Yesterday's Windows update causes IPv4 to be default

[Doug Barton]

On 11/20/2012 5:21 PM, Lorenzo Colitti wrote:
> On Wed, Nov 21, 2012 at 9:56 AM, Doug Barton <dougb at dougbarton.us
> <mailto:dougb at dougbarton.us>> wrote:
>
>     I'm curious about why MS chose not to simply implement the Happy
>     Eyeballs protocol?
>
>
> HE has drawbacks,

To be clear, I am not saying HE is perfect. Learning from actual 
deployment experience is important for us to do it better next time, 
because the transition is going to be a long one.

> such as additional server connection load, tendency to
> prefer IPv4 more than necessary (just look at Apple's implementation,
> which prefers IPv4 most of the time),

FYI, Apple did not implement HE "by the book," they used their own spin 
on several parts of it, this being one of the biggest differences.

> and tendency to mask IPv6connectivity problems.

IMO turning it off altogether, in a way that is not obvious to the user, 
also masks the problem pretty severely. I'm glad to hear that it's only 
successes that are cached for 30 days however.

> MS's implementation fixes the "pathologically broken IPv6" case, which
> is the one that really needs to be fixed, while preferring IPv6 all the
> time everywhere else. I think that's a good tradeoff.

I tend to agree with the spirit of this implementation (and for the 
record, I would have done HE differently in any case), but my immediate 
concern is for the "cached known-good IPv6" state when there is an 
intermittent problem on the link.

Doug