On Fri, 1 Aug 2003 14:32:39 +0100 Michael.Dillon@radianz.com wrote:
I have been plotting the IPv6 ASNs for some time. These should be the ISPs running IPv6. See: http://www.nlnetlabs.nl/ipv6/measurements/index.html
It would be interesting to see an analysis that combines this data with Geoff Huston's IPv4 analysis http://www.potaroo.net/ispcolumn/2003-07-v4-address-lifetime/ale.html and see if we can predict the point at which the number of IPv6 addresses deployed begins to exceed the number of IPv4 addresses deployed? I realize that the IPv6 analysis is routes only, but one should be able to determine how many addresses are available in each ASN.
One could reasonably assume that at the point where the Internet shifts to IPv6 as the core protocol, more than half of the interfaces with an IPv4 address will also have an IPv6 address. So to get to that point, one could make some assumptions about the allocation of IPv6 /48's based on the observed trends in IPv4 /32's.
I'm not sure where one would take this, but I think a lot of people would be interested in seeing some type of well-presented analysis of these questions.
It's not worth doing a fine analysis to predict so far in the future - a back of the envelope will do just fine :) Look at ASN : http://www.nlnetlabs.nl/ipv6/measurements/index.html shows that IPv6 ASN (as seen fron NLNetLabs) are doubling about every 1.75 years, and are about 340 now. So, IPv6 ASN can be modeled as N_6 = 340 x 2^(T/1.75) where T = time - 2003.5 in years. Now, IPv4 ASN withb routing are growing linearly lately (see Figure 2b in http://www.multicasttech.com/status/index.html for example) and can be roughly modeled as N_4 = 15000 + 1750 x (t - 2003.5) = 15000 + 1750 T Set N_4 = N_6 and we see that the number of IPv4 and IPv6 ASN with routing will be equal in a little less than 12 years (T ~ 11.75), or some time in the Spring of 2015. This is far enough into the future that I do not think that it is realistic to be more rigorous than this. Regards Marshall Eubanks
--Michael Dillon