Over-utilisation of v6 neighbour slots

[Daniel Roesen]

On Thu, Oct 24, 2013 at 08:37:25AM -0400, Bill Owens wrote:
> On Thu, Oct 24, 2013 at 01:27:34PM +0200, Daniel Roesen wrote:
> > On Thu, Oct 24, 2013 at 01:20:17PM +0200, Gert Doering wrote:
> > > Apple is free to provide a reasonable implementation right away...  not
> > > that they would change it, just because there is an RFC...
> > 
> > Given their ignorance of collateral damage done to operators, users
> > and IPv6 deployment in general by their implementation of Happy
> > Eyeballs (alias "Hampering Eyeballs"), I have zero hope.
> 
> My recollection of the complaint about Apple's implementation is that
> it doesn't bias the choice in favor of IPv6; is that the root of the
> problem?

Yes. As connection setup latency is the only KPI used. As multiple
operators have outlined, NAT doesn't really add latency above noise
thresholds, so statistically the native, low-cost, unconstrained
(non-NATted) IPv6 path looks "same" like the (possibly tunnelled),
high-cost, crippled (NATted) IPv4 path.

AAPL often prefers NATted IPv4 above native, non-NATed IPv6. Damage
done. A reasonable bias would fix this for most cases, and no really
harm the corner cases as far as I can see.

Anyway, the users will have to pay for that. Too bad users of !AAPL
have to subsidize those decisions. Time for an AAPL user NAT tax? :)

Best regards,
Daniel

-- 
CLUE-RIPE -- Jabber: dr at cluenet.de -- dr at IRCnet -- PGP: 0xA85C8AA0