[RESEARCH] Do research on framework selected for our backend platform services.

[Aries1711]

Context: RESEARCH FRAMEWORK BACKEND API PLATFORM SERVICES

Overview

A new directive was arrived to handle our backend application services for our platform API services. Need to do research on some candidate framework below:

1. Django (python language base)
2. Nest.js (javascript language base)
3. Express.js (javascript language base)

Issues Reference

• none

Platform Targeted

• [x] Server API Application

Definition of done

• [x] Provide details comparison between 1 to 1 or 1 to another on pros and cons based on our system requirement needs
• [x] Provide details comparison between 1 to 1 or 1 to another on pros and cons based on our performance needs
• [x] Provide details comparison between 1 to 1 or 1 to another on advantage development life cycle based on our next feature (WebSocket server)
• [x] Provide details comparison on perfect matching for our Ops infrastructure.

CC: @chiefardi @coinresearchai

relevant resource :

1. (Proposed coinresearch Ops System Infrastructure )[https://github.com/coinresearch-ai/coinresearch-platform/wiki/Infrastructure-Diagram]

[Aries1711]

Key point 1 :

Enterprise-level application API: Fast performance, scalability, robustness (start from MVP)

1. Django

• Enterprise-ready: Well-established for large, complex applications (e.g., Instagram, Pinterest).
• Mature ORM and security: Strong ORM, built-in admin, and security features (CSRF, XSS protection).
• Performance: This can be slower due to Python, though Django Rest Framework (DRF) helps build robust APIs.
• Monolithic but scalable: Scaling is harder compared to microservices-based frameworks.

2. Nest.JS

• Designed for microservices: Out-of-the-box support for scalable, modular applications with microservices architecture.
• Asynchronous event-driven: Leverages Node.js's non-blocking I/O for high concurrency and performance.
• Strong TypeScript support: Ensures maintainability and robustness for enterprise-level systems.

3. Express.JS

• Lightweight and fast: Known for simplicity and fast performance for smaller services.
• Minimalistic: Express is very flexible and requires adding tools manually for features like security and routing.
• Good scalability: Suitable for scaling but often requires more setup than NestJS.

Based on this Key point and the value between the item self I considered choosing nest.js for our API Platform services cause it has strong greater performance along with its scalability and type-safe that is good for next when our team is grown (large team or enterprise) and not monolithic approach beside it has good value on Asynchronous event-driven that was modern schema this programming era. Besides nest.js was built on top of express.js but still it has the same performance and capability to be compatible with microservice tasks cause nest js was designed for the microservices ecosystem (microservices are the magic needs to our API services due to complex services we are going on).

[Aries1711]

Key poin 2 :

Performance in development (scalable for modular and feature context)

1. Django

• Monolithic approach: Not inherently modular. Scaling feature complexity requires more effort compared to modular microservices.
• Battery-included: Comes with built-in features like ORM, admin, and authentication, which may slow down large feature scaling but simplify development.

2. Nest.Js

• Modular by design: Follows a modular architecture with dependency injection, decorators, and services, making it easy to scale individual features.
• Microservice-friendly: Easier to break down into scalable services.
• TypeScript support: Great for large teams and enterprises where type safety matters.

3. Express.Js

• Minimalist but flexible: Express is lightweight, allowing developers to structure the app any way they want, but lacks inherent modularity and requires additional tools to scale in a structured way.
• Good for small services: Excellent for smaller services and rapid development but requires extra work to scale with larger projects.

Based on these key points and their value, I chose Nest.js because it outperformed the others and was the perfect match for our main objective of developing these API services (Modularity, Microservice-friendly, TypeScript Support, Not to native).

[Aries1711]

Key point 3 :

Future Features WebSocket server development

1. Django

• Not optimized for WebSocket: Django needs Django Channels for WebSocket support, but performance can be hindered due to Python’s synchronous nature.
• Complex setup: Setting up real-time features like WebSockets requires more effort compared to Node.js-based frameworks.

2. Nest.Js

• Best for WebSockets: Node.js’s event-driven architecture and non-blocking I/O make NestJS ideal for WebSocket servers and real-time applications.
• Built-in WebSocket support: NestJS has a dedicated module for WebSockets, which allows smooth real-time communication.

3. Express.Js

• Good WebSocket support: Express, combined with libraries like Socket.io, handles WebSockets well but lacks the deeper framework support that NestJS provides.
• Scalability depends on libraries: Express needs external libraries for efficient real-time communication but can handle WebSocket well with proper setup.

Based on the evaluation of these three frameworks—Django, Nest.js, and Express.js—Nest.js stands out as the best choice for our API services. Here's why:

• Django is not optimized for WebSockets and requires additional setup with Django Channels, making real-time features more complex to implement.
• Express.js offers good WebSocket support with external libraries, but lacks built-in WebSocket functionality, which can make scalability more dependent on third-party tools.
• Nest.js excels with its event-driven architecture, non-blocking I/O, and built-in WebSocket module, making it the most efficient and scalable solution for real-time applications. Given these key factors, Nest.js aligns perfectly with our needs for seamless WebSocket integration and scalable API development.

[Aries1711]

Key point 4 :

Ops Infrastructure on GCP

1. Django

• Wide compatibility: Django works well with GCP services like Google Cloud SQL, Cloud Run, App Engine, and Cloud Functions.
• Easy deployment: Supports scaling via Kubernetes but can be slower due to Python overhead in microservices architectures.

2. Nest.js

• Seamless integration: Works very well with GCP infrastructure like Google Kubernetes Engine (GKE), Cloud Functions, and Pub/Sub for event-driven architectures.
• Microservice-ready: The modularity of NestJS suits microservices and containerized deployments on GCP Cloud Run and GKE.

3. Express.js

• Highly compatible: Works well with GCP services like Cloud Run, Kubernetes, and Cloud Functions.
• Lightweight and portable: Easy to containerize and deploy but may need more setup for complex deployments compared to NestJS.

After reviewing the compatibility and deployment features of Django, Nest.js, and Express.js with Google Cloud Platform (GCP), Nest.js stands out as the best choice for our microservice architecture. Here's the breakdown:

Django offers wide compatibility with GCP services but can be slower in microservices environments due to Python's overhead and lacks the modularity required for large-scale, distributed systems. Express.js is lightweight and easy to deploy on GCP, but for more complex and scalable deployments, it requires additional configuration and setup. Nest.js provides seamless integration with GCP infrastructure such as Google Kubernetes Engine (GKE), Cloud Functions, and Pub/Sub, and its microservice-ready modularity makes it an ideal choice for scalable, containerized deployments. Given its strong alignment with GCP services and microservice needs, Nest.js is the most suitable framework for our development and deployment goals.

[Aries1711]

@chiefardi I think Nest.js is the perfect match for our API Services Platform, which will serve as the main application API and system services. After reviewing several reference boilerplates, I found one that will save us time. Here’s the link: https://github.com/brocoders/nestjs-boilerplate/tree/main. I'm already install it on my local and do dockerize the services and push to our repo here : https://github.com/coinresearch-ai/api-platform-services