Keith Moore wrote:
> It wasn't a few vocal people.  It was an overwhelming 
> majority.  There 
> were far more people in favor of deprecating site locals at that 
> meeting, than the set of people in favor of site locals even if you 
> include those on the list those who weren't present at that meeting.  
> It was a clear consensus of the working group.  And yes, you 
> can make a 
> decision in a face-to-face meeting if there is such overwhelming 
> support for the decision at the meeting that there simply 
> aren't enough 
> people on the mailing list to affect the consensus.

There are significantly more people on this list than you could fit in
one of the rooms in SF. The chairs need to raise the question on the
list, because this is not a trival issue. From reports I heard the whole
SF discussion was based on a bogus assertion that SL == NAT. 

> 
> > Site-locals are useful exactly for the case that started 
> this thread,
> 
> Site-locals are useful, but the cost is too high.  The additional 
> complexity that site locals add to nearly every part of the 
> Internet - 
> in apps, DNS, routing, and elsewhere - simply isn't worth the 
> benefit.  
> There are less expensive ways to provide the same functionality.

If they exist then list them. There is currently no other way to provide
disconnected sites address space. There is no other way to provide
stable internal use addresses for intermittently connected sites. There
is no other way to provide consistent consumer-simple route filtering
for local use appliances and applications. 

The cost of allowing niche applications to drive the overall
architecture is too high. The mainstream applications will be consumer
oriented plug-n-play, so the addressing architecture has to take that
into account. Yes that means some apps will have to go out of their way
to specify a global scope address, but that is not a high cost, and it
is limited to the app developer. See if you can figure out a way for
Joe-sixpack to restrict the use of some applications on a machine to
local use while others are global use, using only global addresses. For
the consumer space, the only reasonable way to make this work is to have
the restricted apps use a restricted address space that the ISPs know to
filter. 

Tony


> 
> Keith
> 


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