Re: security policy discovery

["The Purple Streak, Hilarie Orman" <ho@alum.mit.edu>]

At one time there was a vision of the future in which the majority of
network traffic would have cryptographic protection, so the default
behavior would be to negotiate IPSec SA's, falling back to plaintext,
perhaps, if there was no solution.

IPSP has been trying to address security policy discovery, but not as
an automatic side effect of a cleartext communication attempt.  That's
because in the Alice-Bob case, Bob might never see Alice's unprotected
packet - it would be caught and discarded by earlier protection
devices in Bob's network.  IPSP tries to address the problem of
discovering the security gateways trying to find a negotiable set of
SA's that will cover the Alice/Bob traffic, but only in the case that
Alice or Bob has already made a decison to setup IPSec.

If in response to a plaintext packet something emanating from the
Bobside declares "Alice-Bob requires IPSec", then Alice can take that
as a signal initiate some key exchange/IPsec negotiation protcol.
But, given that Alice has expressed a willingness to communicate
without crypto protection, one might suspect her committment to
confidentiality to be half-hearted.

Hilarie