Re: [tsvwg] L4S: Guard DSCP

[Gorry Fairhurst <[e-mail]>]

[As an individual]

On 30/04/2021 04:25, C. M. Heard wrote:
> On Thu, Apr 29, 2021 at 10:46 AM Greg White <[e-mail] 
> <mailto:[e-mail]>> wrote:
>
>     In my opinion this proposal (if mandated) would significantly
>     diminish the L4S experiment.  There is widespread evidence of DSCP
>     bleaching in the Internet,
>
>
> Understood; that comment was anticipated by the reference at the end.
>
>     while there is little evidence of RFC3168 bottleneck deployment.
>
>
> Is that actually true? I got exactly the opposite message from Pete 
> Heist's data.

I haven't seen much evidence that pathology of using RFC3168 is common 
(i.e. the behaviour from 2001), if others have better data that has not 
been presented, I would be interested in seeing this.

>
>     I recall from the lengthy discussion that went on about one year
>     ago (see, e.g.,
>     https://mailarchive.ietf.org/arch/msg/tsvwg/ij5BqzNtMxKV8VbHE8pUevZjx7k/
>     [mailarchive.ietf.org]
>     <https://urldefense.com/v3/__https:/mailarchive.ietf.org/arch/msg/tsvwg/ij5BqzNtMxKV8VbHE8pUevZjx7k/__;!!LpKI!3O_yc-c_keXnDzQr5kIsAGHjsdLwrDdwRmywXodJmxuAEvzsG8GA0pNttYbDFl5f$>)
>     that one of the big objections to the use of a Guard DSCP as a
>     means to contain the experiment is that an L4S experimental domain
>     does not map well to a DSCP domain, and so the Guard DSCP may be
>     inadvertently bleached (e.g. by tunnels) inside the participating
>     network. If correct, that objection is hard to address.
>
>
> Mike Heard
>
I'd personally see a Guard DSCP as an optional mechanism that someone 
can deploy for an operator to contain an experiment. As I recall, the 
original use in RFC 4774, was to allow such a 3-level marking experiment.

Gorry (as an individual)