On Wed, 21 Nov 2001, Phil Karn wrote:
> This is *exactly* analogous to the discussion of how the end-to-end
> reliability provided by Internet transport protocols obviates the need
> for "total" reliability in a subnetwork.
The only comment gorry and I have had so far on ARQ being last-called
is that transport reliability does not provide end-to-end reliability;
it's still point-to-point, though networking types tend not to see it
that way.
Applications must still defend themselves against buggy transport
stacks and include their own message-checking. Users must defend
themselves against buggy applications and will e.g. ring to check an
email got through...
> Quite frankly, Dan, I'm being quite charitable here. The "minority"
> consists of one person: you.
A minority of one (and a half - I'm not fully convinced by either of
you, and I'm beginning to suspect IPSec is an overly subtle joke on
the rest of the planet, since it's too complex to get working. See
Schneier's critique of same.)
> Otherwise it would have been a clear
> consensus. And you didn't even show up in London to make your case;
damn unfair, Phil, unless you paid for your all of your London trip,
meeting fees, and accommodation yourself. Alas, I'm not able to fly to
Salt Lake City to make this particular point to you in person.
> I'm still digesting your substitute text. I may put it in just to get
> this over with.
Do that, and *I'll* go to the chairs and scream blue murder just as
Dan was threatening to.
> Nevertheless, I suspect you will never be satisfied with the result,
> so all I can suggest at this point is that you document your point of
> view in your own document and publish it as an informational RFC with
> your name on it.
That's a reasonable suggestion, and one I believe I've made to Dan
previously.
> [section on traffic analysis]
>
> >What does this add? Suggest removing this paragraph.
>
> You weren't in London, so you missed the discussion about the utility of
> subnetwork security to thwart traffic analysis. I suggested, and the group
> agreed, to add language addressing this point with the caveat that traffic
> analysis is a very difficult and subtle threat.
A difficult threat to counter, you mean. passive traffic
analysis is not difficult at all. You just sit there, picking the
stuff up, and you can spend the rest of eternity analysing it or
correlating it with other information to use it to your advantage.
L.
<L.Wood@surrey.ac.uk>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:29 EST