IPv4 meets the exponential functionIf you didn't watch it yet, you should. It's long, but worth watching in full.
In it Dr Bartlett spends some time describing hypothetical scenarios in which, say, consumption or population grow at a steady rate until there's nothing left to consume or no space left for additional population. Shown graphically, we see a rising exponential curve, which stops abruptly and drops to zero.
Dr Bartlett goes on to say that this is an unrealistic scenario and that, in the real world, such charts can be expected to have a classic bell shape. Consumption (or population &c.) grows exponentially for a while, peaks and then decays along a similar, but reversed exponential path.
Well, this got me thinking about IPv4 - again.
Unlike oil consumption or population, IP is a purely logical artefact. It does not physically exist. Would the consumption of the IPv4 address pool follow a similar, bell shaped curve?
There's a partial answer at this page.
The graphs there of IPv4 consumption do indeed show a broadly exponential growth, as we might expect.
However, despite the fact that a very high proportion of IPv4 is now used (a far higher proportion than has been used of oil or gas), that exponential growth shows no sign of stopping. Figure 18b shows the phenomenon well, and shows the cliff edge in late 2011.
I rather doubt that we will see a catastrophic drop like that, but this would seem to suggest that the IPv4 consumption curve is not a symmetrical bell shape, but is significantly skewed towards the latter days.
It also suggests that IPv4 really will be exhausted in 2011. In other words, the Internet's all used up long before we run out of oil, coal or natural gas.
Now really is a good time to start planning your IPv6 strategy, but how long will that last?
Dr Bartlett also talks about doubling times and the false comfort that can accompany lazy thinking.
If a new reserve of oil is discovered today that is equal in size to the total amount of oil used in all of human history, how long would it last? At 7% annual growth in consumption, the doubling time is ten years.
Ten years are all you would get from that huge new reserve.
IPv6 gives a huge new reserve of IP addresses. Will that be exhausted sooner than we expect?
IPv4 utilisation has, coincidentally, been growing at around 7% per year, doubling every ten years. Projecting continued 7% growth well into the future, how many more doublings can there be before IPv6 is exhausted?
This is very simple mathematics.
IPv4 uses 32 bits and IPv6 128, so there are 96 more bits in an IPv6 address than in an IPv4 address.
Each additional bit in an IP address doubles the size of the address pool so, at the present rate of growth, IPv6 will last 96 * 10 = 960 years.
Let's just call it 1,000 years, give or take rounding.
More than enough time for someone to invent IPv7.
See also: Ruminations on a future IPv4 economy
Unable to post a comment? Please read this for a possible explanation...
-