Tony,

> Tony Hain wrote:
> Maybe the way to solve this is to take the 'must be 0' bits and
> define them as 'locally administered'

That is what we were talking about: site IDs in SL addresses.

> with a clear note that  FE00::/8 will be blocked on the
> public net.

How can you guarantee this? If customers demand their ISPs to leak SL
addresses because customers see SL as a PI address they can't get
otherwise, ISPs will take the money at some point.

It is a terrible responsibility to embed everyone-gets-one-PI-address in
the addressing architecture. If the policy groups decide to give a PI
address to everyone, their call not us me thinks.


> This would allow sites that want the hassle of coordinating multiple
> private interconnects to do whatever they want (since they will
> anyway if the implementations allow it), without leaving any
> expectations that these are in any way globally unique.

The way I see it is: If they are not globally unique, it serves no
purpose. If they are, the fact that they will or will not be seen in the
global routing table is a matter of money, not of addressing
architecture.

So, this can't happen now because there is too big of a temptation. How
do we suppress the temptation? By providing people that want PI
addresses and multihomers with an _aggregatable_ solution.

Michel.


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