Rackspace bare metal to AWS

The company

EVgo runs one of the largest public fast-charging networks for electric vehicles in the US. Drivers depend on the apps working at the moment they are standing at a charger.

The problem

EVgo ran three applications — two Node.js backends and a React frontend — on Rackspace bare-metal servers, deployed by hand. They wanted to consolidate on AWS, reduce cost, and stop deploying manually.

The constraints were the interesting part: a tight deadline, no changes to the core application architecture or code, and minimal downtime, with the cutover confined to a low-usage window.

That combination rules out re-architecting. It calls for lift-and-reshape — change the platform, leave the application alone.

What I did

Discovery first. I mapped what the three applications actually depended on: databases, third-party APIs, the implicit assumptions bare-metal deployment had allowed them to accumulate. This is the step that decides whether a migration window holds.

Containers. I improved the existing Dockerfiles for the Node.js builds, and built a Bitbucket Pipelines CI workflow to build, test and push images to ECR.

Kubernetes. I wrote the manifests — Deployment, Service and Ingress — and a Bitbucket Pipelines CD workflow to deploy to EKS.

Frontend. I provisioned the S3 bucket and CloudFront distribution, moving the React app off a server that had no business serving static files.

Cutover. I orchestrated the traffic migration inside the scheduled maintenance window.

Result

All workloads consolidated into one AWS environment, infrastructure cost down significantly, and the Rackspace footprint decommissioned. Deployment went from a manual procedure to an automated pipeline, which is what turned the migration from an event into a new normal.