A web server's primary purpose is to accept HTTP request messages, process the requests, and respond with HTTP response messages. The web server that is built into the Go net/http package is a wonderful example of a bare bones web server. The web server listens for and accepts new TCP requests, launches a new Go routine for each request, and then runs the developer-specified handler for the request in the new Go routine. The web server implementation watches for any application-level panics that occur within the code and then sends the response generated by the handler code back to the client that initiated the request.
This works due to the lightweight thread scheduling that is done within the Go runtime environment. Go routines are much cheaper to initialize and run than traditional threads and subprocesses, and are designed so that thousands of Go routines can be running at the same time. The following is a high-level diagram of how a request is processed within the Go web server:
As a recap, the Go web server will spawn and execute each handler function within a Go routine for every single request that the web server receives. The Go web server will handle the network's listen and accept life cycles, and has a Transport Layer Security (TLS) built-in if requested by the developer.