Re: [tsvwg] Gorry Fairhurst Individual thoughts on choosing whether/how to advance ECN work.

[Jonathan Morton <[e-mail]>]

> On 15 May, 2020, at 10:50 pm, Joseph Touch <[e-mail]> wrote:
> 
> Is there a brief summary of what you wrote?
> 
> I.e., in 1-2 short sentences each:

> 	- assuming endpoints lie and/or network nodes lie (either by what they indicate or what they rewrite), are either of these options still safe?

I believe SCE is safe under the assumption of lying endpoints and/or middleboxes, to approximately the same extent that RFC-3168 ECN is.  Either the lie has a benign effect, or it ultimately harms the liar, for each class of possible lie that has been brought to my attention.

Conversely, it is straightforward to demonstrate a case in L4S where lying gains a significant advantage for the liar: a sender may mark its traffic ECT(0) but then implement DCTCP-style congestion control.  The demonstrated in-network component of L4S does not protect against this, and at least some existing networks are also vulnerable to it.

> 	- assuming everybody lies (as above), are either of these options still useful? (i.e., more than just safe)

Under the assumption that everybody lies, RFC-3168 ECN is itself pretty useless.  But it is used and useful in practice, so the underlying assumption that everyone lies should be re-examined.  The non-deployment and ultimate retirement of Nonce Sum lends evidence to this viewpoint.

Under the weaker assumption that everybody *can* lie but would only do so given some incentive, I believe SCE is not only safe but remains useful in most circumstances.

Under this same weaker assumption, L4S is neither safe nor useful in my opinion.

 - Jonathan Morton