GopherCon UK -2023

Scaling Coffee with Goroutines

A RESTful API to serves coffees, powered by goroutines.

This project was done during GopherCON UK 2023, during Sadie's Freeman (@sadief) workshop.

Requirements

it has to be scalable and fast!

Implementation steps:

Server coffee without concurrency
Server more than 1 coffee
Add concurrency: use goroutines
Add WaitGroup: Why? to make sure that all goroutines we start will also finish
Add more goroutines: so we can run each function concurrent

How to run it?

in root folder run make run to start the docker container
navigate to http://localhost:8080/serve-coffee/50000 to hit the endpoint and try making MANY coffee. Checkout the logs to see the time it took to make them. In my case, 50000 coffee took 13.390567749s
in root folder run make down to stop the docker container and remove image.

What's next?

Scale vertically by adding more pods

Testing Results

Configuration 1

resources:
            limits:
              cpu: 1
              memory: 3092M
            requests:
              cpu: 1
              memory: 1024M

// Took 5.057038683s to serve coffee, customer no: 50000

Configuration 2

          resources:
            limits:
              cpu: 4
              memory: 3092M
            requests:
              cpu: 2
              memory: 1024M

Took 9.362377223s to serve coffee, customer no: 50000

Why? 🤔

Resources

https://github.com/sadief/gophercon-2023-slides-code/tree/main

oanaOM/coffee-shop