[6bone] Request: two 6bone pTLAs

[Iljitsch van Beijnum]

On 12-mei-04, at 11:08, Pekka Savola wrote:

> I have to oppose this for the simple logic:
>  1) we want to kill 6bone as soon as possible

We want to kill it off 6/6/6. That's two years from now.

>  2) for these prefixes to be useful, they have to implemented

Just configured. This can happen in five minutes. No code required.

>  3) withdrawing implementations takes a long time.

This is something that's only going to be used by a very small number 
of people: those who run IPv6-only at least some of the time and roam 
between different IPv6 network, OR people who specificially want to 
participate in the experiment. When the WKAs are removed, these people 
will no longer be able to resolve names so they'll reconfigure their 
systems fairly quickly. Obviously there will be some stray traffic for 
a long time, but this will be fairly insignificant in the grand scheme 
of things.

Note that the value of these WKAs is that they are available 
everywhere, so as soon as the 6bone becomes unroutable in many parts of 
the network people will move away from the WKAs.

> So, giving out 6bone pTLAs would probably just lead to continuation of
> the 6bone mess.

> There is already a better process in place for something like this --
> for example the experimental allocations:
> http://www.ripe.net/ripe/docs/ipv6policy.html#experiment-assignments

6bone "mess" will have to be cleaned up (or left to be cooled off for a 
long time) anyway, this won't add a whole lot to this. On the other 
hand, doing this in RIR space creates a new mess that will probably be 
permanent. So using RIR space for just this experiment would be a 
mistake.