S Woodside wrote:
That may be true, but there are many people who are talking about the
internet being flat and reachable ... or not. And "not" includes both
packet filtering and scoping, even though they are different!
I don't know how to parse this.
My point was that there are topology locators that are only viable
within a scope defined by the local network manager. That network
manager may choose to implement the boundary of the local scope through
routing protocols or through access controls. It really doesn't matter
which technique is used, the result is the same. Local scope addresses
that are passed outside the defined boundary are useless.
The argument for passing addresses around is that doing so is faster
than passing the label used to get the initial mapping. While it is
probably faster in the case where the topology is consistent between the
sender & receiver, in the case where the topology doesn't match the
argument fails. How is passing a useless value faster than a label that
could be used to construct a useful one? If it is recognized that the
address failed, the only recovery is to try another member of the list,
or go back to the original label.
Tony