This project is an ongoing work to monitor the performance of QUIC implementations. Currently the repository compares the following QUIC implementations : quicly, quiche, msquic, picoquic, picoquic-dpdk. To get a performance reference the repository also contains picotls. This project is an ongoing work to monitor the evolution of QUIC stacks, in the futur other stacks will be added.
This project relies on NPF to build the different repositories, launch the measurements and graph the results. The build process was not tested on many environments and may need modifications to fit your setup. The scripts run_npf.py can be used to launch a comparison between selected stacks.
Example :
#launches a comparson between picoquic and msquic
python3 npf_run.py --stacks picoquic msquic --client client --server server
the cluster repository contains configs to connect to node through SSH. By passing --client=client and --server=server to run_npf.py those files will be used. It is also possible to specified a machine to use directly
Example :
--client=user01@server01.network.edu
picoquc_dpdk relies on the DPDK library. When running the scripts, it is expected that the library is installed and added to LD_LIBRARY_PATH.