Avri and all,
Perhaps a problem statement is necessary? Is the goal just to "make
things better"?
I don't work for an ISP. I don't configure routers. But in my mind the
current BGP-based routing system is too complicated to configure, and too
prone to being hosed by operator error. Maybe all I want is something
like DHCP for routing.
Then there's the multihoming issue. Wanting each site to have N-points of
connectivity to the Internet (with relatively small N, of course), seems
like a reasonable goal. Currently we rely strictly on aggregation, but it
collapses as large number of sites start to multihome (not to mention the
AS number issue). ;)
Perhaps a protocol that allows easier router self-configuration (even
partial self-configuration) and easy multihoming setup would be a good
start? (As in, I shouldn't have to call the tech support of both my
upstream providers and twist arms to get multihomed without using another
AS number.)
As for convergence times, real-time traffic, and so on, I don't know.
Does that need to be solved here? Is the IETF handling those issues?
-- Shane Kerr <shane@ripe.net> Database Software Engineer RIPE NCC +31 20 535 4427
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:04 EDT