[EMAIL PROTECTED] wrote:
> ...
> Our intention is to write this specification as "minimum
> IPv6 functionality for a cellular host", i.e. not defining
> any kind of 'maximum set of functionality'.
> ...
> As a summary, a cellular host (often) has limited size,
> memory, processor capacity and battery capacity / power
> consumption. The users would like to see their terminals to
> be usable for quite a long time (typically for some days, at
> least stand-by time) without re-charging them.

You are focused on the engineering constraints of a particular device
(host) that is *only useful* in a cellular context as the definition. My
point was that there are many other devices that could take advantage of
the cellular link technology, and therefore MUST be considered 'cellular
hosts'. Statements in the document like: 'A cellular host SHOULD NOT
perform Duplicate Address Detection ...' or lack of support for IPsec
that is justified as too hard for a small device to implement are
focused on engineering issues for a particular product, not the
architectural minimum. If there are characteristics of the cellular link
that make the current standards inappropriate as written, we should
address those. If you are looking for concessions simply for marketing
reasons related to a particular target product, this document should die
right here.

The statement 'We can not list all possible services here, and in
general we expect application protocol RFCs to have requirements on what
security services they MUST employ.' is both an unrealistic expectation,
and the fundamental reason that we are being so insistent about the
requirement for IPsec. As a general rule, we don't expect (or have the
means to require) that every application define security requirements;
and even if they did evolving the security infrastructure to meet the
needs of every new application would be insane. On the other hand by
providing a secure channel for use by any application we can build an
infrastructure that scales. The argument that a cellular host is exempt
only ensures that the application developer can't count on that secure
channel being available, so each one has to do something specific to
address their needs. If you want to do as Keith suggested and split off
fixed-function as a special case, I won't object too loudly. Even there
you preclude something like a Java app, that comes in over the TLS
channel, from opening a non-TLS secure channel to get additional
content. Yes it is a fixed function web-browser, but do you really know
how every web site is designed, and can you ensure that all bits in and
out of the device can be protected by TLS?

Tony


--------------------------------------------------------------------
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