Re: [v6ops] An Update to Happy Eyeballs

On 23/03/2017 04:18, Job Snijders wrote:
> Perhaps of interest to this group:
> 
> https://blog.powerdns.com/2014/05/21/a-surprising-discovery-on-converting-ipv6-addresses-we-no-longer-prefer-getaddrinfo/

What a bizarre idea to use getaddrinfo() for format conversion, when inet_pton() and inet_ntop() exist.

My sympathy is limited.

   Brian

> 
> Kind regards,
> 
> Job
> 
> On Wed, Mar 22, 2017 at 4:15 PM, Lorenzo Colitti <lorenzo@google.com> wrote:
>> On Thu, Mar 23, 2017 at 12:04 AM, Fred Baker <fredbaker.ietf@gmail.com>
>> wrote:
>>>
>>>> I think it's extremely hard to argue that waiting for an A query when
>>>> you have an AAAA in cache (or vice versa) is going to reduce latency or
>>>> improve user experience. You can use getaddrinfo if you want, but you're
>>>> going to pay the price in terms of latency. I wouldn't call that a best
>>>> practice.
>>>
>>> I must be missing something. If you already know the AAAA, fine. If you're
>>> using the resolver's cache, what API gets you an IPv6 address other than
>>> getaddrinfo?
>>
>>
>> You don't have to use getaddrinfo to do AAAA lookups. You can use your own
>> code, or even pass binary blobs to res_send if you want.
>>
>> getaddrinfo is not very smart. It keeps trying until it has answers for
>> everything you asked it to do, and then it returns everything. That means
>> that if the A query is in cache and the AAAA query is not, then you have to
>> wait for the AAAA query to succeed.
>>
>> I think the key point here is that integrating the DNS resolver retry loops
>> with the connection attempts provide latency savings.
>>
>> _______________________________________________
>> v6ops mailing list
>> v6ops@ietf.org
>> https://www.ietf.org/mailman/listinfo/v6ops
>>
> 
> _______________________________________________
> v6ops mailing list
> v6ops@ietf.org
> https://www.ietf.org/mailman/listinfo/v6ops
> 