Marek, At 2016-07-11 22:26:00 -0500 Marek Vavruša <mvavr...@cloudflare.com> wrote: > > You get queueing for free, but not pipelining and out-of-order > responses, that has to be defined. > The draft mentions this, but I think at this point it should just > depend on HTTP/2, > as it's the only way to get decent performance from resolvers but high > variance in response time, > and request semantics in both is different and needs to be defined, > i.e. what happens when client cancels > in-flight query, what happens when server resolver pushes response.
There is another option, which is what HTTP developers have done for years, which is keep multiple HTTP connections open. Not ideal, but better than blocking all lookups on a slow resolution. :) I recognize that HTTP/2 is definitely a better option because of out-of-order replies, but I worry about requiring it. It's still quite new and language support may be spotty. But I guess given it's popularity this shouldn't be a huge problem, so maybe that is a reasonable recommendation. I'm also not sure if we need to define all of the semantics in full. Rather, I am worried about over-specifying semantics. Perhaps we can do that as "suggested semantics"? Cheers, -- Shane
pgpWh3QIOWW0T.pgp
Description: OpenPGP digital signature
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
