Closed CBenoit closed 1 month ago
By removing a few to_vec() calls and reusing the Bytes buffer as-is, perfomance of JMUX proxy is increased by ~62.3%.
to_vec()
Bytes
Performance was measured using iperf on local network.
iperf
JMUX proxy performance before this patch:
0.0000-10.0493 sec 8.21 GBytes 7.02 Gbits/sec
JMUX proxy performance after this patch:
0.0000-19.0245 sec 25.2 GBytes 11.4 Gbits/sec
This is still 78.6% slower than jetsocat regular TCP port forwarding.
By removing a few
to_vec()calls and reusing theBytesbuffer as-is, perfomance of JMUX proxy is increased by ~62.3%.Performance was measured using
iperfon local network.JMUX proxy performance before this patch:
JMUX proxy performance after this patch:
This is still 78.6% slower than jetsocat regular TCP port forwarding.