On Fri, Sep 22, 2017 at 12:24 PM, Warren Kumari <warren@xxxxxxxxxx> wrote:
Yup -- RFC7766, S6.2.1.1 says you should do this, as does RFC7858
("Specification for DNS over Transport Layer Security (TLS)").
double yup. https >=h2 offers a couple, really minor in the grand scheme of things, similar advantages beyond that too. One is a rather seamless (no additional specification necessary) upgrade to httpOverQuic when that happens which will help with the tcp level in-order suboptimality.. the other is multiplexing within any partial response (as opposed to just reordering whole responses) that is blocking the wire (slow to generate additional records, or big things like axfr..). They're just nice to haves but they should help with the tail cases.
DPRIVE solves much of this, but one weakness is that it is still
clearly DNS traffic.
yep.