>>905758
Don't forget to add 1.0.0.1 as well.
>>909448
> Maybe we could use HTTP Headers
We actually do. SNI (server name indication) checks the Host header in the HTTP request. Your browser automatically sets this to the DNS name of what you're connecting to. Technically you could do this without DNS somehow.
>>909844
>being so insignificant he wants to go back to the days of hosts files.
No. DNS is useful because it offers a central database of abstract name -> routable address mappings. These mappings change, can vary by requester, and so on. Even if we could boil every network service you will ever connect to down to a single IP or set of IPs, if you think you can actually just write them down and expect to be able to stay on top of every time they change, you either do nothing on the internet, or have way too much free time.
>>909873
So I have a service that makes a large amount of queries, and in order to not get rate-limited, I round-robin it across every public DNS I can find. On thursday (over the course of 24 hours) google's public DNS had 4 failed queries on 8.8.8.8 and 3 failed queries on 8.8.4.4. Google is historically the worst, but clownflare has now taken the crown, with 68 failed queries on 1.1.1.1 and 54 on 1.0.0.1 over thursday.
Also nice is this little gem:
>independent DNS monitor DNSPerf ranks 1.1.1.1 the fastest DNS service in the world.
>DNSPerf: our other projects: jsDelivr (scroll to the bottom of the page to see this)
>jsDelivr operates thanks to our main sponsors: Clownflare and a few others
Really makes you think.
Also notice that a lot of the latency numbers are so short that they would have to be in the same datacenter. I wonder why DNSPerf's monitoring points seem to be in the same datacenters as clownflare's DNS?