ipv6 addressing - non-routable equivalents?

[Peter Håkanson]

On Sun, 8 Oct 2000, Jonathan Guthrie wrote:

> On Thu, 5 Oct 2000, Antonio Querubin wrote:
> 
> > On Thu, 5 Oct 2000, Jonathan Guthrie wrote:
>  
> > > Yes, ::1 is a perfectly fine interface identifier, and I make use but it's
> > > not a GLOBAL interface identifier.  That is, interface identifiers with
> > > the global bit set are guaranteed to be unique on the Internet.  Perhaps
> > > my terminology is incorrect, but unless I can put multiple globally-unique
> > > interface identifiers on the same box, an awful lot of the magic IPv6
> > > promises simply isn't there.
>  
%%%%% snip %%%%%%%%
> 
%%%% snip %%%%%%%
> I'm not talking about a globally unique IPv6 ADDRESS, I'm talking about a
> globally unique bottom 64 bits.  If you can guarantee that the bottom 
> 64 bits are unique, then you can tell if two different IPv6 addresses 
> (which ARE going to be globally unique, just like IPv4 addresses) are
> really from the same computer. There is no equivalent to this in
> IPv4.  There is nothing even close to this in IPv4.

Ok, now we are talking.

I'll say what noone has told openly : THERE IS NO WAY TO GUARRANTEE
A GLOBALLY UNIQUE BOTTOM 64 BITS. 

Mac addresses did not do it, i have seen ethernet boards with duplicate
addresses, i have seen boxes where the adresses comes from otger 
places, i have seen eqipment where no IEEE adresses is used at all.

How on earth could one make any assumtions that the bottom 64 bits
is unique ?  At the best one could say "they are probably unique on
this subnetwork".


> 
> > > I believe that many people will want to have multiple globally-unique
> > > interface identifiers on the same computer.  Is there a standard way of
> > > doing this?
>  
> > The most common way hosting sites assign multiple addresses is to just
> > number them sequentially.  It's pretty simple and mindless and doesn't
> > require knowledge of the ethernet address on any NIC card.  Ie. it's KISS
> > compliant :)
> 
> Since you don't understand what a globally-unique interface identifier IS
> and how it's different from a globally-unique address, you probably won't
> understand why what you describe is not particularly useful, with respect
> to easy multihoming.

It's understandable that noone understands what a "globally-unique interface
identifier IS". During the BIG-IP list days (where IPng was 
discussed), during the period where ipv6 took form, there was
a discussion about a thing called "locators". Whenever
discussion focused on locators, they were supposed to look
like "IEEE-adresses", but with other properties. Noone ever
came with a good description.

Now on top of that, IEEE is no infinite source of addresses. 
The 46 bits available is beginning to be scarce, increasing
them to 62 bits will only delay the lifetime. As opposed to
ip(4) adresses then may not be reused. What will be used
as " globally-unique interface identifiers" when we are
out of IEEE adresses ??


> -- 
> Jonathan Guthrie (jguthrie@brokersys.com)
> Brokersys  +281-580-3358   http://www.brokersys.com/
> 12703 Veterans Memorial #106, Houston, TX  77014, USA
> 
> 
> 

--
Peter Håkanson            | Melissa ? Lovletter ? Joke ? Mothers day ?
tfn 0707 328101              | Not on this mailsystem!
IPSec sverige || ipsec.nu    | It's safe by design !
Lundbystrand  Sweden         | (in contrast to some other)