From Docker Compose to AWS EKS: Building a Real SRE Operations Platform

From Docker Compose to AWS EKS — Building a Real SRE Operations Platform

I recently built and deployed an SRE Operations Platform on AWS EKS — starting from Docker Compose on my laptop and ending with a production Kubernetes cluster running 3 replicas behind an AWS ALB.

The goal was simple:
Build something SRE teams would actually use daily — not another todo app running on Kubernetes.

The platform currently handles 5 core SRE workflows:

• Operations dashboard with live P1 count, SLO breach status, and current on-call engineer
• Incident tracker for managing P1/P2/P3 incidents by severity
• SLO tracking with availability and latency breach detection
• Runbook management for incident response procedures
• On-call rotation management

Deployment flow looked like this:

• Built PHP app + MySQL images locally using Docker Compose
• Pushed both images to private Amazon ECR repositories
• Wrote Kubernetes manifests and deployed everything on Amazon EKS
• Running 3 PHP application replicas + MySQL inside the cluster
• MySQL exposed internally using ClusterIP
• Application exposed externally through a LoadBalancer service
• AWS ALB automatically routing traffic across all replicas

The biggest learning wasn’t writing YAML or deploying pods.
It was understanding the difference between:

“Works on localhost”
vs
“Works in a real Kubernetes cluster.”

Container networking, Kubernetes service discovery, ECR image pulls, pod health checks, debugging failed deployments — almost nothing worked perfectly on the first attempt.

And honestly, that’s where most of the learning happened.

Tech Stack:
Docker | Amazon ECR | Amazon EKS | Kubernetes | PHP 8 | MySQL 8 | AWS ALB

Github Link : Github Repo Link

Post by : Satvik Bodke