Re:A Pure HTTP/3 Alternative to MQTT-over-QUIC in Resource-C...
by rurban on Sunday June 27, 2021 @09:36AM
(#27651464)
Exactly. I'm always sceptical of protocols where the header size is 3x bigger than my body. I just implemented MQTT via NB-IoT, and I had to remove the subscribe receive part.
My other custom protocol fits into a single TCP frame, which fits into about 5 radio frames, I think. Of course I compress the hell out of it, and for MQTT clients I just put a small MQTT bridge on the host translating the body. The battery should last 10 years afterall, and TCP costs 95% of all power. The less packets, the longer it lasts.
With http/3 I would give it 2 years max. And much more dropped packets. I'm also pretty sure that I don't have the ram for http/3
My other custom protocol fits into a single TCP frame, which fits into about 5 radio frames, I think. Of course I compress the hell out of it, and for MQTT clients I just put a small MQTT bridge on the host translating the body. The battery should last 10 years afterall, and TCP costs 95% of all power. The less packets, the longer it lasts.
With http/3 I would give it 2 years max. And much more dropped packets. I'm also pretty sure that I don't have the ram for http/3