Re: [tsvwg] CC/bleaching thoughts for draft-ietf-tsvwg-le-phb-04

[Toerless Eckert <[e-mail]>]

On Tue, Apr 10, 2018 at 09:17:59AM +0000, [e-mail] wrote:
> Hi Toerless,
> 
> you wrote: "How is an LE sender meant to indicate that the traffic is LE 
> if not via an end-to-end DSCP indicating LE ?"
> 
> There is no end-to-end DSCP indicating the same PHB unless all providers, 
> departments, platform owners (and so on) along the end-to-end Transport 
> chain agreed on the value. 

Sure. 
a) thats why we write these drafts/RFC to establish such DSCP standards.
b) If there is no end-to-end DSCP, there is also no other way to
   indicate end-to-end PHB.

> My take of your LE discussion was that the sending party doesn't care, whether 
> there's a Diffserv agreement in place with a receiving domain. In that case, 
> it is a no go to link any service or transport specification to the DSCP value. 

No, its not about parties "not caring" about DiffServ agreement. Its just
that any new diffserv semantic we want to bring into "the Internet" needs to
have a good story-line how it can be be backward compatible with BE in case there is no
end-to-end agreement, and with LE we can IMHO easily do this. I don't think
"good fallback" is the same as "don't care".

E.g.: for me the logic is quite simple: Applications indicate the end-to-end
desired PHB/DB via a DSCP, the path/domains try to figure out if/how that
desire can be met. If such an app is deployed on the Internet, it also
needs to have some fallback to BE - Circuit breakers or BE compatible CC.


Cheers
   Toerless
> 
> Regards,
> 
> Ruediger
> 
> -----Ursprüngliche Nachricht-----
> Von: Toerless Eckert [mailto:[e-mail]] 
> Gesendet: Dienstag, 10. April 2018 11:01
> An: Brian E Carpenter <[e-mail]>
> Cc: Geib, Rüdiger <[e-mail]>; [e-mail]; [e-mail]
> Betreff: Re: [tsvwg] CC/bleaching thoughts for draft-ietf-tsvwg-le-phb-04
> 
> On Tue, Apr 10, 2018 at 08:52:10AM +1200, Brian E Carpenter wrote:
> > On 09/04/2018 19:13, [e-mail] wrote:
> > ...
> > > My general view is: don't link determined end to end transport 
> > > behavior to DSCP values bearing no end to end significance.
> > 
> > +many. The fact that DSCP meanings are domain-specific is so 
> > +fundamental
> > that coupling them to e2e transport behaviour is condemned to failure.
> 
> I don't  know how to read Ruediger and your statements. How is an LE sender meant to indicate that the traffic is LE if not via an end-to-end DSCP indicating LE ? 
> 
> > (Which we spent a lot of time explaining to the WebRTC folk.)
> 
> draft-ietf-tsvwg-rtcweb-qos ? Too bad it doesn't have changelogs. WOuld love to know what was changed in it because of that feedback. No mentioning of interdomain so not sure how this draft is relevant here...
> 
> Cheers
>     Toerless
> 
> 
> Cheers
>     Toerless

-- 
---
[e-mail]