Architecture

Initial Architecture

Initial development will be on a core domain model, persistence, and a RESTful web service to access models.

I selected the following

Component	Rationale	Drawbacks or Negatives
Python for backend	WS tools are more mature, scalable, and secure than Javascript frameworks. Python is simpler than Java REST frameworks with more deployment options.	Async behavior rather complex in Python.
FastAPI for Web Service	Much faster than Django RestFramework, good async support, excellent documentation, large community.	Less consistent and elegant syntax for WS endpoints than Django or Java REST API; requires separate "schema" classes for models that result in some duplicate code.
Postgres DB; Sqlite for testing	Free & well supported. RDBMS is appropriate for structured data in this app. Both have async interfaces.	-

The backend services can be implemented using standard synchronous Python code or asynchronous code.

Asynchronous code has higher throughput -- up to 10X faster for I/O bound applications. I expect the web service and database will be running on a separate cloud services or different container from the web service, hence there may be network latency.

The drawbacks of async code are complexity and less support.

• PostgreSQL 2.0 mainstreams support for async mode, but it requires separate packages and is much less widely used. Some operations don't work.
• Sqlite support for async is by a 3rd party package that still has not reached version 1.0
• Testing async code is more complex and also requires extra packages.
• My own limited understanding of async code in Python may result in code defects.

Design Guidelines

As usual, strive for a component-based design with loose coupling. We should be able to replace the framework or library used in a component with low (or no) change needed in the rest of the code.

Architecture Suggested by AI Engines

I asked DeepSeek and ChatGPT this question:

I want to build a simple application to record residential data such as electric power and water usage. It will have both a web-based and mobile interface.
The major features are:

1. user can input home data with a timestamp,
2. user can view all data for his home.
3. There will be only a few users, and each user may create more than one data source for his home.

What software architecture(s) would be best suited for this?

Later, I rephrased the question with a reference to the wiki page for Requirements instead of writing requirements in the question.

The answers were long and I asked follow-up questions, so I put them in separate files.

I asked follow-up questions, such as:

• how to accommodate a LINE interface using a LINE Mini-App
• comparison of something like Railway (Python & FastAPI) + Supabase (Postgres) with Firebase or Cloudflare Workers + D1.
• in a separate chat I asked for comparison of using Python (FastAPI or Django DRF) and Javascript/Typescript (Express) in terms of performance, development effort, and security. For security and development effort, FastAPI and Python look better.

Deepseek Suggested Architecture
ChatGPT Suggested Architecture

Alternative Stacks for Development and Deployment

I compared Javascript and Python for the backend web service, and a few REST frameworks for each language.

FastAPI and Firebase Comparison