[2956] in cryptography@c2.net mail archive
Re: Cisco, NAI propose new key recovery
daemon@ATHENA.MIT.EDU (Phil Karn)
Tue Jul 14 13:58:30 1998
Date: Tue, 14 Jul 1998 10:19:45 -0700 (PDT)
From: Phil Karn <karn@qualcomm.com>
To: frantz@netcom.com
CC: cryptography@c2.net, karn@qualcomm.com
In-reply-to: <v03110716b1d09fd40354@[209.109.235.248]> (message from Bill
Frantz on Mon, 13 Jul 1998 21:43:57 -0800)
>This proposal doesn't protect against packet sniffing by people on the same
>LAN as the sender or recipient. It also doesn't protect against telephone
>taps for dial up ISPs or packet sniffing on cable TV IP networks. As such,
>it doesn't even begin to solve the confidentiality problem for a large
>class of network users.
This is an argument for end-to-end encryption, not an argument against
what Cisco et al are proposing. As I said yesterday, we've always known
that security is best done end-to-end.
IPSEC can be run in either end-to-end mode or in tunnel mode. Tunnel
mode is mainly intended as an expedient way to build virtual private
networks over the Internet and through firewalls in a way that keeps
most corporate network security folks happy. It is not a replacement
for true end-to-end security that is what most end users should want.
If Cisco's proposal results in the loosening of export regs on routers
capable of tunnel-mode IPSEC, without any deployment of actual key
escrow, then it will have been useful. The more tunnel-mode IPSEC out
there, the more end-to-end IPSEC will likely also appear -- whether or
not it is export controlled.
Phil
