Performance testing

[kuba-moo]

Once we have bare metal HW we can work on tracking performance regressions. This will likely start as a basic set of iperf / uperf / kperf runs. We can also explore XDP testing if we can generate enough PPS.

[martinkarsten]

I like the idea of tracking performance regressions and I would be happy to contribute, if I can be useful. I would hope that a suite of
performance experiments can track performance regressions, but also serve as a reference implementation of best practices and performance benchmarks. I believe this would be quite useful for the kernel community, but also for the academic community.

[borkmann]

Would love to see sth like https://core.dpdk.org/perf-reports/ for XDP in particular to make sure we don't run into regressions.