Re: Fwd: A comment about MAST

["Eugene M. Kim" <gene@nttmcl.com>]

Sorry, you seem to have filtered `this isn't some 10 years ago' part.  I
do admit that MAST is strikingly similar to NAT in a sense you have
pointed out below.  What intended to be a short term solution becoming
prevalent.  Yes, very unfortunate.

Now, what led to that NAT havoc?  IMO there are two reasons.  One is
lack of the long-term solution developed in reasonable timeline, and
the other is difficulty of phasing out NAT.  The former is as simple as
developing a new solution (which I hope is something we are here for :p)
so I don't think it is too terrible a situation for it's a ball in our
hands.

The latter is what people might be worried about, because, if MAST is as
hard to phase out as NAT is, it will be very frustrating when we finally
have the real solution ready for deployment.  But no, IMO, MAST is a lot
less harmful than NAT is in that sense.

NAT affects the entire network design, namely what address is being
allocated to each and every machine in the network.  Having to redesign
and reconfigure an entire network is every network administrator's
nightmare, especially considering what will result in when he makes a
mistake doing so (which is an entire network broken, of course).  So,
unless there is a *very* compelling reason that affects the everyday
operation, they will simply refuse to give up NAT.

However, MAST does not affect the network design and configuration
because it is a host-to-host protocol.  That means there will be a lot
less network structural changes, and in turn, less administrative
overhead, less headache for administrators.  They will have to upgrade
networking stack of operating systems running on their machines, but
they don't have to do it all at once (provided the new solution can
coexist with MAST, which I think must be done and is not terribly hard
to achieve); they can start upgrading less critical machines first, one
by one, then they can tackle more critical machines.  As you see here, a
conservative and gradual phase-out is possible, making its risk a lot
lower than phase-out risk of NAT.

Finally, I'm afraid I cannot fully see what you said in your postscript;
could you elaborate?

Eugene

P.S. Yes, I do hate NAT.  I just try to have a calmed-down view about
it, so I don't hate things just because they resemble NAT in some
aspect. =)

On Wed, Sep 10, 2003 at 05:21:09PM +0859, Masataka Ohta wrote:
> Eugene;
> 
> > Granted, we need some architecturally solid and sound solution, i.e. the
> > ultimate, long-term goal, to the ID/LOC problem.  However, that will
> > probably mean introducing a new namespace for either identifier or
> > locator (depending on which purpose we decide to use the current IP
> > address space for), and applications must adapt itself to the new world
> > order.
> 
> Granted, we need some architecturally solid and sound solution, i.e. the
> ultimate, long-term goal, to the address exhaustion problem.  However,
> that will probably mean changing the IP protocol and applications must
> adapt itself to the new world order.
> 
> And, NAT was introduced.
> 
> > I do hate NAT,
> 
> You don't, really.
> 
> 						Masataka Ohta
> 
> PS
> 
> With proper API, most applications works as is even with ID/LOC
> separation.