On Tue, 2012-10-16 at 22:13 -0500, Brian Wood wrote:
> From: Jima 
> 
> >
> >  What have you done with or for IPv6?
> 
> A week or so of IPv6 programming in the code here -- 
> http://webEbenezer.net/build_integration.html
> .  
>   
> >  No, 64-bit host addresses wouldn't "make more sense."  If you're
> > thinking of the IPv6 address space as 128-bit host addresses, you're
> > doing it wrong.  Think of it as 64-bit network addresses, each with
> an
> > irrelevant number of hosts.  The lower 64 bits were engineered for
> > autoconfiguration, and can be ignored for 90+% of scalability
> discussions.
> 
> An IPv6 packet header has the source and destination addresses --
> both 
> 16 bytes.  The header is 40 bytes total.  If the addresses were 8
> bytes 
> the header would be 24 bytes.   I'm not sure what you mean by ignoring
> the 
> lower 64 bits in scalability discussions.  

You might be confusing bits and bytes. An IPv6 address is 128-bits (16
bytes). The lower 64-bits of a single IPv6 address means the right half
of an IPv6 address when reading it.

Jima's specific reference was to SLAAC, see the wikipedia article on
that for example.
http://en.wikipedia.org/wiki/IPv6_address#Stateless_address_autoconfiguration