Re: [tsvwg] draft diffserv-intercon: Handling of a scavenger class / CS1

[<[e-mail]>]

Mikael,

it took a while to make progress on DiffServ Intercon. You say 

     So yes, let's meet up in Prague and discuss this. I think what we 
     need is a different document than diffserv-intercon because most 
     in there just isn't applicable to Internet peering interconnects.

Do you say:

- there's no need for interconnection based on QoS classes - just forward 
  limited DSCP information to an access and that's fine? Or is your 
  statement no DSCP or IP-Precedence based QoS differentiation in the IP 
  backbone (and only default forwarding there), limit QoS to the access?
- you'd like to support QoS interconnection, but set up individual DSCP 
  based PHB mapping tables per interconnection? In that case, no new 
  standard is required. 
- You just need bleaching to Default - that's a separate issue which has been 
  removed from DiffServ-Intercon as part of WG consent. It is a pre-requisite 
  of DiffServ-Intercon, but not an issue handled by it.
- You look for a rewrite of ranges of DSCPs ("treatment of IP precedence")
  - this is out of scope, as it is a standards track issue touching the
  fundaments of DiffServ. DiffServ Intercon also works well if IP-Precedence 
  rewrite occurs. 

General claims like "isn't applicable to Internet-peering" aren't helpful, if you don't tell us what is applicable to internet peering. DiffServ-Intercon has been applicable to the QoS interconnection agreements I've been involved in so far. 

If you are able to provide something which isn't TL;DR, I'd appreciate your submission before cut-off. I'd also appreciate your indication what needs to be removed from draft diffserv-intercon to increase op-folk attention.

I repeat that I've talked with my RIPE/NANOG colleagues to get feedback - they came back only with one or two names with whom to discuss QoS interconnection. And I never got any useful response (positive or negative) from those.

Regards,

Ruediger