On Wed, 1 Oct 2008 10:45:21 -1000 (HST)
Antonio Querubin <[EMAIL PROTECTED]> wrote:

> On Wed, 1 Oct 2008, Dunn, Jeffrey H. wrote:
> 
> > Antonio,
> >
> > So are you suggesting that we replace IPv4 NAT with IPv6 routing proxy?
> 
> I'm suggesting not relying on NAT as a crutch in your specific example. 
> It seems too easy to just mirror IPv4 practices onto IPv6 and not look at 
> how IPv6 capabilities enable other potential solutions.

This is the exact mental trap a lot of people seem to fall into.

IPv6 is similar to IPv4, but I'd argue that it's even more
similar to Novell's IPX and Apple's Appletalk than it is to IPv4.
IPv6 seems to me to have many of the features of IPv4, IPX, Appletalk
and other protocols before it that have proven to be useful. That means
if you have some operational experience with these other protocols, you
can also think about where and how they were better than IPv4,
and also see if you can apply that thinking to IPv6, for features of
them that have been carried through to IPv6.

When I think about fixed length subnets and whether they're useful or
not, I think back to my pre-IPv4 days of working with Novell's IPX with
it's fixed length 48 bit node addresses, and in particular, how much
easier and simpler it was having fixed length subnets everywhere
on the network, including on point-to-point WAN links. We didn't have a
choice about whether we "wasted" node address bits or not because
it was part of the protocol spec, but it never caused any problems
either - for point-to-point links we never said, "I wish we weren't
forced into wasting those other 47 bits."

Ethernet addressing also "wastes" bits for operational convenience -
who's ever going to put 2^46 hosts on an ethernet segment! It's the
original "plug and play" networking technology, and those design
decisions were made in early 80s. The paper "48-bit Absolute Internet
and Ethernet Host Numbers" describes the considerations behind this
decison.

IPv6 should be at least as easy to use as Ethernet, IPX and Appletalk
are or were. If it isn't then I think we've failed. 64 bit fixed length
subnets is one of the ways this goal can be achieved.

I'd really encourage people who don't have any knowledge of or
experience with any other layer 3 protocols other than IPv4 to have a
look into how those other protocols were different and how they worked.

Regards,
Mark.
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to