On 4/9/19 00:34, Tom Herbert wrote: > On Tue, Sep 3, 2019 at 1:49 PM Fernando Gont <[email protected]> wrote: >> >> On 3/9/19 23:33, Tom Herbert wrote: >>> Bob, >>> [....] >> "fragile" means that it fails in an uncceptably large number of cases. >> ~30 failure rate is not acceptable. ~20% isn't, either. >> > Okay, so then the definition of a fragile protocol is one that has a > failure rate greater than 20%? If so, then that should really be > stated in the draft. But, then I'd point out that per > https://www.google.com/intl/en/ipv6/statistics.html, IPv6 is only > supported by 30% of the Internet at best, so it fails 70% of the time. > So by that same definition, isn't IPv6 fragile?
That's of course a bad analogy. Where there's no IPv6, you just don't use IPv6. -- IPv6 is not broken, because you simply don't use it. IPv6 broken-ness existed when hosts would try it, but it would fail. This was indeed the reason why content providers didn't enable v6 at the time, and the reason why Happy Eyeballs was introduced -- to overcome the <10% IPv6 brokenness at the time, which was unacceptable to content providers. What's the specific number that's unacceptable? -- I'd personally that at the very least, anything over 10%. I wonder if you'd go on of vacation with your car if you new it had a 30% chance of breaking apart. That said, Bob is addressing the comments that were made on this thread, so this discussion is rather spurious, I think -- Fernando Gont SI6 Networks e-mail: [email protected] PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492 _______________________________________________ Int-area mailing list [email protected] https://www.ietf.org/mailman/listinfo/int-area
