RE: requirements draft revision

["Tony Hain" <alh-ietf@tndh.net>]

Joe Abley wrote:

> I have been trying to make the definitions as solution-neutral
> as possible.

This may not be possible because all the terms are already so overloaded in
different contexts. :)

> The word "enterprise" is only really being used to describe an
> entity which is able to multi-home.

So the guy next door with ADSL and Cable modem is an enterprise?

I think the problem is assuming that there is any association between the
size of an attaching entity and the routing policies it will try to inflict.
Even if current operational experience shows a correlation between size and
a particular activity; five years from now the words used to describe the
activity will be the same but the size will be smaller, so the entity being
described by those words will be different.

> On the other hand, if you have a better definition, let's
> hear it.

I have been trying to understand what the intent of the distinctions is. If
the scope is limited to the example of an entity attaching to two (or more)
providers and having the traffic flow over those attachment points without
traversing another provider, then your current evolved definition is
sufficient. If the intent is to eventually describe a variety of routing
policies expressions beyond the attached providers of the entity, then
directionality becomes a problem. B might want to punch a policy through C
to E to influence a mutual customer of D & E, even though the 'global
Internet' is upstream through A. Does this make C a 'transit provider' for B
or E, or a 'transit network', or all of the above? If the definition
requires directionality toward A then does it become none of the above? If C
acquires its address space from B does this change anything?

Tony