Re:A Pure HTTP/3 Alternative to MQTT-over-QUIC in Resource-C...
by belter on Sunday June 27, 2021 @09:10AM
(#27651292)
The authors do mention in their findings that:
"...While QUIC has been proven by different researchers to be more advantageous than TCP for MQTT traffic, our hypothesis was that further gains may be found through a pure H3 solution. The reasoning behind this hypothesis was rooted in QUIC’s original design intent and various optimizations specifically for carrying HTTP traffic. This turned out to be the case for performance, but posed a trade-off for resource consumption.
... Performance indicators were investigated in
addition to network and device level overhead.
MQTT-over-QUIC was found to put marginally less strain
over the network than H3, but H3 offered a key performance
savings of 1-RTT in the time to first data frame..."
"...While QUIC has been proven by different researchers to be more advantageous than TCP for MQTT traffic, our hypothesis was that further gains may be found through a pure H3 solution. The reasoning behind this hypothesis was rooted in QUIC’s original design intent and various optimizations specifically for carrying HTTP traffic. This turned out to be the case for performance, but posed a trade-off for resource consumption.
... Performance indicators were investigated in addition to network and device level overhead. MQTT-over-QUIC was found to put marginally less strain over the network than H3, but H3 offered a key performance savings of 1-RTT in the time to first data frame..."