RE: draft-ietf-multi6-multihoming-requirements-02

["Michel Py" <michel@arneill-py.sacramento.ca.us>]

Joe and multi6 folk,

It seems to me there is a circular reference issue here, two
requirements mutually incompatible that could lead to the impossibility
to provide a proposal that would meet all the requirements.

More specifically, I don't see how it would be possible to meet the
following requirement:

>> 3.1.3 Performance
>> A multi-homed site MUST be able to distribute inbound traffic from
>> particular multiple transit providers according to the particular
>> address range within their site which is sourcing or sinking the
>> traffic.

Without some form of cooperation (exchange of routes for the IPv4 model)
between directly connected transit providers ("direct peers") which is
the following.

>> |3.2.6 Cooperation between Transit Providers
>> |   A multihoming strategy MAY require cooperation between a site and
its
>> |   transit providers, but MUST NOT require cooperation directly
between
>> |   the transit providers.


In other words, the "performance" requirement, also referred to as
"global load sharing" (the capability to receive traffic from other
parties by the link closer to them, avoiding NAPs or interconnects) is
in IPv4 a function of routing. Regardless of the way it will be done in
IPv6, it is highly improbable that it could be achieved without
collaboration between direct peers, for the same reason it is highly
unlikely improbable without cooperation between a site and its transit
providers.


As I mentioned before, this is extremely restrictive and would prevent
the development of any new network-layer protocol. It could be argued
that exchanging routes between directly connected transit providers is a
form of cooperation.

I think both requirements are good, but that the cooperation between a
site and its transit provider must be extended to transit providers that
are direct peers as well.

I encourage every multi6 folk to comment on this,

Michel.

-----Original Message-----
From: Joe Abley [mailto:jabley@nlri.org] 
Sent: Monday, November 19, 2001 12:24 PM
To: multi6@ops.ietf.org
Subject: draft-ietf-multi6-multihoming-requirements-02

The following is the text of the candidate1 draft with minor
revisions relating to the small amount of discussion over the
past week. Main changes:

 + I included the cooperation requirement, since nobody seemed to
   object to it. I didn't include Michael's proposed changes to
   that clause, since there seemed to be some (small amount of)
   dissent about them. Further discussion on that paragraph would
   be useful, perhaps.

 + minor change of "internet" to "Internet".