On 06/26/2017 08:05 PM, JORDI PALET MARTINEZ wrote: > Regarding the different issues that are indicated because Happy Eyeballs, a > possible solution is to increase the timers, so even if IPv6 is “slower” than > IPv4, is not just a matter of 50 milliseconds (for example), but several > seconds, so then IPv6 will get preference anyway. I’ve described this in a > recent email related to the HE-update document (so you could use some of the > text that I mention in that email):
Please be careful about this (for problem 6): I've already seen people advising to "just disable IPv6 to get a speed boost." If the timeout is going to slow down to several seconds, this may make more people switch to using IPv4 only, which would be the complete opposite of the objective. Sure, you also consider making this setting a knob, but people may go the easiest already known way, because "well, who cares about IPv6 anyway?" Then, all this depends on the proportion of the internet that is already quickly IPv6-reachable, but if there is a problem to solve, it can only be solved by making IPv4 slower when IPv6 is enabled... which sounds like a bad idea to me so long as IPv4's sunset (or IPv6's rise?) isn't complete yet (at which point this solution is no longer needed anyway). Best regards, Leo
signature.asc
Description: OpenPGP digital signature
_______________________________________________ sunset4 mailing list [email protected] https://www.ietf.org/mailman/listinfo/sunset4
