[ppml] 240/4

vixie at vix.com wrote:

> we know from botnet hunting that hundreds of thousands of win9x PC's are never
> patched, and that in spite of their incredible infection levels, are still
> used every day by people who buy stuff.  what this means in practice is that
> no matter how easy the patch might be to produce and distribute, there will
> not be sufficient uptake that anyone large enough to need space from 240/4
> would ever(*) be willing to deploy it as "public unicast".  by "ever" i mean
> that the heat death of the universe would be more likely to occur first, or
> else, the wide scale usefulness of "public ipv6 unicast" would occur first.

I consider the requirement that anything using 240/4 would be patched up
to at least 2007 levels a significant feature :)

> so, 240/4 as an extension of RFC1918 space makes sense to me.  large ISP's
> can't do enough private addressing with the space RFC1918 reserves, and so,
> on the assumption that it'll be used for internal infrastructure where the
> deployer will be in better control of the software revision level, that's
> what i think ought to be done with it.

How much of the /4 would we need for rfc1918 type space?  Would a /6 be 
enough?
It might be wise to save the rest of it for future, as yet unimagined use.
I wonder if any of the various encapsulation approaches to id/locator 
split could
make use of some of that space.

- Kevin