Re: 3gpp-analysis document and automatic tunneling

[Fred Templin <ftemplin@IPRG.nokia.com>]

Pekka Savola wrote:

> On Wed, 28 May 2003, Fred Templin wrote:
>  
> > > Pekka Savola wrote:
> > >
> > >      
> > > > ISATAP doesn't allow you you to connect more than a single host.
> > > >        
> > I think I know now why I was confused by your point on isatap not
> > allowing more than a single host. It is not strictly the case that nodes
> > on an isatap link must use isatap addresses as the source/destination
> > addresses in packets. If there is a way for the isatap router to discover
> > host routes, then  nodes on the isatap link can use , e.g., RFC 3041
> > privacy addresses.
> >
> > In other words, the only addresses that need to be in isatap format are
> > those that appear in the next-hop fields of a node's IPv6 routing table
> > entries. (In many cases, those next-hop addresses could simply be link-
> > local.) As such, a single isatap client can connect many additional hosts.
> >    
> Indeed -- but I think this is a rather challenging problem in operator <->
> user interface (where it has been proposed in this context), as routing
> protocol cannot be used.
That's OK. Even better than using a routing protocol is dhcpv6 prefix 
delegation,
with the isatap routers serving as delegating routers. This puts a 
single /64 into
the delegating router's forwarding table instead of a bunch of /128's.

> Was your point about ISATAP in general, or did you have some method for 
> host route discovery in mind for this particular case?
No - my point was that in fact isatap clients can support many hosts (up to
2**64 when prefix delegation is used) contrary to your assertion that only
a single host is supported.

Fred
ftemplin@iprg.nokia.com