Re: [tsvwg] L4S DSCP (was: L4S drafts: Next Steps)

[Jonathan Morton <[e-mail]>]

> On 31 Mar, 2021, at 1:07 am, Bob Briscoe <[e-mail]> wrote:
> 
> We can now summarize the impact of this proposal to use a DSCP to confine the L4S experiment to one network:
> * You've agreed that the DSCP scheme doesn't itself avoid any FIFO RFC3168 ECN AQMs in the network participating in the experiment, or in its customer networks upstream or downstream. It still requires a participating L4S network to follow the l4sops draft to bypass these AQMs, as if they were using the original L4S ECN scheme.

Correct, although you mischaracterise this as a disadvantage.  The point of "containment" is to divide the Internet into a small number of networks that follow l4s-ops and participate in the L4S experiment, versus a much larger number of networks which do not need to know or care about L4S.  This makes deploying your experiment easier, not harder, so IMHO you should be welcoming it.

> * You've also agreed that the DSCP scheme needs changes to all standards track transport protocols, so that they somehow feed back the DSCP as well as the ECN field, even though bits are scarce, and even though the only motivation for these changes would be to make this DSCP idea work for L4S.

No, I have not agreed anything of the sort.  I have instead pointed out that because both endpoints of an L4S connection need to be L4S-aware for it to *be* an L4S connection, building in a DSCP reflection to these endpoints is not a showstopper - even if it means that both ends must be using an L4S congestion control for either end to do so.

Also, this does not prevent using ECT(1) as an indicator that one end is using L4S congestion control while the other is not.  See, there is a workaround.

> * In addition, the DSCP will not necessarily even serve its confinement function if it is within a pipe-mode tunnel.

I assume you mean "a tunnel that hides the DSCP field of its inner packets".  I don't know how prevalent these are, but if they are a feature of the network rather than the endpoint, they should be considered as part of the network border.

> In summary, this DSCP proposal adds very little to safety and it subtracts greatly from viability of the experiment.

Okay, I get it, you don't like the DSCP proposals.  You don't need to argue and rant about every little detail of that.  But what you do need to answer is the simple question I posed earlier today:

> On 30 Mar, 2021, at 11:21 am, Bob Briscoe <[e-mail]> wrote:
> 
> L4S does not involve a DSCP, so there is no need to discuss DSCPs in l4sops.

Okay, so what mechanism do you propose to contain the L4S experiment to the participating networks?

 - Jonathan Morton