One advantage of having scoped addresses defined in the IPv6
architecture from the start is that applications can know not to pass
them outside of their scope.  If we instead suggest that people
firewall/filter off random portions of the global address space, then
apps will blindly pass those addresses around in the data stream,
mistakenly thinking that they are real global addresses.  Only having
dedicated scoped address space allows apps to do the right thing.
I don't agree...

When one sections off a portion of the global address space and
an application can't reach it, that is the intended behaviour.
The application has an unambiguous address, but it just can't
reach the addressed node.  This will be a frequent occurence
(due to firewalls, network outages, etc.) and applications will
have to deal with it appropriately (error message to the user,
or whatever).

I'm not actually against the idea of private networks, and I
agree that they are necessary for security.  I just think that
they should use global addresses, not site-local addresses.

Site-local addresses create _overlapping_ private address
spaces.  When SLs are sent outside of a site, they become
ambiguous (or just plain wrong).  So, it is not possible to
know which host a corresponds to a particular site-local
address unless you also know which site the address applies
to.  This creates a set of problems that I think we should
avoid by not using site-local on globally-connected networks.

Margaret


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