E-commerce API

This project is a mini E-commerce API built with Node.js, TypeScript, Apollo Server, Prisma, PostgreSQL, Redis, and Elasticsearch. The architecture and design decisions focus on scalability, performance, and ease of development.

Architecture and Design Decisions
Setup and Usage
Future Improvements

Architecture and Design Decisions

Using pnpm

I've chose pnpm for its superior performance of handling dependencies. It creates a single version of all packages in the node_modules directory, reducing disk space to improve speed.

My reference: pnpm vs npm

Using Prisma

Prisma is used for database access ORM. It provides a type-safe database client, it also helps in managing database migrations and schema changes seamlessly.

Ioredis

Ioredis is used as the Redis client. It has a rich API that covers all Redis commands, making it easy to integrate Redis functionalities into the application.

Type-GraphQL

Type-GraphQL is used for creating GraphQL schemas, types, and resolvers, avoiding the need to manually create types and interfaces.

Class-validator

Class-validator is used for input validation. It provides a declarative way to validate data, ensuring the integrity and correctness of the input.

Prettier + ESLint for Code Quality

Prettier and ESLint maintain code quality and consistency across the project. ESLint is a static code analysis tool that helps identify and fix problems in the code.

Apollo Standalone Server

Apollo Server is used in standalone mode for quick server startup without the need for HTTP routes, simplifying setup and improving performance.

Pino Log

Pino is used for logging. It is a fast, lightweight logging library providing a consistent and efficient way to log application events.

ts-node-dev for Development

Ts-node-dev is used for running the application in development mode. It provides fast and efficient TypeScript transpilation and hot-reloading.

Storing Only Name and Description in Elasticsearch

Store only the name and description fields in Elasticsearch for efficient search indexing, while the full product details are stored in PostgreSQL.

Avoid circular dependencies on graphql types

I've made use of loadash package to use the function omit, to create an extended class for Product.

Git atomic commits

You can check all the development flow through the atomic commits here.

Setup and Usage

Prerequisites

Node.js and npm
Docker and Docker Compose

Environment Variables

Create a .env file based on the .env.sample provided in the project. Configure the necessary environment variables such as DATABASE_URL, REDIS_HOST, REDIS_PORT, and ELASTICSEARCH_URL.

Docker Instructions

Build the Docker Image:
```
docker-compose build
```
Run the Application:
```
docker-compose up
```

This command will start the Apollo Server application, PostgreSQL, Redis, and Elasticsearch, ensuring they are all running together.

Suggestion for future improvements

Include Husky for Pre-commit Hooks

Add Husky to run linting and formatting checks before committing code. This ensures code quality and consistency.

Use Redis Password in Production

Configure Redis to use a password in production for enhanced security.

Manage Cache Timing

Implement proper cache timing management to ensure data is up-to-date while reducing the load on the database and Elasticsearch (currently is set up as 30 minutes)

Add esbuild to the Project

Esbuild is one of the best JavaScript bundlers and minifiers. Adding it to the project can improve build times and optimize the output.

Migrate to Apollo Server Express

Consider migrating to Apollo Server Express if there is a need to combine REST and GraphQL endpoints.

Omit Stack Trace from Apollo in Production

Ensure that stack traces are omitted in production to avoid leaking sensitive information. This should be the default behavior, but it's worth double-checking.

Enable Security and Use SSL Certificates in Elasticsearch

Enable security features in Elasticsearch and use SSL certificates to ensure secure communication and data integrity.

Kubernetes for Orchestration

I strong suggest using ks8 for the container orchestration, that provides automated deployment, scaling and management of container applications, ensuring the scalability and high availability.

Sentry

Implement sentry for api monitoring, metrics and stack traces management.

ecommerce-api