OK, I AM intrigued by this:
--- Keith Moore <moore(_at_)cs(_dot_)utk(_dot_)edu> wrote:
I don't think we're likely to be able to dispense with ICMP.
can find more ways to make it more trustworthy. Or maybe we
to discourage people from filtering it.
Any thoughts on the effectiveness of a BCP that says, message
type by message type and return code by return code, "filtering
this at a network edge does/does not make sense"?
We've been digging our toes in the sand on a Phil Karn proposal
to do TCP far-end link-up notification via retransmitted TCP
packets (for a few years). Architecturally, this is the kind of
thing ICMP is meant for, but I'm told repeatedly that new
ICMP-based mechanisms are non-starters on today's network...