From: Pekka Savola (pekkas_at_no.spam)
Date: Tue Mar 19 2002 - 08:19:36 PST
On Tue, 19 Mar 2002, Octavio Medina wrote:
> I totally agree. Adding NAT to DSTM is a bad idea. Adding NAT no
> anything is a bad idea anyway. I think the other authors of DSTM agree
> with me that one of the main goals of DSTM is to AVOID translation. I
> will never support adding a NAT to DSTM to solve the
> lack-of-IPv4-addresses problem.
But the site which does this had a problem with lack of IPv4 addresses to
being with!
I've yet to hear of an organization (except for a very few examples) which
would have to give back IPv4 addresses if it starts using IPv6!
> I believe DSTM ports is a better choice
> for that.
Sure, it _sounds_ an interesting idea.
But I don't think it's going to _work_ in a robust, reliable way except
under some very specific scenarios.
Therefore I think ports option can be very bad: I wouldn' want to define
even an optional _IPv6_ mechanism that may be unreliable, etc.
> Stig Venaas wrote:
> >
> > On Tue, Mar 19, 2002 at 02:16:50AM -0500, Bound, Jim wrote:
> > > thats good logic and I can't refute it. and we cannot change the DSTM
> > > assumptions that would be unacceptable. The assumption is really a
> > > deployment strategy and thats the whole point of DSTM. So if we cannot
> > > make any DSTM option support that assumption in my view the option is
> > > DEAD.
> > >
> > > but why not at least discuss the other benefit which is this.
> > >
> > > with the port option is it possible to give the user back e2e with its
> > > limited IPv4 address space. Kind of like right-sizing a bad situation
> > > and helping the user along.
> > > Just in case they cannot move to a pure Native IPv6 network till say
> > > 2005.
> >
> > I agree the port option is a bit ugly, but it's much much better than
> > NAT which would be the alternative. If someone has enough IPv4 addresses
> > they shouldn't use this option. If you have very few IPv4 addresses, you
> > may be forced to share an address between multiple hosts. Doing this the
> > DSTM+port/RSIP way breaks very few applications compared to NAT(-PT).
> > In particular addresses and ports embedded in the application protocol
> > is less of a problem. I think much of what is said about RSIP in RFC
> > 3103 also applies to DSTP+port.
> >
> > Stig
>
-- Pekka Savola "Tell me of difficulties surmounted, Netcore Oy not those you stumble over and fall" Systems. Networks. Security. -- Robert Jordan: A Crown of Swords
This archive was generated by hypermail 2.1.7 : Fri Oct 06 2006 - 00:00:31 PDT