Re: [EXTERNAL] Re: Limited Domains:

[Toerless Eckert <[e-mail]>]

On Fri, Apr 16, 2021 at 04:22:33PM -0700, Tom Herbert wrote:
> > Yes, and the scale of those large domains renders any proposal that relies
> on a perfectly synchronized configuration across all the elements to
> inevitably be problematic in deployment. Even in a limited domain, It is
> better to use explicit code points for new features to avoid any ambiguity
> and allow incremental deployment (i.e. no "flag days").

I think its more subtle than that. What you say is IMHO very true for
a non-tagged field in a packet of (often incorrectly) assumed global scope
that often may be used in non-well bounded networks, if not the Internet. 
Aka: for IP/IPv6 base header fields. Those conditions are IMHO what makes
 overloading a non-starter.  (and i am leaving out a couple more issues)

In MPLS the base functionality of every label (except SPL) is: NONE, so you
MUST have a control plane to go along with the forwarding plane for anything
to happen. And the scope of consistency of course differs across the
different semantics of label, and most times the IETF defined such a semantic,
the problems of consistency where analyzed/understood, and sometimes/often?
even documented in RFCs. In this model of separating encoding from
semantics, the core problems of IP/IPv6 for untagged "overloading" of
fields does not really exist. But of course, the issue of e.g.:
"eventual consistency" becomes larger and larger the more the domain where
consistent interpretation is required grows. So IMHO a protocol that would
allow to pick and choose tagging/registration based semantic vs
untagged/control-plane based semantic would really be cool.

Toerless

> Tom
> 
> > Thanks
> >
> > Gyan
> >
> > On Fri, Apr 16, 2021 at 6:12 PM Manfredi (US), Albert E <
> > [e-mail]> wrote:
> >
> >> From: ipv6 <[e-mail]> On Behalf Of Robert Raszuk
> >>
> >> > I think this this thread nicely demonstrates that we need to first
> >> define what a "limited domain" is.
> >> >
> >> > To some it seems to be 1980s definition of an IGP network boundary.
> >> More modern folks would consider as "limited domain" a set of IGP ASNs
> >> areas interconnected by p2p BGP still under the same administration.
> >> >
> >> > For me "limited domain" is an arbitrary collection of sites anywhere in
> >> the world using Internet for inter-connectivity.
 s
> >> Good point! Whereas to me, "limited domain" means, only inside this
> >> platform.
> >>
> >> > So any protocol which claims to be defined for "limited domain" and
> >> which claims that it is backwards compatible with nodes not supporting it
> >> is equal to allow it to traverse Internet.
> >>
> >> "Backwards compatible" may mean different things to different people, and
> >> it seem dubious in this case (because flow label is to be a random value,
> >> per IPv6). My main view is, if the domain is truly limited, firewalled or
> >> even air gapped, then what is the motivation to seek approval in a
> >> standards body?
> >>
> >> Bert
> >>
> >> --------------------------------------------------------------------
> >> IETF IPv6 working group mailing list
> >> [e-mail]
> >> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> >> --------------------------------------------------------------------
> >>
> > --
> >
> > <http://www.verizon.com/>
> >
> > *Gyan Mishra*
> >
> > *Network Solutions A**rchitect *
> >
> > *Email [e-mail] <[e-mail]>*
> >
> >
> >
> > *M 301 502-1347*
> >
> > --------------------------------------------------------------------
> > IETF IPv6 working group mailing list
> > [e-mail]
> > Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> > --------------------------------------------------------------------
> >

> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [e-mail]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------


-- 
---
[e-mail]