[Previous] [Next] [Index] [Thread]

WWW Security and Standards Efforts

The following comments are based on my current understanding of
various security "standards" efforts underway that are either related
specifically to WWW applications or related to Internet client/server
applications in general.  The intent is to spur some discussion
regarding future interoperability among secure WWW applications.

It appears that there are multiple, parallel efforts underway
regarding various aspects of WWW security.  While each may be based on
RSA and DES technologies, their differing approaches to other aspects
of the security problem could impede interoperability.  In particular,
CERN, EIT, and the IETF are all in the process of specifying
approaches for one or more of the following: security mechanism
negotiation, data encapsulation, and key management.

Wrt security mechanism negotiation, EIT is espousing the S-HTTP
specification for negotiating authentication/encryption options.  The
IETF Common Authentication Technology (CAT) WG is also working on this
issue, under the guise of the GSSAPI standard.  The GSSAPI is intended
to be a generic application security API, to be used by a variety of
applications (not just the Web) and to support a variety of security
technologies (Kerberos, public key - RSA, DSA, etc.).  In addition,
the latest email from CERN indicates some notion of security
"negotiation" via URL syntax/pattern-matching.

Wrt to data encapsulation, I assume the EIT/RSA-based approach will
use the PKCS specifications regarding digital signatures and digital
envelopes.  Alternatively, there is an IETF CAT Internet Draft
specification of the Simply Public Key Mechanism (SPKM), which
describes similar concepts for a public key implementation of the
GSSAPI.

Wrt key management, there are issues regarding public key certificate
(PKC) formats and the validation of PKCs.  While many PKC formats are
based on X.509, some specifications include attribute extensions (e.g.
PKCS).  Then, there's the problem of how PKCs are validated.
Conceptually, a Certification Authority (CA) issues/signs PKCs for
users and services on the network.  However, even if a CA exists that
enough people trust, there's still an issue of how certificates are
revoked in the event of periodic update, compromise/theft, etc.  It's
unclear whether the certificate revocation list (CRL) approach is
sufficient for on-line services (e.g. WWW applications, versus privacy
enhanced mail).  In addition, users will want to use their
certificates to access a variety of secure applications, not just the
Web.

Now, for some informational comments regarding RSA, DSA, etc.  It's my
understanding that the RSAREF toolkit can be freely obtained in the US
for non-commercial use, whereas one must obtain a license from RSA for
one of their toolkit products for commercial use.  Also, I believe the
US government recently announced (or at least implied) that the
DSS/DSA is not in violation of RSA patents (at least from the
government's perspective).

Finally, a few questions:
* What/who is W3O?
* What "deal" with IETF "to gather all the standards making groups
  under W3O" was being referenced?


- Doug

Doug Rosenthal
MCC EINet                    |  Email: rosenthal@mcc.com
3500 W. Balcones Center Dr.  |  Voice: 512-338-3515
Austin, TX USA 78759         |  Fax:   512-338-3897
Follow-Ups: