🐳 How I Reduced My Docker Image from ~1GB to ~200MB for a Python App

While containerizing a Python web application using Docker, I encountered a common problem: my image size had ballooned to almost 1 GB.

Before diving into optimization, let’s talk about why large Docker images are a real problem, especially when you're:

• Building images in CI/CD locally

• Deploying to cloud environments

• Working remotely on slow networks, like mobile hotspots

A 1 GB+ image means:

• Longer build and push times

• Slower deployments

• Wasted bandwidth and computing resources

• Higher storage and transfer costs

After some cleanup and optimization, I reduced the image size to ~200MB — a massive improvement. Here’s exactly how I did it.

---

✅ Step 0: Create a .dockerignore File

Before we optimize anything in the Dockerfile, the first step is to add a .dockerignore file.

Why? Because Docker will copy everything in your project directory into the image unless you tell it not to, including things like .git, virtual environments, build artifacts, and editor configs.

Here’s a basic example of a .dockerignore:

Edit__pycache__/
*.pyc
*.pyo
*.pyd
*.db
*.sqlite3
*.log
*.env
.env*
venv/
.git/
.idea/
*.egg-info/
dist/
build/

This drastically reduces build context size and ensures cleaner images from the start.

---

⚙️ Step 1: Use a Slim Base Image

In my original Dockerfile, I used the full python:3.10.11 image. It’s convenient but includes a lot of unnecessary tools.

I switched to the slim version:

FROM python:3.10.11-slim

This one change cut down hundreds of megabytes from the image. But using slim it also means explicitly installing build dependencies for some Python packages.

---

🧰 Step 2: Install Build Dependencies in Stage One

Some Python packages like psycopg2, cryptography, or numpy require native libraries or compilation. So we install system-level build dependencies in a build stage:

# Stage 1: Build dependencies and install packages
FROM python:3.10.11-slim as builder

WORKDIR /app

# Install build dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
    build-essential \
    gcc \
    libpq-dev \
    && rm -rf /var/lib/apt/lists/*

# Copy requirements file and install packages into a separate directory
COPY requirements.txt .
RUN pip install --upgrade pip \
 && pip install --prefix=/install-packages --no-cache-dir -r requirements.txt

We're installing all Python dependencies to /install-packages later copy just the results into the final image, keeping it lean.

---

🏗️ Step 3: Multi-Stage Build for Minimal Final Image

Next, we create a second stage where we build the final runtime environment. This is what runs your app in production:

# Stage 2: Final runtime image
FROM python:3.10.11-slim

# Environment settings
ENV PYTHONDONTWRITEBYTECODE=1
ENV PYTHONUNBUFFERED=1

WORKDIR /app

# Install runtime dependency required by psycopg2
RUN apt-get update && apt-get install -y --no-install-recommends \
    libpq5 \
    && rm -rf /var/lib/apt/lists/*

# Copy only the necessary files
COPY --from=builder /install-packages /usr/local
COPY . /app/

# Run using Gunicorn
CMD ["gunicorn", "--config", "gunicorn_config.py", "wsgi"]

Key benefits of multi-stage builds:

• No build tools in the final image

• No source files, tests, or unused code

• Only runtime dependencies are copied over

---

📉 Final Result

After applying these changes, here's the size comparison:

💡 That's an 80% reduction in image size!

---

🔁 Recap

Here’s a quick recap of what we did to reduce the Docker image size:

1. ✅ Added a .dockerignore file to shrink the build context

2. ⚙️ Switched to python:3.10.11-slim to avoid unnecessary bloat

3. 🧰 Installed build tools in a temporary stage and removed them later

4. 🏗️ Used multi-stage builds to produce a clean, minimal final image

🔵 Original Dockerfile (Unoptimized)

FROM python:3.10.11

ENV PYTHONDONTWRITEBYTECODE 1
ENV PYTHONUNBUFFERED 1

RUN mkdir /app
WORKDIR /app
COPY . /app/

RUN pip install --no-cache-dir -r requirements.txt

CMD ["gunicorn", "--config", "gunicorn_config.py", "wsgi"]

✅ Improved Dockerfile (Optimized)

# Stage 1: Build stage
FROM python:3.10.11-slim as builder

WORKDIR /app

# Install build dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
    build-essential \
    gcc \
    libpq-dev \
    && rm -rf /var/lib/apt/lists/*

# Copy requirements and install packages into a separate folder
COPY requirements.txt .
RUN pip install --upgrade pip \
 && pip install --prefix=/install-packages --no-cache-dir -r requirements.txt

# Stage 2: Final runtime image
FROM python:3.10.11-slim

# Environment settings
ENV PYTHONDONTWRITEBYTECODE=1
ENV PYTHONUNBUFFERED=1

WORKDIR /app

# Install runtime dependency (libpq5 needed by psycopg2)
RUN apt-get update && apt-get install -y --no-install-recommends \
    libpq5 \
    && rm -rf /var/lib/apt/lists/*

# Copy installed Python packages from builder stage
COPY --from=builder /install-packages /usr/local

# Copy application code
COPY . /app/

# Start the application
CMD ["gunicorn", "--config", "gunicorn_config.py", "wsgi"]

Happy Learning 😊

#Docker #Python #DevOps #Dockerfile #WebDevelopment #PerformanceOptimization #Mutlistage build