Thanks Margaret for opening up discussions on each of
the specific points you raised. I think we do need
the discussion.

> Now, I don't actually live under a rock, so I do understand that most
> of today's IPv6 nodes don't actually implement IP Security.  In the
> past, however, the IESG had mandated that IP Security would be a
> mandatory part of IPv6, and I don't believe that they've changed that
> statement.

First I'd just like to note that the main approach
to security is outlined in the beginning of section 3.
Section 2.4 text is just a consequence of that.

Now, I actually believe there's today a better understanding
of what security is required when, and what are the
true capabilities of various security mechanisms than back
when some of the base IPv6 RFCs were created.

What we describe in section 3 tries to reflect some of
that new understanding. In particular, we realize that
there are a few different security mechanisms, and often
IETF application protocol RFCs specify a particular security
mechanism to be used. This isn't always IPsec. In other
cases a single mechanism is simply the most appropriate
one. As an example of this, consider the use of TLS vs.
IPsec for web access; I think the appropriate security
mechanism is quite clear in this case. 

Secondly, we now have a better understanding of what
IPsec can do in the area of protecting IPv6 control
signaling -- it can do things, but not perhaps quite
as much as was originally believed [1, 2, 3]. IPsec
is however, obviously the used in e.g. corporate intranet
access and VPN solutions.

Therefore we feel that on size- and power-constrained
devices one should select the implemented security
mechanisms based on the requirements of the applications
rather than a general rule. The general rules aren't
very good if in the end the application RFC you were running
would demand something else than the general rule, or if
the deployment at the other end was something else.

This conclusion seems to be supported by other approaches
to host requirements [4].

In any case, I'd be very happy to get feedback on
exactly what we should write as the host requirements
for security. I'm reasonably happy with what we have
written -- in particular I believe this is roughly
where the world seems to be heading -- but if you
have other ideas or improvement suggestions, please
present them!

Jari
----
[1] http://search.ietf.org/internet-drafts/draft-kempf-ipng-netaccess-threats-00.txt
[2] http://www.arkko.com/publications/draft-arkko-manual-icmpv6-sas-00.txt
[3] http://www.ietf.org/internet-drafts/draft-arkko-mipv6-bu-security-01.txt
[4] http://search.ietf.org/internet-drafts/draft-okabe-ipv6-lcna-minreq-01.txt
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to