Usage:

• _21_multi_thread_server [COMMAND] or cargo run [COMMAND]

Commands:

• spawn-infinite-threads Run spawning unbounded threads (default behavior)
• thread-pool [size] Run using a fixed-size thread pool of [size]
• single-thread-tokio Run using single threaded tokio runtime
• help Print this message or the help of the given subcommand(s)

Options:

• -h, --help Print help
• -V, --version Print version

Benchmarking:

• Start up a desired server in release mode with cargo run --release -- [COMMAND]
• And in a separate terminal window run wrk -t4 -c100 -d30s -s scripts/root-60-sleep-20-404-20.lua http://127.0.0.1:7878
• Above would simulate 100 users constantly sending requests for all 3 available routes (root, sleep, 404) in 3:1:1 proportion respectively for 30 seconds using 4 threads to generate load.
• To test all of the available options you can use scripts/benchmark.sh which would run all of the server implementation options and write benchmark results to files within ./benchmarks.

Things to complete in this project

• Learn how to limit the value of the size argument with clap
• Add proper error handling with anyhow
• Remove unwrap from the project.

• Do not use thread::spawn as it may panic!(), instead use thread::Builder.

• Create a separate option to deal with server requests using:
  • Fork-join model.
  • Async I/O model with tokio.
    • Multi-threaded.
    • Single-threaded. This could be done using #[tokio::main(flavor = "current_thread")].
  • Low-level event-driven I/O and mio.
• Brighten up the server, having more different routes.
• Add proper server logs, that persist for the user.
• Understand how can I write tests for this application
• Understand how can I measure the performance of each of the options to run the server with.
  • Use wrk to simulate HTTP/1.1 requests.
  • Use .lua scripts to simulate calls to all 3 endpoints the server has in somewhat mixed fashion.
  • Write a script to automate starting up the server, running the wrk load, shutting down the server for different server options and storing the result in ./benchmarks
  • Graph to visualize results.

Learning takeaways

Terminology

• Thread Pool - a group of spawned threads that are ready and waiting to handle a task. When a program has a task it needs to complete it asignes an available thread to it.
• Worker - a structure that picks up the code and runs it in its thread.