Blazingly fast UDP server using

[@kylecroarkin1371]

Really fun to watch and informative. Thank you a lot for publishing this.

[@cheebadigga4092]

This is aaaaaaaaaawesome! Thanks!

[@dragoons_net]

Your content is incredible. Thanks a lot!

[@nandita.nambiar]

Quite informative, thank you!

[@viewtifulstart]

are you sure you're reading the flamegraph right? I'm pretty sure right at the top is not the tokio scheduler.

[@uquantum]

So informative and like your relaxed style that keeps us engaged as you lead us through your various optimizations. To your point to massive scaling possibilities at the end of the video about if clients could also use XDP...would a workaround to client-side XDP be webAssembly+eBPF ? Thank you so much Sandeep!

[@_sandeepnambiar]

client side xdp exists! its called af-xdp! you can use xdp to receive AND send. whole programs living in xdp land. would you like a video on that topic?

[@zeroZshadow]

Very cool!

[@vaibhavrai1796]

much informative :)

[@techsyndrome3291]

yay finally !

[@irlshrek]

fun!

[@vcankeklik]

Why did you prefer tokio::select! instead of spawn?

[@_sandeepnambiar]

Good question! I have a branch somewhere that does spawn as well. I think I lost a few points in performance immediately. So I reverted that. Would love a comparison between spawn vs select that is more concrete. I already began testing with monoio instead of going down that route.

[@vcankeklik]

@@_sandeepnambiar I see. So can we say that it is better to rely on metrics when making such decision?

[@_sandeepnambiar]

@@vcankeklik I think so. My findings are contrary to what I would have assumed if I hadn't done the tests.

[@vcankeklik]

@@_sandeepnambiar thanks!

[@nevgeniev]

well, it's not an impossible at all guys :)

[@glennfiedler6236]

Hired