Hello Richard,

Reading your note, I thought of something that might be useful
in the discussion.

Richard Carlson wrote:

>                                                         That is, some
> addresses only work within a limited scope (link-local, site-local, and
> global).  I believe this limited scope view matches the reality, were
> network managers and institutions block certain connections.  I also
> believe this is a correct architectural position to take.

In this excerpt, I suggest replacing the concept of addresse "working"
by "are routable".  This does make sense, and moreover it allows a
possibly important refinement on the meaning of "addressability".

I would like to suggest that a host is "addressable" at a certain
address if a packet, with that particular destination address, would
be accepted by the host.

A host is "routable" at the address if routers know how to route
packets to some link where the host is able to receive it.

I hope the distinction is clear.  For instance, I am typically
addressable as Charles E. Perkins, but some messages addressed
with that as a destination address cannot be delivered to me,
depending on the transmission medium and location of the source
and so on.   Maybe the distinction has already been made, but
I don't remember seeing it lately.

If you like that distinction, then I think it means that site-local
addressability as a concept should be replaced by site-local
routability.  That puts it squarely in the domain of routing
configuration (and protocol), and takes it even farther away
from proper visibility by applications.

Anyway, I prefer globally unique but not-necessarily-routable
addresses to replace site-local addresses.  I don't see why
they shouldn't have prefix FEC0:/10, and I don't see why whatever
document deprecates site-locals should not also reserve that
prefix for the provider-independent uses that have been proposed.
The allocation mechanism could be discussed further, but
the addresses would be taken on the understanding that they
would not be globally routable.  If a particular recipient wishes
to make private routing arrangements, then that's their business.

Regards,
Charlie P.
--------------------------------------------------------------------
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