Anonymity by mechanism ?

[Rich Salz]

The IETF "common authentication technology" working group (the folks
behind GSSAPI and, now, basically, all application-level security
protocols on the Internet) are looking at providing first-class support
for anonymity.  Here's a reent message.

cat-ietf-request at mit.edu to join.
	/r$
---------- Begin Forwarded Message ----------
Message 173:
>From owner-cat-ietf at cam.ov.com Mon Jan 30 07:15:20 1995
Date: Mon, 30 Jan 1995 12:35:24 --100
From: danisch at ira.uka.de (Hadmut Danisch)
Message-Id: <9501301135.AA29078 at elysion.iaks.ira.uka.de>
To: cat-ietf at mit.edu
Subject: Anonymity by mechanism ?
Content-Length: 1142
Status: R

Currently there is a discussion about providing anonymity. 

To avoid the problems of recognizing the name string syntax
for anonymous names and the necessity to implement the 
anonymous way into every gssapi mechanism I want to
propose the following:

- No anonymity within any gssapi mechanism except one
  special anonymous-only mechanism.

- Allow to use distinct authentication methods for
  both sides.

If an application accepts anonymous connections, it
uses the anonymous mechanism. If it accepts non-anonymous
connections also, it chooses both the anonymous and the
non-anonymous mechanism and asks for negotiation.

If only one peer wants to be anonymous, it can use the
anonymous mechanism and the other side a non-anonymous
mechanism (e.g. an ftp-server where the client wants to
be sure to be connected to the requested server).

This make programmers life easier and improves security:
A context is anonymous if and only if it was generated
by the anonymous mechanism. And the other mechanism don't have
to be extended for anonymity.

Anonymous mechanism doesn't mean no security. Just no
authentication of the peer.

Hadmut


----------- End Forwarded Message -----------