[6bone] Re: routing concern

[Robert Kiessling]

John Fraizer <tvo@EnterZone.Net> writes:

> Rob from Sprint hit on something that I don't quite agree with.  He wants
> to limit transit.  Coming from Sprint (no offense Rob), that makes perfect
> sense.  I live on the other end of the spectrum.  The more transit
> possibilities you have to a network you don't DIRECTLY peer with, the more
> likely you will be able to reach that network.

Actually this is not true, and limiting transit does make much sense,
for purely technical reasons. More transit in the sense of BGP
advertisements does NOT automatically give you better connectivity.

In practice, we saw a number of cases where someone provides IPv6
transit (e.g. advertises full BGP table to everyone), but in fact
fails to route packets because of some misconfiguration. Such this
"transit" caused a blackhole for you.

On a performance issue, a transit provider with lots of tunnels across
the world will give you short BGP paths, but not good actual paths for
the packets.

The lesson learned from this should be: look carefully at your transit
providers. For good connectivity, your transit provider should be
dedicated to providing this service (as opposed to just running it
experimentally and not caring), have cluefull staff for IPv6, and
should care to have good connectivity itself (as opposed to collecting
as many tunnels as possible).

A handful of providers fulfulling this will give you much better
connectivity than dozends of full BGP peerings.

Robert