Keith,

>> Michel Py wrote:
>> So what is the difference between this new GUPI block
>> and GUSLs, except that you want a possible evolution
>> to globally routable for the new GUPI block?

> Keith Moore wrote: 
> problems with GUSLs as an alternative to GUPIs:
> - the proposed uses of PI globals change too many
>   assumptions about how SLs were to be used to reuse
>   the FEC0::/10 prefix for that purpose.  

GUSL does not create PI globals. As its name implies, it creates
globally unique site-locals, which would avoid NAT and/or renumbering
when two sites connect. The scope would not be global, although it would
go behind the site's boundaries.


> - there were too many conflicting or intractable
>   assumptions about SLs anyway

Assumptions in current drafts?



> - we need PI globals that can be routed between sites
> by private agreement anyway.

We need _something_ that can be routed between sites by private
agreement only. This does not need to be global.


>> The obvious answer is that we create a new block _then_, not now.
> no, because we need the block now to solve the problems with SLs.

You are missing the point. If network administrators do not like
restricted SLs, they will use 2002:0A00::/24 instead, a one-way ticket
to NAT. Everybody would be better off having them use site-locals and
try to limit the damage.

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