This probably has a very simple answer, but I always wonder how the provide load on these sorts of tests. Can you get by with 2-4 other servers with 400Gb/s links and just tons and tons of simulated IPs/ports to activate LACP balancing? Because you probably want to simulate simultaneous clients that stream at varying rates, probably in the range of 0.3 - 10 Mbps, which means hundreds of thousands of clients to saturate at 800 Gbps, right?
Just an interesting observation I had about this once when I noticed that kernel quic implementations weren't very fast.
KTLS is mostly useful if paired with sendfile (I'm ignoring io_uring because I'm not as up to date on that). Otherwise you have to context switch back to userspace constantly.
Assuming the files are encrypted anyway for DRM reasons: why should static content like movies be TLSed? I know I know, "TLS all the things", but it sounds like a high cost at Netflix scale.
Nice seeing BSD s getting some use.