Re: multihoming without BGP

[Vadim Antonov <avg () pluris com>]

Paul A Vixie <paul () vix com> wrote:

> Shorter AS paths are a silly way to choose a path for a connection.  *IF*
> BGP carried all the end-to-end bandwidth and delay stuff that EIGRP does,
> it might be possible to make this decision intelligently.  But only if all
> nets described by a routing element were internally homogeneous -- that is,
> only one exit gateway rather than different exit gateways in each region.

Unfortunately, EIGRP-style "physical" metrics are patently useless in
the real networks where traffic is engineered.

Also, there are theoretical results in multicommodity flow routing
basically showing that "physical" metrics do not do much good; some
exponential function from "renormalized" load is needed to achieve
useful load distribution.  Unfortunately, nobody knows how well it works
together with cooperative congestion control schemes.

> None of these requirements hold in the case of BGP.  While BGP is a fine
> way to route packets, it's a horrid way to select paths for connections.

BGP lacks real metrics.  I'm still waiting for people to understand
that.

The protocol is not important.  What's important is how it compares
different paths.

--vadim