Re: Why Scopes? (was: Re: [saad] About saad)

> On Friday, October 17, 2003, at 01:03 PM, James Kempf wrote:
> > One of the things I'd like to see is a list of why people use scoped
> > addresses (RFC 1918) in IPv4.
>
> I've talked to a very large number of people about this (or
> rather why they use NATs, which is a slightly different
> question), and the most common reasons are:
>
> 1) don't want to buy more addresses
> 2) simplification of network management/renumbering
> 3) security/firewalling/unreachability
>
> The first two are already being dealt with in one form
> or another.  The third is only peripherally being addressed
> and certainly not satisfactorily (for whatever value of
> "satisfactory").  The reality is that some large number
> of users, including some users who consider themselves
> relatively expert (network administrators, etc.) don't want
> their hosts to be reachable by default but they do want
> them to be able to initiate connections themselves.  I'm
> not sure there's a good answer to this question, since
> the users' wishes are incompatible with the IETF's working
> assumptions about reachability.
>

But there are other ways that one could imagine doing this and still
maintain global routability only on the outbound connection.

For example, I've got a NAT at home on my 802.11/802.3/DSL access box. Now,
as a consumer, I don't have much choice in the matter: it's the only
technology out there that provides the functionality I want. And, it was
really easy to set up: plug it in, configure via a Web page, and it worked.
I suppose I could pay my DSL provider for more addresses, but I typically
just use one machine at a time (it might not be the same machine) but not
always and I don't leave it on all the time (due to electricity cost).

Suppose that, instead of a NAT, I could buy a box that had some number of
globally routable IP addresses preconfigured into it (and I could get more
by downloading them from the manufacturer via their Web page, maybe paying a
small fee). Suppose also there were some way for that box to communicate
with my service provider, without requiring a complex human intermediated
(and perhaps suits intermediated) business and technical conversation to set
up routing characteristics between the box and my ISP's network. The
communication would allow global routing outbound for purposes of initiating
a connection, but not inbound, and would be driven off my service profile
with the ISP (so that, for example, if I had a server, was paying more, and
needed the inbound connectivity, that would happen).

This would have no impact on the address architecture. The addressess on the
box could be globally routable, they could be HH IPv6 provider-independent
(if I'm using IPv6), but they need not be, since the manufacturer of the box
could apply to their local RIR for the address block like anybody else that
wants addresses. It would only require the routing reachability to be
configured properly, and in a way that is considerably more automated than
today.

                jak

_______________________________________________
Saad mailing list
Saad@ietf.org
https://www1.ietf.org/mailman/listinfo/saad