> > > This shows IPv4 thinking, where the network has a single 
> > > prefix / L2. 
> > > While I agree the initial deployments will likely mirror the IPv4 
> > > network, there is no reason to preclude having additional 
> > > prefixes / L2, where the reachability characteristics are different.
> > 
> > ...except that apps won't have any way of knowing which 
> > prefixes to use for which purposes.
> > 
> 
> Only due to the refusal to accept that hinden/haberman prefixes would
> provide a clue, and those used with the bellovin/zill RA provide a
> clear indication to any app that cares to look. 

hinden/haberman prefixes as currently defined do not provide a clue.
just because a PI prefix is in use does not mean that it's more
or less suitable for a particular purpose than a PA prefix.  in practice
the app needs several attributes: stability of the address-to-host
binding, routability of the address among the hosts that will need to
use it, etc.  depending on which nodes the app will need to talk to and
their locations in the network, a PI prefix might do better in some
cases and a PA prefix in others, and there's really no good way to tell.

I am still in catchup mode and haven't read the bellovin/zill RA
proposal yet, and actually am having trouble figuring out which I-D
people mean when they say this.  please send a pointer in private mail.

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