4 ms·

i was only responding to your question whether there is any difference between the two beside the port number used which there is, in a significant way. You can

i was only responding to your question whether there is any difference between the two beside the port number used which there is, in a significant way. You can not point a DoH client to a DNS over TLS server or vice versa. That said there is no reason a DoH server could not be recursive or consult a traditional (probably recursive) resolver for its response. I think the only benefit of DoH is that is unlikely to be blocked. Otherwise it feels more like a hack and at least i am not concerned about OSI models ;) Actually, i tried to setup a DNS over TLS server some weeks ago but while it worked like intended i had a hard time to get reasonable response times even after somewhat serious fine tuning for the session setup (i used haproxy at server side). But this was mainly caused by the fact that systemd-resolved established a new connection for each and every request and i did not find a stub resolver that did not either.

CogitoCogito 7y ago
> i was only responding to your question whether there is any difference between the two beside the port number used which there is, in a significant way. You can not point a DoH client to a DNS over TLS server or vice versa. That said there is no reason a DoH server could not be recursive or consult a traditional (probably recursive) resolver for its response. > I think the only benefit of DoH is that is unlikely to be blocked. Otherwise it feels more like a hack and at least i am not concerned about OSI models ;) Thanks for the response, but I'm still a little confused. If you live in the DoH stack you must stay there sure, but the same apparently applies to the DNS over TLS. What to you exactly is hacky about the DoH method if you don't care about the layering violations?