I recently heard about DOH (https://hacks.mozilla.org/2018/05/a-cartoon-intro-to-dns-over-https/ …) and was wondering whether I'm the only one shocked by the design. For instance, @jedisct1 what's your opinion about that? @bortzmeyer you look pleased; are you?
-
-
Replying to @aifsair
I think it’s great and extremely promising. DNS has traditionally been a bottleneck for what CDNs can do to reduce latency. DoH can remove that bottleneck.
1 reply 0 retweets 0 likes -
Replying to @jedisct1
From an early reading, I dislike the mandatory TCP + TLS + http boilerplate as opposed to packed datas, inside UDP datagram. Should it even be called DNS? Plus, that's just tunneling DNS inside another protocol.
2 replies 0 retweets 0 likes -
At least QUIC should remove some of this overhead. Still the same underlying idea about universal HTTP though.
3 replies 0 retweets 0 likes -
Replying to @aifsair
The overhead is negligible, as HTTP/2 provides multiplexing, prioritization, etc. similar than UDP, except better implemented. And it opens new possibilities such as https://tools.ietf.org/html/draft-nottingham-doh-digests-00 …
1 reply 0 retweets 0 likes -
Replying to @jedisct1
I agree on that part for http/2. Didn't get the requirement cause only read "dns over https" though.
2 replies 0 retweets 0 likes
DNSCrypt remains a faster and simpler protocol. But including DNS in HTTP/2 flows is not a bad idea. Virtually all DNS queries are made just to resolve names found in data downloaded using HTTP. Using distinct transport protocols doesn’t really make sense, only slows things down.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.