Decide tech stack

[joseporiolayats]

In order to fulfill the requirements, as specified in the documentation problem, let's take some considerations for fulfilling each part of the problem.

In short:

1. Python
2. Modern
3. Reliable
4. Permissive license
5. Fast to iterate and easy to mantain
6. Flexible enough to open up to new reqs

The components to be chosen are:

1. Database
2. Api serving backend
3. Web-api REST-ful backend
4. Simple frontend

[joseporiolayats]

Databases

It has to be free, opensource, deployable, secure, cloud-based, modern, fast. The candidates are:

• SQLite
• MSSQL
• MySQL
• MongoDB
• Postgres

All of them can be deployed on the cloud. SQLite is the least functional, so it can be limiting. MSSQL is based on proprietary software and tied up to Microsoft. A good option but not the most modern. MySQL is battle-tested and can be deployed anywhere. Limitations are on not being able to scale horizontally. MongoDB is modern, designed for web-apps and can scale and do some compute inside. Good candidate. Postgres is very functional but also limited in scaling as MySQL.

I'm choosing MongoDB.

There is a service called MongoDB Atlas which offers free database instances for trying out the capabilities. I will be using it. It's deployed in AWS Frankfurt servers.

[joseporiolayats]

Server

As it will be coded in python, the server is going to be deployed also in python. It has to be able to handle lots of requests, so better be compatible with ASYNC methods, and for the rapid iteration of the product it has to be easily configurable. Candidates:

• Flask. Easy but slow.
• Apache HTTP. Battle tested but not modern.
• Gunicorn. Functional and easy but not able to handle many requests.
• Uvicorn. Async version of gunicorn. A winner.

I'm choosing uvicorn as web server.

[joseporiolayats]

WebAPI RESTful backend

It has to be able to work async and easy and fast to iterate. As it will be deployed in production setting, it has to be battle-tested.

Candidates:

• Flask. Very easy to configure but slow and limited to python.
• Django: Hard to configure although it's all python
• FastAPI: It's fast, can interconnect with many frontends and works async.

I'm choosing FastAPI for the backend.

[joseporiolayats]

Tech Stack

The tech stack I've selected is as follows:

• Language: Python
• Database: MongoDB
• Database hosting: MongoDB Atlas deployed on AWS Frankfurt servers.
• Python Web Server: Uvicorn
• API framework: FastAPI

This stack is usually known as FARM stack (acronym for FastApi-React-Mongodb)