On May 12, 2011, at 9:06 AM, Scott Whyte wrote: > On Wed, May 11, 2011 at 23:10, Franck Martin <fmar...@linkedin.com> wrote: >> I think the yahoo test should just differentiate between no IPv6 and IPv6 >> is slow (test between 3s and 10s). Like: >> >> We have detected that you have IPv6 and will be able to access our site on >> IPv6 day, but your user experience may not be as good as with IPv4, you >> may consider disabling IPv6. >> > > Measurements during the experiment won't be directly comparable to > those before/after, at least as far as I can see. So they will be > informative, but its the slope of the brokenness line before/after > that will determine when IPv6 is not an impediment to itself. > > -Scott
I think it's a little more complex. I think there are two lines. A line representing brokenness with AAAA records enabled and a line representing brokenness without AAAA records. The first line is trending downwards while the second line is trending upwards and wil soon be making a rather pronounced increase in its slope. When these two lines cross, I think it will become virtually inevitable that those who are ready to do so will publish their AAAA records. Owen
