Turning Chaos into Control with Cloudflare

This talk followed the move from scattered infrastructure to a more consistent way of accessing and protecting services. I described the problems created by Node.js and Python services spread across Cloud Functions, Cloud Run, Kubernetes, and Firebase, with multiple URLs and uneven access controls. I then worked through the foundations of the solution: a CDN, a web application firewall, authentication, authorisation, single sign-on, and a Zero Trust approach.

I showed how bringing APIs, events, and interfaces behind Cloudflare and a shared domain gave us a common entry point, with access policies spanning development through production. I also discussed the practical trade-offs we encountered, including login delays, access costs, domain configuration, and keeping end-to-end tests working behind Zero Trust. The session connected the architecture to the operational details that determine whether a security change works for the people using it.