System Design Capstone

Backend for Questions and Answers API

Goal

Support an existing retail web application with a modernized system able to withstand the web scale traffic loads.

Development Process

1. Research and choose the most suitable database for storing questions and answers datasets: PostgreSQL
2. Design ETL process and seed 15 million+ data into PostgreSQL databse
3. Design and optimize read/write queries to interact with database
4. Deploy server and database on AWS EC2
5. Repeat stress test in production and optimize the system

Optimizations made to solve database and server bottlenecks

• Create B-tree indexes on tables in PostgreSQL database
• Apply NGINX load balancer to achieve horizontal scaling with 6 AWS EC2 servers (load-balancing method: least-connected)
• Enable node caching to improve server performance

Outcome

Achieved 3000 RPS under 20ms latency by load balancing 6 server instances with node caching

Tech Stack

• PostgreSQL
• Node.js
• Express.js
• node-cache
• AWS EC2
• NGINX
• Loader.io
• Frisby

Installation and Setup

To download a copy of this project to your local machine:

$ git clone repo_url

Runs the app in the development mode.

$ npm install

Set up a .env file with the following

• DB_PORT
• SERVER_PORT
• USER
• PASSWORD

$ npm run start