Multi-cloud Cluster

One app. Two clouds. GitOps end to end.

A project that runs a single application across AWS EKS and Azure AKS with shared observability and traffic control.

Business Challenge & Solution

Relying on a single cloud provider is a hidden business risk, because:

How do businesses protect themselves from cloud dependency?

This project solves the problem by:

Benefits

  • Resilience Survive a regional or provider-level outage without service interruption.
  • Cost control Combine savings plans across providers and avoid lock-in pricing.
  • Flexibility Place each workload on the cloud where it performs best.

Trade-offs

  • More moving parts Multiple control planes and networking stacks to manage across providers.
  • Consistency overhead Keeping both environments in sync requires extra tooling and discipline.

Feature: Multi-cloud Deploy

Architecture diagram

Key components

How it works

  1. Register the clusters AWS EKS and Azure AKS are registered in ArgoCD, each labelled with its cloud provider
  2. One config, two targets A single ArgoCD ApplicationSet reads those labels and deploys to every matching cluster automatically
  3. Cloud-specific settings Each cluster receives its own Helm values, so cloud differences are handled without duplicating code
ArgoCD app list

Application list: prefixed with cluster name

ArgoCD app status

Application: Healthy and Synced

Feature: Traffic Control

Cloudflare Load Balancer UI

Cloudflare UI

Test script

Test script

Test result

21:78 matches the load balancing configuration (2:8)

Feature: Cloud-agnostic Monitoring

Both clusters ship metrics and logs to a single Grafana Cloud workspace, making monitoring cloud-agnostic.

Cluster overview

Cluster Overview

Cluster list

Cluster list

Cluster EKS

Cluster: eks