ietf
[Top] [All Lists]

Fw: IPv6: Past mistakes repeated?

2000-04-24 13:40:02
If what you suggest should be implemented, then
probably the entire software of all the switches
and hubs need to be upgraded (if not entirely scrapped) .

That's what has to be done, anyway.  I'm not sure that I see what you are
saying.

As also everytime the source and destination addresses are
upgraded, all the systems and the related software needs to
be upgraded.

If you design it correctly in the first place, this isn't necessary.

Think of a railroad network as an analogy.  The current design for IP
addressing allows a fixed number of tracks, and you have to allocate them
all in advance.  If the future evolution of the network is such that your
allocation turns out to be less than optimal, you have to redo entire
sections of the network to reallocate tracks.

Now compare this with an open-ended addressing scheme.  All you have to do
in this case is allocate the first track.  As additional tracks are needed,
you build new ones, branching off from the first track.  If some branches
evolve more than others, no problem--they can just add additional branches
of their own.  No branch impinges on any other branch.  You might have only
two branches leading away from your original track.  One of them might lead
to a total of fifty stations (endpoints), but the other might lead to ten
trillion stations.  It doesn't matter, and you don't have to care, since
when you route trains on your section of the network, all you'll look at is
the first digit of the destination, which will tell you which of your two
branches the train must follow.
And you might be on a branch yourself, for that matter.  The network can be
restructured upstream or downstream of your little section of track, and it
remains transparent to you, as long as the digit designations for you and
the two branches you serve remain the same.

Personally my telephone number has changed
3 times within the last couple of years.

Probably because the telephone numbering scheme was not truly open-ended.
In the U.S., for example, attempts to fix the number of digits in telephone
numbers have caused great problems, with things like area codes being
exhausted, exchanges being exhausted, and so on.  A truly open-ended scheme
wouldn't have this problem--you'd just add more digits in the areas that
needed more numbers.  This open-ended scheme is actually in place for
international calling.  (Equipment usually has fixed-length buffers for
telephone numbers, but all you have to do is boost the size of the buffers
if you ever come across numbers that won't fit.)

But the million dollar question is that whether the
protocol and switch vendors would like to scrap the
years and amount of investment that they have already
made in the existing system.

Looks like they're pretty much doing that with IPv6 now!  And with a
fixed-length addres, they'll be doing it again in 15 years, only it will
cost a thousand times more than it did on this pass.

  -- Anthony