Re: [tsvwg] Call for comments on the suggested publication timeline for draft-white-tsvwg-l4sops

[Jonathan Morton <[e-mail]>]

> On 22 Mar, 2022, at 11:39 am, Sebastian Moeller <[e-mail]> wrote:
> 
> So the one big issue I see is that the whole length section
> 6.  Operator of a Network Employing RFC3168 FIFO Bottlenecks  . .  13
> is aimed at an audience which is in no way guaranteed to even notice this draft at all.

> All networks employing RFC3168 (FIFO) Bottlenecks will see unexpected behavior once L4S-type ECN response traffic traverses their network and one of the rfc3168 nodes becomes the bottleneck. Those networks that are also actively deploying L4S (either for edge testing or in the core) will likely notice all L4S related drafts/RFCs, but networks not participating in the L4S experiment are IMHO unlikely to do so.

Yes, this is symptomatic of the larger problem with L4S as a whole.

> P.S.: My preferred solution to all of this is not to ratify the main L4S drafts any time soon since I consider L4S not ready for internet-wide testing* which makes the decision about the OPs draft also less urgent.

This would also be my preferred solution.

However, as I understand it, this is now the decision of the IESG to make, not the WG.  It is crucial that the IESG has all the information necessary to make the appropriate decision.  Some of that is in the drafts themselves, some in the shepherd's writeup, and some in the l4s-ops draft.  But the IESG does not look at drafts.

My considered opinion is that the L4S drafts cannot safely be used without the precautions that l4s-ops is intended to detail.  (The adequacy of these precautions is a matter for separate discussion.)  Therefore, publication of the L4S drafts - and consideration by the IESG - should not proceed until l4s-ops is passed by the WG as ready for publication alongside them, as a complete unit.

 - Jonathan Morton