Most of them? 500k per minute is around 8k reqs/sec. $6 will get you 2 vCPU, so each CPU only needs to handle 4k requests per second.

I'd at least expect that to be possible in Node, and would be somewhat surprised if a Go or Rust framework didn't hit that mark.

I think 4k requests per second is quite ambitious even for native languages or garbage collected languages like Go.

Hello world will work, but once parsing, validation, and database queries get involved, you'll need some beefier hardware to get those numbers.

I can't get the cloud benchmarks for 2022 to work (a dedicated PowerEdge is the standard physical benchmark and that's not very realistic here), but according to these benchmarks: https://www.techempower.com/benchmarks/#section=data-r20&hw=... you're going to need some very optimized code to get close to 8k reqs/s.

OK, 10 minutes then ;)

My point was that the article quotes orders per month, while other systems are quoting requests per second. I found it funny that they tried to mask the low speed by using big numbers.