Subject: Re: IPv6 Comment
To: None <>
From: Sean Doran <>
List: current-users
Date: 09/01/2000 20:50:26
thorpej writes:

| ...right, and IPv6 is a perfect solution for the problem.  It enables
| me to access, in a secure way, remote systems from, for example, my
| wireless network at home.

IPv6 is not a perfect solution for this problem, because 
you are not the only factor in the cost/benefit analysis of it.

What you are trying to do is engage a number of resources
in a distributed computation, the result of which is that 
data from your laptop arrives at the remote system of your

The "secure" part comes in two pieces.   One part has to
do with bits going to the right place, the other has to
do with the bits being the right ones upon arrival.

The former presupposes that it is possible to
secure the entirety of that distributed computation.
The problem is that the computation's boundary is often
very large (e.g., the entire Internet).

The latter can be done in any number of ways: ssh works well
in some cases, ESP mode works in some cases, hashes carried
along with unencrypted data works well sometimes, and so forth.
IPv6's security stuff is not the only way to skin this cat.

IPv6 is also far from perfect in skinning the first cat: you end
up adding costs into the network in comparison to IPv4 now, and
risk exacerbating some parts of the distributed computation that
are known to be NP-hard.

To make it very clear: the benefits and costs of transitioning
fully to IPv6 are anisotropic.   IPv6 deployment and operating costs
will always outweigh the new benefits, when examining the entirety
of the Internet.