Skip to content

Environmental Impact

Jump to bottom
erwan edited this page Mar 10, 2026 · 2 revisions

1. Context & Objective

Digital technology accounts for 4% of global greenhouse gas emissions.
As engineers, we must ensure StemHub is designed responsibly.

We analyzed our architecture through the three pillars of Eco-Conception:

  • Hosting
  • Data Transfer
  • Compute

2. Eco-Conception Analysis

A. Hosting: Low-Carbon Infrastructure

Challenge:
Cloud servers and data centers are energy-intensive.

StemHub Strategy:
We strategically select Google Cloud Regions with low carbon intensity (e.g., europe-west9 Paris - Carbon Free Energy) to host our GCS Buckets and PostgreSQL database.

Impact:
By leveraging Google's commitment to 24/7 carbon-free energy, we significantly reduce the carbon footprint of our storage and compute compared to traditional data centers.

B. Data Transfer: The "Low-Hops" Architecture

Challenge:
Transferring heavy audio files (.wav, stems) consumes massive network energy.

StemHub Strategy:
We use Streaming Processing in our Python backend. Files are streamed directly from the client through the server to Google Cloud Storage (GCS) without saving temporary copies on disk.

Impact:
This avoids extra "Disk Write/Read" cycles on the server, saving I/O-related energy. While it adds a server hop for validation (Security/Consistency), the lack of intermediate disk persistence keeps the energy footprint per GB optimized.

C. Compute: Efficient Algorithms & Client-Side Rendering

Challenge:
Server-side processing of audio visuals is CPU-heavy.

StemHub Strategy:

  • Frontend:
    We use React and Wavesurfer.js to render waveforms directly on the user's device, relieving our servers of graphical processing.

  • Backend:
    We use FastAPI with optimized libraries (struct, PyFLP_enhanced) and Alembic managed migrations to ensure efficient execution and minimal database overhead.

3. Measuring Impact

To validate our strategy, we utilize the following tools:

  • EcoIndex:
    To audit our React frontend performance and aim for a score of B or higher (70/100).

  • Google Cloud Carbon Footprint:
    To monitor and minimize the CO2eq emissions of our GCP infrastructure.

4. Conclusion: Optimization Choices (Target)

Category Optimization Choice Environmental Impact
Hosting GCP europe-west9 (Paris) 100% carbon-neutral energy (matched)
Network Streaming Uploads Zero intermediate disk I/O, optimized energy per GB
Compute Client-Side Rendering Offloads energy cost to client devices; reduces server heat/AC needs
Parsing Binary Structs (PyFLP_enhanced) Minimizes CPU cycles/electricity required for file analysis
Storage GCS Lifecycle Policies Automated cleanup of temporary files to reduce storage waste

Clone this wiki locally