1. Why Hybrid Cloud?

Few enterprises are 100% cloud. Common drivers for hybrid architecture:

  • Compliance & data sovereignty — regulations mandate data stays on-prem or in-country.
  • Latency — factory floor, retail POS, or medical devices need sub-millisecond response.
  • Migration timeline — lift-and-shift takes years; you need unified management during transition.
  • Existing investment — hardware with years of depreciation remaining can't be abandoned overnight.

2. Azure Arc Overview

Azure Arc projects non-Azure resources into Azure Resource Manager (ARM). Once projected, you manage them with the same tools: Portal, CLI, Policy, RBAC, Monitor, and Defender for Cloud.

Azure Control Plane
ARM · Portal · Policy · Monitor
Arc Agent / Extension
On-Premises
VMs · K8s · SQL
AWS
EC2 · EKS
GCP
GCE · GKE

The Arc agent establishes an outbound HTTPS connection — no inbound firewall rules needed. Resources appear as ARM objects with resource IDs, tags, and location metadata.

3. Arc-Enabled Servers

Install the Connected Machine agent on Windows or Linux to project physical or virtual servers into Azure.

  • Onboarding — single script, at-scale via service principal, or Configuration Manager / Ansible.
  • Guest Configuration (Machine Configuration) — audit and enforce OS settings (password policy, installed software) using Azure Policy.
  • Extensions — deploy Log Analytics agent, Dependency agent, Custom Script, Key Vault for certificates.
  • Azure Policy — apply the same compliance policies to on-prem servers as to Azure VMs.
  • Microsoft Defender for Servers — threat detection on Arc-enabled machines.

4. Arc-Enabled Kubernetes

Connect any CNCF-conformant cluster — on-prem, EKS, GKE, Rancher — to Azure.

  • Azure RBAC — grant Azure AD identities access to namespaces without managing kubeconfig.
  • GitOps (Flux v2) — declare cluster configuration in Git; Arc reconciles automatically.
  • Azure Policy for K8s — enforce pod security, image registries, resource limits via Gatekeeper.
  • Cluster Extensions — deploy Azure Monitor Container Insights, Defender, Open Service Mesh.
  • App Services / ML on Arc — run App Service, Functions, Logic Apps, or ML inference on your own cluster.

5. Arc-Enabled Data Services

Run Azure-managed database engines on any Kubernetes infrastructure with Azure billing and evergreen updates.

  • SQL Managed Instance (Arc) — full SQL MI feature set running on your K8s cluster, with automated patching.
  • Azure Arc-enabled PostgreSQL — Hyperscale PostgreSQL on any infrastructure.
  • Connectivity modesDirectly connected (real-time telemetry) or Indirectly connected (air-gapped, periodic upload).
  • Billing — pay-as-you-go or SQL Server license with Software Assurance; metered through Azure.

6. Azure Stack Family

Azure (Public Cloud)
Full services · Global scale
Azure Stack HCI
On-prem Azure VMs · Validated hardware · AKS-HCI
Azure Stack Hub
Disconnected / sovereign · Full IaaS+PaaS · Own marketplace
Azure Stack Edge
IoT / AI at edge · Managed appliance · ML inference
  • Azure Stack HCI — hyper-converged on-prem cluster; run Azure VMs, AKS, and Azure Virtual Desktop locally with Azure billing.
  • Azure Stack Hub — fully disconnected Azure environment for sovereignty, submarine, or classified workloads.
  • Azure Stack Edge — Microsoft-managed appliance for edge compute, ML inference, and data transfer.

7. Multi-Cloud Patterns

Azure Arc treats AWS and GCP resources as first-class Azure objects:

  • Unified inventory — see all EC2 instances, GKE clusters in the Azure portal alongside Azure resources.
  • Consistent governance — one Azure Policy definition enforces tagging, allowed regions, and security baselines everywhere.
  • Single pane monitoring — Azure Monitor and Defender aggregate metrics/alerts from all clouds.
  • Cost visibility — while native billing stays per-cloud, Azure Resource Graph queries span all projected resources for governance.

Pattern: Use Arc for governance & compliance; keep workloads running natively in each cloud to avoid re-platforming.

8. Arc vs Azure Stack HCI vs Native Azure

9. Real-World: Unified Factory Management

Scenario: A manufacturing company has 12 factories with on-prem servers, 3 AWS accounts (acquired company), and Azure for corporate IT.

  • Arc-enabled servers on 200+ factory machines — Azure Policy enforces CIS benchmarks; Defender detects threats.
  • Arc-enabled K8s on factory-floor clusters running MES software — GitOps deploys config updates across all sites.
  • Arc-enabled SQL MI stores production data locally (latency); telemetry replicates to Azure for analytics.
  • AWS EC2 instances projected into Arc — same tagging and compliance policies as Azure resources.
  • Azure Stack Edge at remote factories with unreliable internet — runs ML quality-inspection models offline.

Result: One security team monitors all environments from Azure Portal; compliance reports generated from a single Resource Graph query spanning 3 clouds.

10. Exam Tip

🎯 AZ-305 loves scenarios: "The customer has servers in AWS and on-prem and needs unified policy." Answer = Azure Arc. If they need to run Azure VMs on-prem, that's Azure Stack HCI. If fully disconnected, that's Azure Stack Hub. Arc doesn't create resources — it projects existing ones into ARM for management.

11. Knowledge Check