[Gllug] Public IPs - When are they appropriate

[David Damerell]

On Wednesday, 14 Nov 2001, Paul Brazier wrote:
>> the surface of the earth. But only some of the IPv6 addresses 
>> are assigned
>> for general use and of these ones there is only one for every 
>> square meter
>> of land on the surface of the Earth.
>there should be 256^4 = 2^32 = 4,294,967,296 normal IP addresses
>and 256^6 = 2^48 = 281,474,976,710,656 IPv6 addresses

Er, no; MAC addresses are 6 octets (and you'll notice no-one is
worried about running out of those.)

IPv6 addresses are 16 octets - 2^128 addresses; 3.4x10^38.

The "Aggregatable Global Unicast Addresses" represent 1/8 of the total
address space (according to RFC2373), so some 2^125 IP addresses
(however, it's worth noting that about 3/4 of the range remains
unassigned, so we could release some of those for toasters if
necessary); 2^125 is about 4.3x10^37.

If we assume the Earth is a sphere 10,000km in diameter, its surface
area is 4 x pi x (5 x 10^6)^2; 3.1 x 10^14 m^2. So it's quite
massively bogus that there's only one for every square meter.

Even if we assume that the last 6 octets of the IPv6 address will be
the MAC address - so the unicast addresses effectively only offer
2^(125-48) addresses - 2^77 - that's still 1.5x10^23, or 5x 10^8 per
square meter of the Earth's surface.

[There's an obvious fallacy there in that not all those 2^77 devices
can have one of the 2^48 MAC addresses - but even if all devices have
a MAC address, 2^48 MAC addresses is 2.8 x 10^14. Which is probably
where this 'one per square meter' confusion comes from.]

-- 
David Damerell <damerell at chiark.greenend.org.uk> flcl?

-- 
Gllug mailing list  -  Gllug at linux.co.uk
http://list.ftech.net/mailman/listinfo/gllug