Question about tools

[kckeiks]

What set of tools did you use to obtain the data for your graphs, like the one below? I haven't been able to get data for "traffic forwarded".

Also, what tool did you use to simulate this flood attack where the pps rate increases? I would like to run some tests, do some experiments and have data to compare. Thank you for making your solution and code open source! It is very educational.

[lmb]

@sauercrowd produced these, maybe he still has the source code. Note that the implementation has evolved a lot since those graphs were made, so ymmv.

[kckeiks]

@lmb what profiling tool are you using? You mentioned here #20 that you could tell the % usage of bpf_ktime_get_ns.

[sauercrowd]

Hey @kckeiks, for these tests we generated artifical packets and used pre-determined timestamps rather than relying bpf_ktime_get_ns() to have full control over the simulation. This might be a good reference. On the userspace side we used ebpfs Test function to pass the artifical data into.

We generated the actual test-data ourselves without any additional tooling by adjusting packet timestamps appropriately to match our target pps.

[lmb]

what profiling tool are you using?

Nothing fancy, I just observed the run time when faking time vs when using bpf_ktime_get_ns.