How to Use DockerSlim to Reduce Image Sizes

Posted in: Technical Track
Docker's Official Logo

Learn how to test the capabilities of docker-slim on Java and Python images and containerized Spring and Flask apps.

 

If you’ve ever worked with Docker, there’s likely been at least one time when it started taking up significant storage space on your computer. For example, some of your images took a long time to download in a CI/CD pipeline. Some common approaches to this problem are to:

  • Swap out the base image for something lighter
  • Reduce the number of RUN statements in your Dockerfile
  • Remove cached package manager artifacts as part of your Dockerfile

These steps, while helpful, can take up a significant amount of time and effort. Thankfully, there are open source tools that can automatically minify an existing Docker image. Enter docker-slim.

This post aims to showcase and test out the capabilities of docker-slim for several use cases:

  • Optimizing a larger, debian based Docker image running a Java app.
  • Optimizing a smaller, alpine-based Docker image running a Python app.

I highly recommend checking out docker-slim using the link above; here’s the repository’s description:

“Don’t change anything in your Docker container image and minify it by up to 30x (and for compiled languages even more) making it secure too! (free and open source)”

I know what you’re thinking.

Surely this is too good to be true. 

Either that, or you’re wondering if this applies to heavily layered images with many RUN statements.

Let’s have a look.

Dockerfile 1: Containerized Spring (Java) app

 

First, let’s take a quick look at the Dockerfile supporting my HelloWorld Java app:

FROM openjdk:8
EXPOSE 8080
COPY . /usr/src/myapp
WORKDIR /usr/src/myapp
CMD ["./mvnw", "spring-boot:run"]

When built, it yields an image this size:

REPOSITORY   TAG       IMAGE ID       CREATED         SIZE
javatest     latest    dff8c8aa3566   7 seconds ago   520MB

That’s pretty hefty. Admittedly, this could be shaved down by using a smaller alpine-based image, but we’ll explore that in the Python example. For now, let’s see what docker-slim can do for us.

$ docker-slim build --target javatest:latest


$ docker images
REPOSITORY      TAG       IMAGE ID       CREATED         SIZE
javatest.slim   latest    76c7a799c027   6 seconds ago   214MB
javatest        latest    dff8c8aa3566   3 minutes ago   520MB

So, from running one command and without refactoring the Dockerfile manually, we went from 520MB (with what was a pretty simple Dockerfile) to 214MB (41% of its original size!).

The million-dollar question of course is: does the app the image is serving still function?

$ docker run --name javatest -P -d javatest.slim:latest
3c2148270edac2a19a7f1df1c26ab7daa29bae7243a25fe1bfc20a6261f4f9f8

$ docker ps
CONTAINER ID   IMAGE                  COMMAND                  CREATED         STATUS         PORTS                     NAMES
3c2148270eda   javatest.slim:latest   "./mvnw spring-boot:…"   7 seconds ago   Up 6 seconds   0.0.0.0:55030->8080/tcp   javatest

$ curl http://localhost:55030
<p>Hello there</p>%

Yes, it does.

Dockerfile 2: Containerized Flask (Python) app

 

So, the results from the JDK image were pretty impressive. Let’s try something more challenging and move on to our next Dockerfile:

FROM python:3.8-alpine3.13
ADD . /api
WORKDIR /api
RUN pip install -r requirements.txt
EXPOSE 80
CMD ["python", "src/main.py"]

This time, let’s use a much smaller (alpine-based) image, running a bare-bones Flask app.

Let’s build the image as-is and see what we have to work with:

REPOSITORY   TAG       IMAGE ID       CREATED             SIZE
pythontest   latest    55c25c2b963a   About an hour ago   64.1MB

The initial image is 64.1MBnot a lot to work with there. However, let’s still see what docker-slim can do for us:

$ docker-slim build --target pythontest:latest

$ docker images
REPOSITORY        TAG       IMAGE ID       CREATED             SIZE
pythontest.slim   latest    4ae57b3849c0   14 seconds ago      18.4MB
pythontest        latest    55c25c2b963a   About an hour ago   64.1MB

Starting with an already slimmed down image (64.1MB), we still see a reduction down to 18.4MB (28.7% of its original size).

Conclusion

 

Having run through the above two examples, we can say the following of docker-slim:

    1. It most certainly does work, and
    2. It greatly optimizes small images that didn’t even need it in the first place.

Again, you can (and definitely should) check out DockerSlim’s GitHub repository for a deeper dive into usage, as well as instructions on how to install it. Happy optimizing!

 

Don’t forget to sign up for more updates here.

email

Author

Want to talk with an expert? Schedule a call with our team to get the conversation started.

About the Author

Cloud Architect
Evan is a GCP certified Cloud Architect and Data Engineer located on Canada's West Coast. He brings with him years of experience working in the Real Estate Technology sector, as well as DevOps consulting. In his spare time, Evan spends his time playing video games, attending car shows, and spending time outdoors.

No comments

Leave a Reply

Your email address will not be published.