Informal meeting at 3TF of interested people on key initialization

[by way of "Chris M. Lonvick" <[EMAIL PROTECTED]>]

This afternoon Dan Nessett (DN), Bill Vroman (WV), Carl Madison (CM) and
myself (AB) met with Boby Joseph (BJ) from NSBU and James Bind (JB) from
PCBU on key initialization and management.  BJ is interested because he's
been investigating a similar scheme but based on public key technology;  JB
is interested because of similar work in NICs.

DN described the basics of the simple key initialization procedure, making
it clear that the embedded 128b key is never seen on the wire, and is used
only once, for authentication, not encryption, in a semi-automatic
administrative procedure that enforces a single use of the key in
initialization.   The procedure uses a simplified Diffie-Hellman key
exchange rather than full IKE procedures, to reduce memory footprint and
perhaps key exchange time, for small devices with limited CPU resources.

BJ described an alternative that uses PK certificates fully, assuming an
embedded PK certificate and secret key, and some 3Com PK infrastructure to
support recording and retention or possibly derivation of per-device
certificates and keys.  Among other complexities of the key management
problem, DN pointed out that the company assumes a huge legal liability for
the confidentiality of this key database in its CA.  Moreover the PK
approach uses more CPU resources than can be afforded in the smallest
devices.

JB described a similar embedded key in the Typhoon GE NIC which is intended
primarily to serve as a mechanism for enabling software features;  it's
called a "Crypto Enabling Feature".  Some thought has been given to other
uses including secure management initialization;  he is interested in
looking into this key management approach.

(BJ also described similar feature enabling crypto applications in the NSBU
products.)

DN promised more information in the draft 3FC;  no commitments on date due
to travel and conflicting activities.

(AB also reported briefly on progress on the IETF BOF on syslog security.)

Alex Brown <[EMAIL PROTECTED]> +1 508 323 2283