Multi-Cloud Kubernetes Deployment

🎯 Project Overview

This project demonstrates a production-ready, multi-cloud Kubernetes deployment strategy that leverages the managed Kubernetes services from AWS (EKS), Azure (AKS), and Google Cloud (GKE). By deploying the same application across multiple cloud providers, we achieve vendor independence, increased reliability, and the flexibility to optimize costs and performance.

The architecture implements best practices including infrastructure as code (Terraform), GitOps workflows (ArgoCD), automated CI/CD pipelines, centralized monitoring (Prometheus/Grafana), and service mesh integration (Istio) for advanced traffic management and observability.

3

Kubernetes Clusters

24

Total Nodes

156

Running Pods

18

Services

99.9%

Uptime SLA

<50ms

Avg Latency

🏗️ Architecture Components

☸️

Kubernetes Clusters

AWS EKS, Azure AKS, GCP GKE running identical workloads

🔧

Infrastructure as Code

Terraform modules for consistent provisioning

🔄

GitOps

ArgoCD for declarative deployments

📊

Monitoring

Prometheus, Grafana, and centralized logging

🌐

Service Mesh

Istio for traffic management and security

🔒

Security

RBAC, network policies, and secrets management

☁️

Amazon EKS

Elastic Kubernetes Service

AWS EKS provides a highly available and secure Kubernetes control plane with seamless integration into AWS services like ALB, EBS, EFS, and CloudWatch. Offers the most mature ecosystem with extensive third-party integrations.

✅ Managed control plane with automatic upgrades
✅ Native AWS service integration (IAM, VPC, ALB)
✅ EC2 and Fargate compute options
✅ Multi-AZ high availability
✅ Integrated with AWS CloudWatch and X-Ray

EKS EC2 Fargate ALB

☁️

Azure AKS

Azure Kubernetes Service

Azure AKS offers a serverless Kubernetes experience with the deepest integration with Microsoft ecosystem. Features advanced networking options and excellent support for Windows containers alongside Linux workloads.

✅ Free control plane management
✅ Azure Active Directory integration
✅ Virtual nodes (serverless containers)
✅ Windows and Linux container support
✅ Azure Monitor and Application Insights

AKS Azure CNI Virtual Nodes AAD

☁️

Google GKE

Google Kubernetes Engine

GKE is built by the creators of Kubernetes, offering the most advanced features and fastest updates. Provides exceptional autoscaling capabilities and the best developer experience with built-in CI/CD integration.

✅ Autopilot mode for hands-off operations
✅ Advanced pod autoscaling and node auto-repair
✅ Binary Authorization for deployment security
✅ Built-in Config Connector for GCP resources
✅ Cloud Operations (Logging, Monitoring, Tracing)

GKE Autopilot Cloud Build GCP Load Balancer

🚀 AWS EKS Deployment

Terraform Configuration - AWS EKS

module "eks" { source = "terraform-aws-modules/eks/aws" version = "~> 19.0" cluster_name = "multi-cloud-production-eks" cluster_version = "1.28" vpc_id = module.vpc.vpc_id subnet_ids = module.vpc.private_subnets # Managed node groups eks_managed_node_groups = { production = { min_size = 3 max_size = 10 desired_size = 6 instance_types = ["t3.large", "t3.xlarge"] capacity_type = "SPOT" labels = { Environment = "production" Workload = "general" } tags = { ExtraTag = "multi-cloud-k8s" } } } # Cluster addons cluster_addons = { coredns = { most_recent = true } kube-proxy = { most_recent = true } vpc-cni = { most_recent = true } aws-ebs-csi-driver = { most_recent = true } } tags = { Environment = "production" Project = "multi-cloud-kubernetes" } }

AWS EKS Deployment Output

$ terraform apply -auto-approve

module.eks.aws_eks_cluster.this[0]: Creating...

module.vpc.aws_vpc.this[0]: Creation complete after 2s

module.eks.aws_eks_cluster.this[0]: Still creating... [2m30s elapsed]

✓ module.eks.aws_eks_cluster.this[0]: Creation complete after 11m42s

module.eks.aws_eks_node_group.this["production"]: Creating...

✓ module.eks.aws_eks_node_group.this["production"]: Creation complete after 3m15s

$ kubectl get nodes

NAME STATUS ROLES AGE VERSION

ip-10-0-1-45.ec2.internal Ready <none> 5m v1.28.3-eks-abc123

ip-10-0-2-78.ec2.internal Ready <none> 5m v1.28.3-eks-abc123

ip-10-0-3-92.ec2.internal Ready <none> 5m v1.28.3-eks-abc123

✓ AWS EKS cluster is ready with 6 nodes

🚀 Azure AKS Deployment

Terraform Configuration - Azure AKS

resource "azurerm_kubernetes_cluster" "main" { name = "multi-cloud-production-aks" location = azurerm_resource_group.main.location resource_group_name = azurerm_resource_group.main.name dns_prefix = "multicloud-k8s" kubernetes_version = "1.28.3" default_node_pool { name = "production" node_count = 6 vm_size = "Standard_D4s_v3" enable_auto_scaling = true min_count = 3 max_count = 10 upgrade_settings { max_surge = "33%" } } identity { type = "SystemAssigned" } network_profile { network_plugin = "azure" load_balancer_sku = "standard" outbound_type = "loadBalancer" } azure_active_directory_role_based_access_control { managed = true azure_rbac_enabled = true } oms_agent { log_analytics_workspace_id = azurerm_log_analytics_workspace.main.id } tags = { Environment = "production" Project = "multi-cloud-kubernetes" } }

Azure AKS Deployment Output

$ terraform apply -auto-approve

azurerm_kubernetes_cluster.main: Creating...

azurerm_resource_group.main: Creation complete after 1s

azurerm_kubernetes_cluster.main: Still creating... [4m30s elapsed]

✓ azurerm_kubernetes_cluster.main: Creation complete after 8m23s

$ az aks get-credentials --resource-group rg-multicloud --name multi-cloud-production-aks

Merged "multi-cloud-production-aks" as current context in ~/.kube/config

$ kubectl get nodes

NAME STATUS ROLES AGE VERSION

aks-production-12345678-vmss000000 Ready agent 3m v1.28.3

aks-production-12345678-vmss000001 Ready agent 3m v1.28.3

aks-production-12345678-vmss000002 Ready agent 3m v1.28.3

✓ Azure AKS cluster is ready with 6 nodes

🚀 Google GKE Deployment

Terraform Configuration - Google GKE

resource "google_container_cluster" "primary" { name = "multi-cloud-production-gke" location = "us-central1" # GKE Autopilot mode for fully managed experience enable_autopilot = false # Manual node pool management remove_default_node_pool = true initial_node_count = 1 network = google_compute_network.vpc.name subnetwork = google_compute_subnetwork.subnet.name # Cluster features addons_config { http_load_balancing { disabled = false } horizontal_pod_autoscaling { disabled = false } network_policy_config { disabled = false } } release_channel { channel = "REGULAR" } workload_identity_config { workload_pool = "${var.project_id}.svc.id.goog" } } resource "google_container_node_pool" "primary_nodes" { name = "production-pool" location = "us-central1" cluster = google_container_cluster.primary.name node_count = 2 autoscaling { min_node_count = 3 max_node_count = 10 } node_config { machine_type = "e2-standard-4" oauth_scopes = [ "https://www.googleapis.com/auth/cloud-platform" ] labels = { environment = "production" } tags = ["gke-node", "multi-cloud-kubernetes"] } }

Google GKE Deployment Output

$ terraform apply -auto-approve

google_container_cluster.primary: Creating...

google_compute_network.vpc: Creation complete after 45s

google_container_cluster.primary: Still creating... [3m0s elapsed]

✓ google_container_cluster.primary: Creation complete after 7m15s

google_container_node_pool.primary_nodes: Creating...

✓ google_container_node_pool.primary_nodes: Creation complete after 2m30s

$ gcloud container clusters get-credentials multi-cloud-production-gke --region us-central1

Fetching cluster endpoint and auth data.

kubeconfig entry generated for multi-cloud-production-gke.

$ kubectl get nodes

NAME STATUS ROLES AGE VERSION

gke-prod-pool-default-pool-a1b2c3d4-e5f6 Ready <none> 2m v1.28.3-gke.1098000

gke-prod-pool-default-pool-a1b2c3d4-g7h8 Ready <none> 2m v1.28.3-gke.1098000

gke-prod-pool-default-pool-a1b2c3d4-i9j0 Ready <none> 2m v1.28.3-gke.1098000

✓ Google GKE cluster is ready with 6 nodes

📊 Provider Comparison

Feature	AWS EKS	Azure AKS	Google GKE
Control Plane Cost	$0.10/hour (~$73/month)	Free	$0.10/hour (~$73/month)
Kubernetes Version	1.28.3	1.28.3	1.28.3
Upgrade Policy	Quarterly	Bi-monthly	Weekly (Regular channel)
Autoscaling	Cluster Autoscaler	Cluster Autoscaler + Virtual Nodes	Node Auto-provisioning
Serverless Containers	Fargate	Virtual Nodes (ACI)	Autopilot / Cloud Run
Windows Support	Yes	Yes (Best support)	Yes (Limited)
GPU Support	Yes (NVIDIA)	Yes (NVIDIA)	Yes (NVIDIA, TPU)
Network Plugin	AWS VPC CNI	Azure CNI / Kubenet	GKE CNI
Load Balancer	ALB / NLB	Azure Load Balancer	GCP Load Balancer
Storage Options	EBS, EFS, FSx	Azure Disk, Azure Files	Persistent Disk, Filestore
Monitoring	CloudWatch Container Insights	Azure Monitor	Cloud Operations Suite
SLA	99.95%	99.95% (99.9% free tier)	99.95% (Regional), 99.5% (Zonal)

💰 Cost Comparison (Monthly Estimate)

$892

AWS EKS Total

Control Plane: $73
6x t3.large nodes: $819

$748

Azure AKS Total

Control Plane: $0
6x Standard_D4s_v3: $748

$826

Google GKE Total

Control Plane: $73
6x e2-standard-4: $753

🎯 Best Use Cases

Choose AWS EKS When:

✓ Already heavily invested in AWS ecosystem
✓ Need deep integration with AWS services (RDS, S3, etc.)
✓ Require Fargate for serverless containers
✓ Want the most mature marketplace and ecosystem

Choose Azure AKS When:

✓ Using Microsoft enterprise stack (.NET, Azure AD)
✓ Need Windows container support
✓ Want free control plane to reduce costs
✓ Require Azure-specific services integration

Choose Google GKE When:

✓ Want the most advanced K8s features (built by Google)
✓ Need Autopilot mode for hands-off management
✓ Require ML workloads with TPU support
✓ Prefer the best developer experience and tooling