At the end of the day it doesn't matter what you call it, the fundamental
problem is that the widespread assumptions about a single-interface /
single-address are not realistic in today's Internet, and there are
independent policy realms influencing each end system. The available tool
set is built on a single-interface / single-policy-realm model, so it
routinely fails. Any effort that does not recognize that there needs to be a
way to describe the policy attributes is going to create a short-sighted
mess.
Keith is also right, in that aspects of the problem from the application
perspective still fail even when there are multiple choices from the same
administrative entity. That said, there should be a way to describe the
local policy attributes to sort between choices from a common
administration.
The bottom line is that we need to stop the fantasy that there are -any-
global attributes. This alone is the source of most of the confusion, and if
removed it would be fairly obvious to most people how to sort between the
choices. The IPv6 address selection effort is suffering from the same
'merged global attribute' mindset, and needs to be able to describe policy
attributes to each source of addresses, then a hierarchy for sorting through
them.
I do object to having the charter spend any more time on IPv4 than to simply
document current practice as a reference. It is long past time for the IETF
to stop wasting cycles on a dead end, and this should be one place where we
say 'it just doesn't work in IPv4, so move to IPv6 if you need multiple
interfaces'. Other than that I think it is fine.
Tony
-----Original Message-----
From: ietf-bounces(_at_)ietf(_dot_)org
[mailto:ietf-bounces(_at_)ietf(_dot_)org] On Behalf Of
Melinda Shore
Sent: Tuesday, April 21, 2009 3:09 PM
To: Dean Willis
Cc: IETF Discussion
Subject: Re: [mif] WG Review: Multiple InterFaces (mif)
Dean Willis wrote:
Consider that peering policy is often driven by things that are well
beyond the scope of protocol. Its potential range of expression is
unlimited; in fact driven by a natural-language contract and
heuristic
operations on underspecified constraints derived from that
natural-language contract.
Good heavens - I was not proposing, nor would I
propose, that what's needed here is the development
of a policy language. If the word "policy" is
making people uncomfortable perhaps it would be
better to drop it in favor of "properties."
Melinda
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf