We’ve been looking for an analogy for two days to explain the size of the IPv6 pool compared to IPv4, but IPv6 space is so huge none of the comparisons are immediately comprehensible. See below for Jacob’s light years attempt.
However Kim Davies reckons he has come upon an analogy that works and, if the maths is correct, we may finally have a comprehensible explanation of the comparative sizes of the two pools. Before we start using it though, it seems only sensible to get others out there to check on the calculations. So the analogy and associated maths is below. Please review it and get back.
Volume of earth: 1.0832073 × 10^12 km³
=1.0832073 × 10^21 m³
=1.0832073 × 10^27 cm³
IPv6 space: =3.4028236 × 10^38
IPv4 space: =4.2949672 × 10^9
IPv6/IPv4 space =7.9228162 × 10^28
So, if all the IPv6 space was the size of earth, then you could fit IPv4 in 73.142 cubic centimetres.
An iPod is 2.4″×4.1″×0.55″ = 5.41 cubic inches = 88.68 cubic centimetres.
So, ball park, if all the IPv4 space would fit in an iPod, then all the IPv6 space is the size of the entire Earth.
{ 9 comments }
Fergie 06.17.07 at 11:21 am
I think you meant [above] that “…if all the IPv4 space would fit in an iPod, then all the IPv6 space is the size of the entire Earth.”
- ferg
Watches 06.17.07 at 5:47 pm
And how about the Zune!?? Heheheh It’s a little bigger!
Michele 06.18.07 at 2:41 am
Kieran
It’s nice to see you guys writing about IPV6, but it would be really really nice if you could give clear accessible examples of why IPv6 is cooler than IPv4
Michele
Paul Levins 06.18.07 at 1:26 pm
Michele
This is on it’s way in the form of a Factsheet on IPv6 that we hope to produce for distribution and use at our next f2f meeting at San Juan, Puerto Rico.
So it’s high on the priority list
Paul Levins
Sean Siler 06.19.07 at 12:13 pm
Not that I think these kind of examples are relevant, but I tried to come up with the same sort of scenario last year. Here’s what I figured out:
If you shrunk the entire IPv4 address space into a single Hydrogen nucleus (10^-14M) how big would IPv6 be in comparison? I will spare you the math, but you would have to travel at the speed of light for 30 days to get the end of the string of nuclei that would be the size of the IPv6 address space.
The numbers are just too large; there is no way to comprehend how large 2^128 is.
Sean Siler
IPv6 Program Manager
Microsoft
http://blogs.technet.com/ipv6
David Conrad 06.19.07 at 1:46 pm
Interesting, but…
The way IPv6 has been designed to work, it actually works out to 2^64 LANs. In theory, you could have 2^64 interfaces on those LANs, but if you have more than a couple of thousand devices, you’ll likely deeply regret it. The reason for the insane amount of address space inside a LAN is to permit something called “stateless auto-configuration” and because it is architected into the protocol, half the IPv6 address space is used for it whether you use stateless auto-configuration or not.
Further, the IETF has mandated 16 bits to be used for “local topology”. This means if the IETF mandates are followed (and they aren’t universally), there would be a theoretical maximum of 2^48 “sites”, with each site having 2^16 LANs and each LAN capable of addressing up to 2^64 interfaces.
Of course, the IETF has only designated 1/8th of the entire IPv6 address space to be used for what people consider “normal” IP addresses, so this means there are “only” 2^45 sites that can be addressed under current policies (each site having 2^16 LANs, and each LAN addressing 2^64 interfaces.
Still a very, very big number.
Rgds,
-drc
Butler 06.21.07 at 7:18 am
I’m very far away from being a math major so that comparison is great and very easy to understand. Just forget about all the numbers and lead any press/pr info with that comparison. You can put the math whiz stuff in the footnotes.
Dennis 07.02.07 at 4:40 am
I agree with Michelle. I saw a powerpoint talk on IPv6 in which the author said it would take longer for a router to process an IPv6 packet because of the longer addresses. From my readings, and what I had put in my presentation, is that it would take less time for a router to process a packet b/c of the fixed header size. I was asked how much less time it would take, but I don’t know.
I work in modeling and simulation, and if packets can move across the country faster, it gives the simulation more realism. The same would be true of online games.
Enough with the more address space stuff! Let’s hear about packet processing speed.
iPod 08.21.07 at 11:10 am
This only means there’s still a huge market share for Apple to conquer
Comments on this entry are closed.