Okay, so, to parse this — you have a highly performant http/1 app using a fresh connection setup/teardown at 100ms per cycle?
Have you tested that same app over http/3 with tuning to reuse the underlying channel, and if so, what were your outcomes?
H1 can be pretty awesome but my core curiosity here is whether reducing the connection overhead through H3 matters or not. I am not out to disprove H1 full-page HTML as a delivery mechanism.
Have you tested that same app over http/3 with tuning to reuse the underlying channel, and if so, what were your outcomes?
H1 can be pretty awesome but my core curiosity here is whether reducing the connection overhead through H3 matters or not. I am not out to disprove H1 full-page HTML as a delivery mechanism.