Five Security Flaws in IPv6

Geoff Huston gih at apnic.net
Thu May 10 21:20:06 CEST 2007


At 01:13 AM 11/05/2007, Nick Hilliard wrote:
>Geoff Huston wrote:
>>As the amount of remaining addresses space that these models look 
>>at diminish in size, then the degree of possible variance in the 
>>modelling exercise also diminishes - i.e. the convergence of these 
>>predictive models is also a relatively strong signal that this is 
>>no longer a long range prediction with high levels of potential 
>>variance, but is now working as a short term prediction with very 
>>limited potential variance.
>
>It might be more accurate to say that the upper bounds of the 
>address pool sizes can be predicted more accurately as time goes 
>by.  Your model does not take into account a possible stampede near the end.


Actually a quadratic model may better equipped to model the 
characteristics of a last chance rush  than an exponential growth 
model using least squares best fit curve fitting algorithms, but, 
yes, its still probably too conservative. So you comment appears to 
be that even this model is perhaps too conservative. Fair enough. 
What does that observation imply to a projected IANA pool exhaustion date then?

    regards,

    Geoff




More information about the ipv6-ops mailing list