:03, Donald Eastlake wrote:
>
>     Hi,
>
>
>     I'm not subscribed to the tsvwg mailing list but I have reviewed
>     draft-ietf-tsvwg-ecn-encap-guidelines-13 and though you might be
>     interested in my comments.
>
>     Overall, this is a very clear and well-written draft. The comments
>     below are minor. Whether or not they are incorporated into the
>     draft, I hope that it can be advanced soon.
>
>     Section 1. I suggest just deleting the one occurrence in the draft
>     of "[RFC1323]" and the corresponding reference section entry. It
>     seems unnecessary and just leads to a nits checker warning which
>     will have to be explained, etc.
>
>
>     Section 1.1. Very minor but I believe the usual way, inside a
>     draft, to refer to the RFC which that draft might become is "[this
>     document]" (without the double quotes) rather than "[RFCXXXX]".
>     Changing to the more common notation would, I believe, enable the
>     RFC Editor note to be removed as "[this document]" is well
>     understood by the RFC Editor.
>
>
>     Section 2. The initial paragraph on implementation keywords should
>     be updated to the following as per RFC 8174:
>
>         The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL
>         NOT",
>
>         "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL"
>         in this
>
>         document are to be interpreted as described in [RFC2119] [RFC8174]
>
>         when, and only when, they appear in all capitals, as shown here.
>
> Done all the above
> (BTW, I always baulk at having to cite RFC8174, when the following 12 
> words succinctly state the sum total of its content.)
>
>
>
>     Section 2. Suggest putting the Terminology entries in alphabetic
>     order..
>
> I haven't done this. They are more for reading through than for being 
> looked up individually, most of them fall into logical little groups, 
> and there are not so many that it's hard to find one.
>
>
>
>     Section 4.2, page 18. "802.1p" has been merged into 802.1Q ages
>     ago. Values of the priority field are commonly referred to in IEEE
>     802.1 as Priority Code Points (PCPs) and in any case this seems a
>     bit inconsistent to the way that the merger of 802.1ah into 802.1Q
>     is recognized in the draft. Perhaps the last sentence of Section
>     4.2 could be: "An operator can define certain [IEEE802.1Q]
>     Priority Code Points to indicate non-QCN frames and an ingress
>     bridge is required to map arriving not-QCN-capable IP packets to
>     one of these code points."
>
> OK. I've taken on board the spirit of your edit, but changed it slightly:
>
>     An operator can define certain
>     Priority Code Points (PCPs [IEEE802.1Q]; previously 802.1p) to
>     indicate non-QCN frames and an ingress bridge is required to map
>     arriving not-QCN-capable IP packets to one of these non-QCN PCPs.
>
> This is then consistent with the other references to 802.1Q, which 
> also give the number of the constituent part before it was wrapped up 
> into the mega-standard. If you think this is clumsy, pls say. I did it 
> this way, because many people know these 802.1 drafts much better by 
> their old name (well, for 'many people' read 'me', or perhaps read it 
> as 'old farts like me').
>
>
>     Section 4.4, point 1, first starred subpoint, there is something
>     odd about "the packet MAY be forwarded, but it SHOULD be dropped".
>
> Any better (I've added some of the context for the list)?:
>
>            If the congestion marking is the
>            most severe possible, the packet MUST be dropped.  However, if
>            congestion can be marked with multiple levels of severity and
>            the packet's marking is not the most severe, this requirement
>            can be relaxed to: the packet SHOULD be dropped, but it MAY be
>            forwarded.
>
>     Section 7. It doesn't matter much but IANA would prefer that
>     sections saying there are no IANA actions be left in the final RFC
>     (see Section 9.1 of RFC 8126).
>
> I'm learning something new every day.
>
>
>     Section 9. Should "the document" in the first line of this section
>     by "this document"?
>
> Yes. Done.
>
>     Appendix A. I did not review this update history.
>
>     Authors' Addresses: I don't think Pat Thaler can be listed as a
>     front page "author" in the RFC sense unless at least an email
>     address is listed for her. All authors should be pollable about
>     IPR they know and when the draft gets to the AUTH48 state before
>     RFC publication, the RFC editor must be able to contact all the
>     authors. If no email address is known, she should be moved to a
>     "Contributors" section or the like.
>
> Yes. I discovered that (too late) last night, when the draft got 
> rejected on this point!
> I've added a Contributors section for her.
>
> Thank you again.
>
>
>
> Bob
>
>
>
>
>
> -- 
> ________________________________________________________________
> Bob Briscoehttp://bobbriscoe.net/

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