Re: IPv6 allocatin (was Re: ARIN Policy on IP-based Web Hosting)

[Vadim Antonov <avg () kotovnik com>]

On Fri, 1 Sep 2000, Scott Francis wrote:

> > Are we building production networks or doing experiments?  IPv6 exhibits
> > no added functionality over IPv4 + NAT, so why bother?
>
> what?
>
> *scratches head*
>
> that's odd, I was under the impression that an order of magnitude more
> address space, all of it PUBLICLY ROUTEABLE,

If that was of any importance, a trivial addition of an "extra bits"
IPv4 option would suffice.  In fact, majority of network hosts are not
"Publicly Routeable" for the simple reason that they're sitting behind
firewalls, or have dynamic addresses.

> without any translation
> bottlenecks (and yes, there *are* translation bottlnecks on a setup
> with several thousands IPs running through a single NAT box),

Are you trying to do NAT at OC-3? :)  Actually, a newer faster PCs
can, probably, do that at even higher bitrates :)

It is _very_ easy to install as many NATs in parallel as you wish, simply
by segmenting private address space, and routing different segments
through different NAT boxes.

(And any application-level firewall is already a "NAT" :)

> with support for encryption in the packet format, constituted 'added
> functionality'.

Mmmm... I apparenty have a delusion of having a working IPSEC in my
box...

> Apparently I was mistaken.

No, I agree that having these things available in the initial design
is nice; but that alone does not justify redoing the entire network
from scratch, since pretty much the same effect can demonstrably be
achieved using the already-deployed technology.

--vadim