• The first server is a simple Node.js project with a /simple route that returns a JSON message. 16:36 16:54
  • The uploader stresses that the language is not the main idea at this point; the same route shape could exist in Python, Go, Rust, Java Spring, or another backend stack. 17:11
  • The local server runs on port 3001, connects to Redis and PostgreSQL, and returns message: hi from /simple. 17:33 17:56
  • One request means the client sends bytes to the server, receives a 200 response plus the body, and counts that complete round trip. 17:56 18:25
  • AutoCannon is installed with npm i -g autocannon and is used to simulate thousands or millions of requests against the API. 18:45
  • The first local benchmark uses 20 connections, 20 seconds, pipelining of 2, and a GET request to /simple. 19:17
  • The first result is about 18,000 requests per second, 300,000 total requests over the run, no errors, and about 90 MB transferred. 20:04

What this section teaches

  • Before scaling anything, define what counts as a request and start with the smallest measurable route. 18:25
  • Benchmark outputs are only useful if errors, duration, total request count, and transferred bytes are read together. 20:04