← Perspectives

Cloud Engineering

Azure: Enterprise Foundation Patterns

Azure Landing Zones provide the architectural foundation for enterprise cloud adoption—but implementing them in regulated industries requires more than following Microsoft's reference architectures. Here's how we approach Azure foundations for organizations where security and compliance aren't optional.9 min read

Azure Landing Zones provide the architectural foundation for enterprise cloud adoption—but implementing them in regulated industries requires more than following Microsoft's reference architectures. Here's how we approach Azure foundations for organizations where security and compliance aren't optional.

Why Foundations Matter

We've seen it repeatedly: organizations rush to deploy workloads in Azure, then spend the next two years retrofitting governance, security, and networking. The cost of this approach isn't just financial—it's the organizational drag of constantly explaining why things need to change, why access needs to be restricted, why that "quick deployment" now needs to be rebuilt.

A well-designed Azure foundation doesn't slow you down. It provides the guardrails that let teams move fast without creating security debt. The key is getting the hierarchy, policies, and network topology right from the start.

Management Group Hierarchy

Azure's Management Group structure is the skeleton of your governance model. Unlike AWS's account-centric model, Azure uses a hierarchical approach where policies and access controls flow downward through the tree.

Recommended Azure Management Group Structure Root Tenant Platform Landing Zones Sandbox Identity Management Connectivity Corp (Internal) Online (Public) Prod Non-Prod Prod Dev Key Components Root: Enterprise-wide policies (security baselines, allowed regions) Platform: Shared services (Identity via Entra ID, Connectivity via hub VNet, Management via Log Analytics) Landing Zones: Application workloads (Corp = internal apps, Online = public-facing) Sandbox: Experimentation with limited connectivity (no production data)

Figure 1: Azure Management Group hierarchy for regulated industries

Platform vs. Landing Zones

The critical distinction is between Platform (shared infrastructure managed centrally) and Landing Zones (where application teams deploy workloads). Platform subscriptions handle identity, connectivity, and management—the infrastructure that application teams consume but don't manage.

Why Corp and Online?

The Corp/Online split reflects different security postures. Corp landing zones connect to on-premises via ExpressRoute or VPN—they're for internal applications that need access to corporate resources. Online landing zones are for public-facing applications with stricter isolation from internal networks.

Azure Policy: Governance at Scale

Azure Policy is where governance intentions become enforceable rules. The challenge isn't writing policies—it's structuring them so they're maintainable and don't create conflicts as they inherit down the hierarchy.

Policy Assignment Strategy

Level Policy Focus Examples
Root Universal security baselines Allowed regions, required tags, deny public IPs on VMs
Platform Infrastructure standards Diagnostic settings to Log Analytics, NSG flow logs required
Landing Zones Workload guardrails Require private endpoints, deny unapproved SKUs
Corp Internal app requirements Require ExpressRoute connectivity, Azure AD auth only
Online Public app requirements WAF required, DDoS protection enabled
Example: Deny Public IP on VMs (Bicep)
resource policyDef 'Microsoft.Authorization/policyDefinitions@2021-06-01' = {
  name: 'deny-vm-public-ip'
  properties: {
    policyType: 'Custom'
    mode: 'All'
    displayName: 'Deny Public IP addresses on VMs'
    policyRule: {
      if: {
        allOf: [
          { field: 'type', equals: 'Microsoft.Network/networkInterfaces' }
          { field: 'Microsoft.Network/networkInterfaces/ipConfigurations[*].publicIpAddress.id', exists: true }
        ]
      }
      then: { effect: 'deny' }
    }
  }
}

Hub-Spoke Network Topology

For regulated industries, we recommend the hub-spoke topology with Azure Virtual WAN or a traditional hub VNet. The choice depends on scale, existing investments, and specific requirements around routing control.

Hub-Spoke Network Topology with Centralized Inspection On-Premises Data Center ExpressRoute Hub VNet Azure Firewall VPN GW ER GW Azure Bastion Private DNS Zones Spoke: Prod App Services SQL Databases Storage Spoke: Non-Prod Dev/Test Environments Spoke: Data Databricks Synapse Internet ER/VPN Peering Traffic Flow: All spoke traffic routes through Azure Firewall for inspection • Private endpoints keep PaaS traffic on Microsoft backbone

Figure 2: Hub-spoke topology with centralized security inspection

Private Endpoints: The Non-Negotiable

In regulated environments, we treat Private Endpoints as mandatory for any PaaS service that supports them. This keeps traffic on Microsoft's backbone network rather than traversing the public internet, even with encryption.

The overhead is real—Private DNS zones need proper configuration, and you'll need to plan for the IP address consumption. But the alternative (public endpoints with IP restrictions) creates audit findings and limits your ability to implement zero-trust architectures.

Key Takeaway

Configure Private DNS zones in the Hub subscription and link them to all spoke VNets. This centralizes DNS management and ensures consistent name resolution across your environment.

Identity Foundation

Azure's identity story centers on Microsoft Entra ID (formerly Azure AD). For enterprises, this typically means hybrid identity with synchronization from on-premises Active Directory.

Privileged Access Strategy

We implement tiered access based on the Enterprise Access Model:

  • Control Plane: Azure roles for subscription and resource management (limited to platform team)
  • Management Plane: Entra ID roles for identity administration (security team)
  • Data Plane: Application-specific access (development teams via RBAC)

Privileged Identity Management (PIM) is essential—no standing access to privileged roles. Users request just-in-time elevation, approvals are logged, and access expires automatically.

Compliance as Code

Microsoft Defender for Cloud provides the continuous compliance assessment that auditors expect. The key is mapping your regulatory requirements to Azure Policy initiatives and Defender recommendations.

Compliance Framework Built-in Initiative Additional Configuration
CIS Azure Benchmark ✓ Available Customize for organization-specific controls
SOC 2 ✓ Available Map to Trust Service Criteria
PCI DSS ✓ Available CDE scoping, network segmentation validation
HIPAA ✓ Available BAA in place, PHI handling procedures
ISO 27001 ✓ Available Evidence collection for annual audits
Built-in Doesn't Mean Complete

Azure's built-in compliance initiatives provide a starting point, not a complete solution. You'll need to supplement with custom policies for organization-specific requirements and validate that technical controls actually satisfy the regulatory intent—not just the checkbox.

Lessons from Implementation

After deploying Azure foundations across multiple regulated clients, patterns emerge:

  • Start with connectivity: The hub VNet and ExpressRoute/VPN configuration block everything else. Prioritize this.
  • Policy inheritance surprises: Test policy effects in sandbox subscriptions before applying at higher levels. Conflicts create confusion.
  • Private Endpoint DNS is complex: Plan your DNS architecture early. Retrofitting is painful when you have 50 private endpoints.
  • Subscription limits are real: Plan your subscription strategy around service limits and administrative boundaries, not just billing.
  • Training matters: A perfect foundation means nothing if application teams don't understand how to use it. Budget for enablement.

Looking Ahead

Azure Landing Zones continue to evolve. The Azure Verified Modules initiative provides Terraform and Bicep modules that encode Microsoft's best practices. Sovereign clouds and data residency requirements are driving region-specific configurations. And AI workloads are creating new patterns for GPU compute isolation and model security.

The foundation you build today needs to accommodate these future workloads. That means designing for flexibility while maintaining the governance that regulated industries require.