[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Consensus check



I believe that if we have any consensus, it is as Marcelo mention
focusing on solutions for an "edge site", i.e. a site whose routers do
not relay packets on behalf of other sites. As Noel noted, any short term
work is constrained by the "catenet/datagram" model of IPv6; we could
debate at length whether this is a bug or a feature. The limitations are
clear: we can use multi-homing for redondancy, but the amount of policy
they can exercize is by and large limited to picking the correct exit
router in their own site.
A common configuration in the (US, at least) academic community is for several universities to connect to an aggregator (GigaPoP). The GigaPoP has connectivity to one or more research networks, and, often, to one or more commodity providers. Many have IPv6 allocations from the Abilene PA block and are working on obtaining them from commodity providers. Thus the GigaPoP will be multi-homed, and will pass prefixes from each provider to the universities. For administrative reasons, selection of source and destination addresses will have to be done by the universities, not by the GigaPoP.

I think this scenario (from both the university and GigaPoP perspectives) should be considered by this WG.

Michael

+-----------------------------------------------------------------------+
| Michael H. Lambert, Network Engineer Phone: +1 412 268-4960 |
| Pittsburgh Supercomputing Center FAX: +1 412 268-8200 |
| 4400 Fifth Avenue, Pittsburgh, PA 15213 lambert@psc.edu |
+-----------------------------------------------------------------------+