Launching Option Calculator
Built calculatemyoptions.click website. It’s entirely hosted on AWS. It’s a SPA where static files are on S3 bucket and are served with CloudFront. Backend is R code hosted on Lambda. All of the infra is created/updated with AWS CDK.
In order to release this project, I had to figure out how to host R inside of a container and serve it with AWS Lambda. I’ve already done something similar in Running R on AWS Lambda, so I could re-use parts of learning from there and build on top of it.
Challenges
There were a couple of challenges that I encountered when working on this project:
R Libraries and Docker Image Size
Not all R libraries were available for AWS Lambda image, so I had to compile a couple of them from source code. When compiling, too many intermediate artifacts were created which put the final image over 10GB (Docker images hosted on AWS Lambda have a limit of 10GB [1]).
I reduced the size of the Lambda container by using multi-stage Docker build process and copying only compiled binaries into a final AWS Lambda image. I was able to go from 11GB to around 4GB, and I could run R container with all libs on AWS Lambda (yay).
Frontend Development
The second challenge was the frontend since I’ve never done it before. Luckily ChatGPT helped me setup the React template that I could then modify and shape.
Also, CloudFront was a bit tricky to configure, specifically for configuring routes to Lambda function and making sure that the SPA could talk to Lambda and work across Firefox and Chrome.
Testing and Release
After parts of the whole project had been configured, I did a couple of runs of integration testing and fixing. Once I checked that the skeleton and parts work together, I did a mini release on LinkedIn, to see what people say and if I can catch any errors with real traffic.
Takeaways
Overall, it was a fun learning experience, and now I have deployment templates that I can leverage for future projects as well as knowledge about how website hosting on AWS is done.