[ppml] Big numbers

Bill Darte billd at cait.wustl.edu
Tue Apr 8 09:18:48 EDT 2003


Of course I can recall when the thought of needing more than 256k of RAM and
20MB hardrives was unimaginable and then how could we ever run out of
addresses with over 4 billion in v4.... the issue is always we can't imagine
the need when we establish the standard.  The pain of transition from v4 to
v6 is going to me long and intense... the only thing worse than no standard
is a widely deployed, but obsolete standard.  

Would you have guessed in 1982 that GPS would direct cruise missiles to
targets within feet over hundreds of miles?  Would you have believed that as
a standard feature of higher-end automobiles you would have navigation and
interactive voice technologies or even the pervasiveness of wireless PDAs?
Did you plan to have access to email and millions of webservers around the
globe in 1982?  The Internet and global communications is just beginning.
It took 15 years for PCs and Networks to be truely integrated into
businesses.  It will take a similar time period to understand the potential
for the Internet.  Massive integration of systems will demand unique
addressing and massively integrated systems are a certainty because
automating processes is reliable and inexpensive ways to manage complexity.

NAT is obtrusive and costly because it requires human intervention and it
precludes important technologies today. 

In 20 or even years when the world or communications integration has
matured, who is to say what the needs will be. Security may always be a
limiting factor for exposure of systems globally, but I doubt it. 
Why make the same mistakes we made with earlier estimates of need?  We have
evidence that resources are not inexhaustable and that the pain of
transitions are disruptive and ugly.  Why set ourselves uf for it yet again
with a communication vehicle which will be orders of magnitude more embedded
and pervasive...running unimaginably important systems.... let's not have to
change again, because the ease of waste seemed correct in the face of an
inexhaustable address space.


Bill Darte

-----Original Message-----
From: Brian Bergin
To: 'ppml at arin.net '
Sent: 4/7/03 6:25 PM
Subject: RE: [ppml] Big numbers

With NAT technology getting better and better do you really think every
TV and toaster needs a public IP address?  Every individual and company
already has access to millions of private IPs under IPv4.  Encouraging
the use of public IPs on devices/computers with absolutely NO need to be
on the public Internet is only going to allow hackers to ruin your
holiday dinner when they hack your oven and change the temp to 600°F and
you get burnt turkey or they turn off your furnace or turn off your
water heater when it's -20°F outside.  Any way you look at it, putting
an IP on "every electrical and electronic component as well as
subsystems elements" is a bad idea, IMHO....

Maybe I'm just missing the big picture.  Conservation, IMHO, just isn't
that crucial for IPv6.  I remember when IPv4 was going to be gone
"tomorrow" or "next year" only a couple years ago.  I believe NAT has
had a big part in the life extension of IPv4.

Just my 3¢ (inflation you know...)

Brian Bergin
ComCept Solutions, LLC

At 18:28 07 04 03 Monday, Bill Darte wrote:

Of course, we are not talking about numbering individuals, but
every electrical and electronic component as well as subsystem elements
perhaps....  there is no census data for these things, but undoubtedly
represents a very large number as well.

Bill Darte

-----Original Message-----
From: David Conrad
To: ppml at arin.net
Sent: 4/7/03 2:56 PM
Subject: [ppml] Big numbers

Apropos a comment I made during the Q&A during the IPv6 working group 

According to the latest IPv6 architecture drafts:
- 35,184,372,088,832 /48s currently available for assignment
- a bit under 246,290,604,621,824 /48s available under the other format 

Just for fun, according to the US Census bureau:
- Estimated world population as of 4/7/03, 15:29 GMT+5: 6,285,260,947
- Estimated world population in 2050: ~9,000,000,000

Taking the 35,184,372,088,832 /48s currently available for assignments, 
this means:
- 5600 /48s per person today
- 3909 /48s per person in 2050

And then there are the other format specifiers...

Note that those are /48s (each capable of addressing 64K /64s or, if 
you want ignore the auto-configuration goop that eats the lower 64, 
1,208,925,819,614,629,174,706,176 /128s).

As such, I don't believe address conservation is or will be an issue.  
At least for the lifetime of IPv6.  Keeping the routing system 
constrained undoubtedly is, although I'm not convinced this is the RIRs 
job (after all, RIRs explicitly do not guarantee routability)...

(Hope I got my math right... :-))


More information about the ARIN-PPML mailing list