Re:Measuring open DNS resolver use
by denton-scratch on Tuesday September 24, 2019 @01:31PM
(#21062573)
"Unalloyed" is a strong claim - you are saying effectively that it is pure, flawless gold.
[ Parent
| on HN ]
I can't think of anything bad about it. One popular strain of criticism is clearly bogus, that it removes a measure of visibility from network operators (that's the point). Another is that it centralizes DNS services at companies like Cloud Flare, which, no, you can run your own DoH resolver server (and probably shouldn't use Cloud Flare regardless). Finally, people say that it's clunky compared to datagram-based alternatives; from my perspective, DNS has been hampered by the UDP service model for decades, and revisiting it so that, at least at the retail level, it's TCP/HTTPS like everything else is an opportunity, not a problem.
>you can run your own DoH resolver server
How does this work in your mind?
I don't understand your question. The first Google SERP will give you a multiplicity of tutorials on setting up a DoH server.
At the point when I would be "run[ning] my own DoH resolver", I'm already at the point of setting up my own resolver. Why wouldn't I just turn on unbound and call it a day?
The point of running your own DoH resolver is that you're doing it somewhere off-network, to tunnel DNS out of an ISP network you don't trust to somewhere else you trust more.
You can also just not run DoH if you trust your network. I'm saying DoH is a good thing, not that everyone has to use it.
Alright, but the reason for the complaint about centralization you're dismissing is that most people are encountering DoH via Firefox's plan to default resolve via DoH to Cloudflare, rather than in the context of being deft Unix hands searching for a way to tunnel their DNS out of an untrustworthy network.
I do not think a feature that delegates DNS to Cloud Flare is an unalloyed good. I think the protocol is. The distinction is important.
Re:Measuring open DNS resolver use
by denton-scratch on Tuesday September 24, 2019 @05:30PM
(#21064840)
Oh. Well, I lean to the view that X-over-HTTP is often a bad model; partly because HTTP has been heavily-commercialized.
I can see that there are problems with DNS; there always have been, and it's always been obvious. But for now, I think I prefer running my own resolver. Unbound is really easy to set up.
I'm not seeing how unbound solves the problem of your network provider spying on and recording all your DNS lookups.
Unbound does support DoT.
Sure, but you're not speaking DoT to the authority servers. DoH and DoT aren't magic security dust; they're just a means of tunneling requests from an untrusted network to a more trusted network. If you run your own recursive resolver locally, DoT is mostly theater.
Re:Measuring open DNS resolver use
by ectospheno on Wednesday September 25, 2019 @09:14AM
(#21070427)
Your ISP knows where you go without looking at your DNS. Because after you do the lookup you go there.
Serious question - why do you prefer DoH vs say, DNS over TLS? The former always seemed rather hacky to me.
I don't think there's anything wrong with DoT, but DoH is deliberately more difficult to block, which is a sensible design goal for a privacy and anti-censorship protocol. Both are better than directly using legacy plaintext DNS.