me/hyper-procrastination

Because there are so many things I should be doing instead

Description

Compares speed of different server-side tech in a following test scenario: request a large JSON file 200 times concurrently, then reduce each response to some value.

Each push to master triggers a comparison test that runs on a dedicated cloud server (2 cpus, 8Gb ram). The results are then published here:

Try this at home

Requires docker-compose and I'd say at least 4Gb of ram (may need docker settings change). Then:

docker-compose run go # node ruby

Add more tech

Rules

• all files go in a subfolder named after the new tech
• test JSON - cityloads.json - should be requested over HTTP from the Nginx container bundled in this project
• docker-compose start tech should run the test
• local development environment should not require presence of tech
• no caching
• the results should be stored in results.json and mapped as volume in docker compose
• results.json should ideally look like this: {"request":4,"parse":3,"process":22,"total":42}, but at the very list contain total: time in seconds.

You may find docker-compose start nginx useful in development. It starts nginx and binds it onto the host port 8889.

Test spec

Each thread/worker/whatnot should independently request cityloads.json and print out the number of unique properties.FROM_ST values from features array.