Azure Local Architecture Deep Dive

Table of Contents

  1. Overview
  2. Physical Infrastructure Requirements
    1. Node Configuration
    2. Server Hardware Specifications
    3. Validated Hardware Partners
  3. Hardware Topology and Placement
    1. Physical Rack Layout
    2. Data Center Requirements
  4. Networking Architecture
    1. Network Traffic Types
    2. Network Topology Options
    3. RDMA (Remote Direct Memory Access)
    4. Switch Requirements
  5. Security Layers and Encryption
    1. Hardware Security Foundation
    2. Encryption at Rest
    3. Network Security
    4. Operating System Security
    5. Azure Arc Security (Connected Mode)
  6. Control Plane vs. Data Plane Separation
    1. Data Plane (Always Local)
    2. Control Plane
    3. Why This Matters for Sovereignty
  7. Integration with Azure Cloud Control Plane
    1. Azure Arc Integration
    2. Azure Monitor Integration
    3. Azure Policy Enforcement
    4. Azure Update Manager
    5. Azure Backup Integration
    6. Azure Site Recovery
  8. Network Connectivity Requirements
    1. Required Azure Endpoints
    2. Disconnected Mode Considerations
  9. Disaster Recovery and Business Continuity
    1. Cluster-Level Resiliency
    2. Storage Resiliency
    3. Backup Strategies
    4. Recovery Scenarios
  10. Scalability Architecture
    1. Vertical Scaling (Scale-Up)
    2. Horizontal Scaling (Scale-Out)
    3. Multi-Cluster Scaling
  11. Performance Optimization
    1. Storage Performance Tuning
    2. Network Performance
    3. Compute Performance
  12. Monitoring and Maintenance
    1. Health Monitoring
    2. Maintenance Best Practices
  13. Next Steps

Overview

Azure Local architecture combines physical infrastructure with software-defined technologies to create a hyper-converged platform. Understanding the architecture is essential for planning deployments, sizing infrastructure, and troubleshooting issues.

View Diagram: Azure Local Architecture Stack
![Azure Local Architecture showing the full stack from hardware through Azure integration](/microsoft-sovereign-cloud-brain-trek/assets/images/level-100/azure-local-architecture.svg) _Figure 1: Complete Azure Local architecture with all layers from hardware to Azure cloud integration_

This page provides an in-depth look at:

  • Physical infrastructure requirements and topology
  • Logical components and their interactions
  • Networking architecture and design patterns
  • Security layers and encryption mechanisms
  • Integration with Azure cloud services

← Back to Azure Local Overview

Physical Infrastructure Requirements

Node Configuration

An Azure Local cluster consists of 1-16 physical servers (nodes), with most production deployments using 2-4 nodes for high availability.

Single-Node Clusters:

  • Suitable for branch offices or edge locations
  • No cluster-level redundancy
  • Lower cost entry point
  • Limited high availability

Multi-Node Clusters (Recommended):

  • 2-4 nodes: Most common configuration
  • 5-8 nodes: Large deployments with high capacity needs
  • 9-16 nodes: Hyper-scale scenarios

Server Hardware Specifications

Minimum Requirements (per node):

CPU:

  • 64-bit x86 processor with:
    • Intel VT-x or AMD-V virtualization extensions
    • Second Level Address Translation (SLAT)
    • 8+ cores recommended (16+ for production)
  • Examples: Intel Xeon Scalable, AMD EPYC

Memory:

  • Minimum: 128 GB DDR4
  • Recommended: 384 GB - 1 TB
  • Production: 512 GB - 6 TB per node
  • ECC (Error-Correcting Code) required

Storage:

  • OS drives: 2x 240 GB+ SATA SSD (mirrored)
  • Cache tier: 2-8x NVMe SSD (800 GB - 3.2 TB each)
  • Capacity tier: 4-16x SSD or HDD (1-16 TB each)
  • Minimum total: 4 drives per node
  • Maximum: 400+ drives per cluster

Network Adapters:

  • Minimum: 2x 10 GbE adapters
  • Recommended: 2x 25 GbE or higher
  • RDMA-capable adapters preferred (iWARP, RoCE, or InfiniBand)
  • Examples: Mellanox/NVIDIA, Intel, Broadcom

Other Requirements:

  • UEFI 2.3.1c or newer firmware
  • TPM 2.0 for security features
  • Redundant power supplies
  • BMC/iDRAC for remote management

Validated Hardware Partners

Microsoft maintains a catalog of validated Azure Local solutions from partners:

Tier 1 OEM Partners:

  • Dell EMC (PowerEdge servers)
  • HPE (ProLiant servers)
  • Lenovo (ThinkSystem servers)
  • Fujitsu (PRIMERGY servers)

Tier 2 System Builders:

  • Supermicro
  • DataON
  • QCT (Quanta)
  • Wortmann AG

Benefits of Validated Hardware:

  • ✅ Pre-tested configurations
  • ✅ Guaranteed compatibility
  • ✅ Simplified support
  • ✅ Firmware and driver validation
  • ✅ Performance optimization

Reference: Azure Local Catalog

Hardware Topology and Placement

Physical Rack Layout

Typical 4-Node Cluster Layout:

graph TB
    subgraph ToR[Network Layer]
        Switch1[Top of Rack Switch 1]
        Switch2[Top of Rack Switch 2]
    end

    subgraph Compute[Compute Nodes]
        Node1[Node 1<br/>CPU/RAM/Storage]
        Node2[Node 2<br/>CPU/RAM/Storage]
        Node3[Node 3<br/>CPU/RAM/Storage]
        Node4[Node 4<br/>CPU/RAM/Storage]
    end

    subgraph Infrastructure[Infrastructure]
        Mgmt[Management Switch]
        Power[UPS / PDU]
    end

    Switch1 -.->|Redundant| Node1
    Switch1 -.->|Redundant| Node2
    Switch1 -.->|Redundant| Node3
    Switch1 -.->|Redundant| Node4

    Switch2 -.->|Redundant| Node1
    Switch2 -.->|Redundant| Node2
    Switch2 -.->|Redundant| Node3
    Switch2 -.->|Redundant| Node4

    Node1 --> Mgmt
    Node2 --> Mgmt
    Node3 --> Mgmt
    Node4 --> Mgmt

    Mgmt --> Power
    Compute --> Power

    style ToR fill:#E8F4FD,stroke:#0078D4,stroke-width:2px,color:#000
    style Compute fill:#FFF4E6,stroke:#FF8C00,stroke-width:2px,color:#000
    style Infrastructure fill:#F3E8FF,stroke:#7B3FF2,stroke-width:2px,color:#000

Key Design Principles:

  • Dual top-of-rack (ToR) switches for redundancy
  • Each node connected to both ToR switches
  • Separate management network
  • Redundant power distribution
  • Cable management for airflow

Data Center Requirements

Physical Space:

  • Standard 42U rack sufficient for most clusters
  • 1-2U per node (typical)
  • Additional space for networking gear
  • Consider growth (future node additions)

Power:

  • Per-node: 300-800 watts typical, 1200+ watts max
  • 4-node cluster: ~5-10 kW total
  • UPS recommended for graceful shutdown
  • PDU with monitoring capability

Cooling:

  • Maintain ambient temperature: 10-35°C (50-95°F)
  • Hot aisle/cold aisle configuration recommended
  • Monitor temperature and humidity
  • Adequate airflow (front-to-back or back-to-front)

Connectivity:

  • Access to corporate network
  • Internet access (for Connected Mode)
  • Out-of-band management network
  • Consider future WAN upgrades

Networking Architecture

Azure Local networking is the most critical aspect of the architecture. Poor network design leads to performance issues and reduced reliability.

Network Traffic Types

Azure Local has four primary network traffic types:

1. Management Traffic

  • Cluster management and monitoring
  • Azure Arc communication (Connected Mode)
  • Remote administration (RDP, PowerShell)
  • Bandwidth: Low (< 100 Mbps typical)
  • Latency: Not critical (< 100ms acceptable)
  • Recommendation: 1 GbE sufficient

2. Storage Traffic (SMB)

  • Storage Spaces Direct (S2D) communication
  • Volume replication between nodes
  • Highest bandwidth requirement
  • Bandwidth: 10-100+ Gbps
  • Latency: Critical (< 1ms required, < 0.5ms optimal)
  • Recommendation: 25 GbE minimum, RDMA required

3. VM Compute Traffic

  • Virtual machine network communication
  • Tenant workload traffic
  • Live migration traffic
  • Bandwidth: 10-40 Gbps typical
  • Latency: Moderate (< 5ms)
  • Recommendation: 10 GbE minimum, 25+ GbE preferred

4. Cluster Communication

  • Failover clustering heartbeat
  • Cluster Shared Volumes (CSV) communication
  • Bandwidth: Low-Medium
  • Latency: Critical (< 2ms)
  • Recommendation: Dedicated interfaces

Network Topology Options

Option 1: Fully Converged (Minimum)

  • Single NIC team (2x 25 GbE)
  • All traffic types share bandwidth
  • Use with caution - limited scalability
  • Suitable for small deployments only

Option 2: Storage Isolated (Recommended)

  • Separate adapters for storage (2x 25 GbE RDMA)
  • Management and compute converged (2x 10 GbE)
  • Good balance of performance and cost
  • Most common deployment

Option 3: Fully Separated (Optimal)

  • Dedicated adapters per traffic type:
    • Management: 2x 1 GbE
    • Storage: 2x 25 GbE or higher (RDMA)
    • Compute: 2x 10-25 GbE
  • Best performance and isolation
  • Higher cost (more NICs, more switches)

RDMA (Remote Direct Memory Access)

RDMA is essential for production Storage Spaces Direct performance:

Benefits:

  • Reduces CPU utilization for storage traffic
  • Achieves sub-millisecond latency
  • Enables full network throughput
  • Required for optimal S2D performance

RDMA Protocols:

  • iWARP: Works over standard Ethernet, easier to configure
  • RoCE v2: Higher performance, requires lossless Ethernet (DCB)
  • InfiniBand: Highest performance, less common

Configuration:

  • Enable DCB (Data Center Bridging) for RoCE
  • Configure priority flow control (PFC)
  • Set up bandwidth reservation
  • Test with Test-RDMAConnectivity cmdlet

Switch Requirements

Top-of-Rack (ToR) Switches:

For Storage Networks (RDMA):

  • Line-rate switching at 25/40/100 GbE
  • Low latency (< 5 microseconds)
  • RDMA support (RoCE v2 or iWARP)
  • DCB capable (for RoCE)
  • Jumbo frames support (9000+ MTU)

For Management/Compute:

  • Standard enterprise-grade switches
  • Layer 2/3 capabilities
  • VLAN support
  • Link aggregation (LACP)

Popular Switch Vendors:

  • Cisco Nexus series
  • Arista 7000 series
  • Dell EMC S-series
  • Mellanox/NVIDIA Spectrum

Security Layers and Encryption

Azure Local implements defense-in-depth security across multiple layers.

Hardware Security Foundation

Secure Boot:

  • UEFI firmware-based boot protection
  • Verifies boot components before execution
  • Prevents rootkits and bootkits
  • Required for Azure Local certification

Trusted Platform Module (TPM 2.0):

  • Hardware root of trust
  • Stores encryption keys securely
  • Attestation and measurement
  • BitLocker key protection

Processor Security Features:

  • Intel VT-x / AMD-V (virtualization)
  • Intel VT-d / AMD-Vi (I/O virtualization)
  • Execute Disable (XD) bit
  • Data Execution Prevention (DEP)

Encryption at Rest

BitLocker Drive Encryption:

  • All volumes encrypted by default
  • Uses AES-256 encryption
  • TPM-protected keys
  • Transparent to applications and users

Encryption Scope:

  • Operating system volumes
  • Storage Spaces Direct volumes
  • Virtual machine disks (optional)
  • Configuration data

Key Management:

  • Keys stored in TPM
  • Recovery keys backed up to Azure (Connected Mode)
  • Manual backup for Disconnected Mode
  • Regular key rotation recommended

Network Security

SMB Encryption:

  • SMB 3.x with AES-128-GCM encryption
  • Protects storage traffic between nodes
  • Enabled by default on S2D volumes
  • Transparent performance impact (< 5%)

IPsec (Optional):

  • Encrypts cluster communication
  • Adds defense-in-depth
  • Slight performance overhead
  • Useful for untrusted networks

Network Isolation:

  • VLANs for traffic segmentation
  • Firewall rules between networks
  • Micro-segmentation with SDN
  • Zero trust network principles

Operating System Security

Windows Server Security Baseline:

  • CIS Benchmarks compliance
  • STIG (Security Technical Implementation Guide) hardening
  • Minimal attack surface
  • Regular security updates

Credential Guard:

  • Protects credentials using virtualization
  • Prevents pass-the-hash attacks
  • Uses hardware virtualization
  • Recommended for all deployments

AppLocker / Windows Defender Application Control:

  • Restrict application execution
  • Allow-list approach
  • Code integrity policies
  • Protection against malware

Azure Arc Security (Connected Mode)

Managed Identity:

  • No stored credentials
  • Azure AD authentication
  • Automatic token rotation
  • Least privilege access

Azure Policy Integration:

  • Enforce security configurations
  • Compliance reporting
  • Drift detection and remediation
  • Central governance

Azure Security Center:

  • Threat detection
  • Security recommendations
  • Vulnerability assessment
  • Integration with Azure Sentinel

Control Plane vs. Data Plane Separation

Understanding the separation between control and data planes is critical for security and operational sovereignty.

Data Plane (Always Local)

The data plane handles actual workload processing and data storage:

Components:

  • Hyper-V virtual machines
  • Storage Spaces Direct volumes
  • Virtual networks (SDN)
  • Container workloads (AKS)
  • Application data and processing

Characteristics:

  • ✅ Always runs locally on Azure Local hardware
  • ✅ No dependency on Azure cloud
  • ✅ Data never leaves premises (unless explicitly configured)
  • ✅ Sub-millisecond latency
  • ✅ Continues operating during internet outages

Important: In both Connected and Disconnected modes, the data plane is always local.

Control Plane

The control plane manages configuration, monitoring, and updates:

Connected Mode - Hybrid Control Plane:

  • Azure-managed functions:
    • Cluster registration and inventory
    • Monitoring and telemetry
    • Update orchestration
    • Azure Arc management
    • Policy enforcement
    • Security Center integration
  • Locally-managed functions:
    • VM creation and management
    • Storage configuration
    • Network configuration
    • Failover and recovery
    • Performance optimization

Disconnected Mode - Local Control Plane:

  • All management functions run locally
  • Windows Admin Center for management
  • PowerShell for automation
  • No Azure dependency
  • Manual update deployment

Why This Matters for Sovereignty

The control/data plane separation provides:

Data Sovereignty:

  • Workload data stays on-premises
  • No data transmitted to Azure without explicit action
  • Compliance with data residency requirements

Operational Sovereignty:

  • Critical operations (VM failover, storage access) are local
  • Continue operations during Azure outages
  • Reduced dependence on cloud connectivity

Hybrid Benefits:

  • Centralized management when connected
  • Local autonomy when needed
  • Best of both worlds

Integration with Azure Cloud Control Plane

When deployed in Connected Mode, Azure Local integrates with Azure services for management, monitoring, and governance.

Azure Arc Integration

Azure Arc is the foundation for Azure Local’s cloud integration:

Registration Process:

  1. Install Azure Arc agents on cluster nodes
  2. Register cluster with Azure subscription
  3. Create Azure Resource Manager representation
  4. Enable Azure management services

Arc-Enabled Capabilities:

  • Resource inventory in Azure portal
  • RBAC (Role-Based Access Control)
  • Azure Policy assignment and enforcement
  • Resource tagging and organization
  • Cost tracking and billing

Reference: Azure Arc Overview

Azure Monitor Integration

Metrics Collection:

  • CPU, memory, storage, network utilization
  • VM performance metrics
  • Storage latency and IOPS
  • Cluster health status
  • 1-5 minute granularity

Log Collection:

  • System event logs
  • Application logs
  • Security audit logs
  • Storage logs
  • Queryable via Log Analytics

Alerting:

  • Threshold-based alerts
  • Anomaly detection
  • Action groups for notifications
  • Integration with ITSM tools

Dashboards:

  • Pre-built cluster health dashboards
  • Custom workbooks
  • Cross-cluster views
  • Trend analysis

Azure Policy Enforcement

Built-in Policies:

  • Require BitLocker encryption
  • Enforce security baselines
  • Restrict VM sizes
  • Network security rules
  • Compliance standards (PCI-DSS, HIPAA, etc.)

Custom Policies:

  • Define organization-specific rules
  • JSON-based policy definitions
  • Audit or enforce mode
  • Exemption handling

Compliance Reporting:

  • Real-time compliance status
  • Historical compliance trends
  • Remediation recommendations
  • Export to Excel/CSV

Azure Update Manager

Update Orchestration:

  • Scheduled update windows
  • Pre/post-update scripts
  • Phased rollout (cluster-aware updating)
  • Rollback capability
  • Update approval workflow

Update Types:

  • Windows Server updates
  • Azure Local-specific updates
  • Driver and firmware updates
  • Security patches (priority)

Benefits vs. Manual Updates:

  • Reduced administrative overhead
  • Consistent update process
  • Lower risk of errors
  • Audit trail
  • Centralized control for multiple clusters

Azure Backup Integration

VM-Level Backup:

  • Application-consistent backups
  • Scheduled backup policies
  • Long-term retention
  • Cross-region replication
  • Quick restore capability

Backup Storage:

  • Data stored in Azure Recovery Services vault
  • Encrypted in transit and at rest
  • Redundancy options (LRS, GRS)
  • Cost-effective long-term storage

Hybrid Benefits:

  • On-premises VMs backed up to Azure
  • No need for local backup infrastructure
  • Disaster recovery to Azure
  • Compliance with retention policies

Azure Site Recovery

Disaster Recovery:

  • Replicate VMs to Azure
  • Failover during outages
  • Failback when on-premises restored
  • Recovery time objective (RTO): minutes
  • Recovery point objective (RPO): minutes

DR Scenarios:

  • Site-to-Azure DR
  • Site-to-site DR (between Azure Local clusters)
  • Compliance with BCO/DR requirements

Network Connectivity Requirements

For Connected Mode, Azure Local requires outbound connectivity to specific Azure services.

Required Azure Endpoints

Azure Arc Services:

  • management.azure.com - Azure Resource Manager
  • login.microsoftonline.com - Azure AD authentication
  • *.his.arc.azure.com - Hybrid Instance Metadata Service
  • *.guestconfiguration.azure.com - Guest configuration

Azure Monitor:

  • *.ods.opinsights.azure.com - Operations data
  • *.oms.opinsights.azure.com - Operational management
  • *.monitoring.azure.com - Monitoring data

Azure Update Manager:

  • *.download.microsoft.com - Update downloads
  • *.windowsupdate.com - Windows Update
  • *.delivery.mp.microsoft.com - Content delivery

Azure Backup:

  • *.backup.windowsazure.com - Backup service
  • *.blob.core.windows.net - Blob storage

Other Services:

  • *.vault.azure.net - Key Vault (for secrets)
  • *.servicebus.windows.net - Service Bus (for events)

Port Requirements:

  • Outbound HTTPS (TCP 443) for all services
  • No inbound connections from internet required
  • Proxy server supported

Bandwidth Recommendations:

  • Minimum: 1-5 Mbps per cluster
  • Recommended: 50-100 Mbps
  • Optimal: 1+ Gbps (for backup/DR)

Latency Tolerance:

  • Target: < 150ms to Azure region
  • Maximum: < 250ms for acceptable performance
  • Higher latency impacts management responsiveness

Disconnected Mode Considerations

In Disconnected Mode:

  • No Azure connectivity required
  • All management is local
  • Windows Admin Center for management
  • Updates deployed manually
  • Local backup solutions needed

Periodic Sync Option:

  • Connect periodically for updates
  • Download updates to USB/portable media
  • Air-gapped transfer to cluster
  • Apply updates manually

Disaster Recovery and Business Continuity

Azure Local provides multiple mechanisms for business continuity.

Cluster-Level Resiliency

Node Failure Tolerance:

  • 2-node cluster: 1 node failure (50% capacity loss)
  • 3-node cluster: 1 node failure (33% capacity loss)
  • 4-node cluster: 1 node failure (25% capacity loss)
  • N-node cluster: Up to 2 node failures with three-way mirroring

Automatic Recovery:

  • VMs automatically restart on healthy nodes
  • Live migration when possible (planned maintenance)
  • Storage healing begins immediately
  • Cluster reforms after network partition

Storage Resiliency

Resiliency Types:

Two-Way Mirror:

  • 2 copies of data
  • Tolerates 1 disk/node failure
  • 50% storage efficiency
  • Faster rebuild times
  • Recommended for most workloads

Three-Way Mirror:

  • 3 copies of data
  • Tolerates 2 simultaneous disk/node failures
  • 33.3% storage efficiency
  • Higher durability
  • Recommended for critical workloads

Erasure Coding (Parity):

  • Data + parity stripes
  • Tolerates 1-2 failures (depends on configuration)
  • 50-80% storage efficiency
  • Higher CPU usage
  • Recommended for capacity-optimized storage

Storage Repair:

  • Automatic detection of failed components
  • Background data rebuild
  • Parallel rebuild for faster recovery
  • Throttling to avoid performance impact

Backup Strategies

Local Backups:

  • Secondary Azure Local cluster
  • Traditional backup software (Veeam, Commvault, etc.)
  • Azure Backup to local Azure Local
  • Replica volumes

Cloud Backups (Connected Mode):

  • Azure Backup for VMs
  • Azure Site Recovery for DR
  • Azure Blob replication for files
  • Long-term retention in Azure

Hybrid Approach:

  • Local backups for fast recovery
  • Cloud backups for DR and long-term retention
  • Best of both: RPO/RTO + offsite protection

Recovery Scenarios

Scenario 1: Single Node Failure

  • VMs automatically restart on other nodes
  • No administrative action required
  • Replace failed node when convenient
  • Automatic storage rebuild

Scenario 2: Complete Cluster Failure

  • Restore from backup to new cluster
  • Failover to Azure (if using Site Recovery)
  • Failover to secondary Azure Local cluster
  • RTO: Hours to days (depends on strategy)

Scenario 3: Datacenter Disaster

  • Failover to Azure using Site Recovery
  • Recover from Azure Backup to Azure VMs
  • Failover to remote Azure Local cluster
  • RTO: Hours (with proper planning)

Scalability Architecture

Vertical Scaling (Scale-Up)

Increase capacity of existing nodes:

CPU:

  • Upgrade to higher core count processors
  • Requires node shutdown
  • Consider memory capacity when upgrading

Memory:

  • Add memory modules to existing nodes
  • Can be done with rolling updates
  • Monitor memory usage before scaling

Storage:

  • Add drives to existing nodes
  • Hot-add supported for data drives
  • Automatic rebalancing

Network:

  • Upgrade to faster NICs (10→25→40→100 GbE)
  • Requires brief outage per node
  • Improve throughput and latency

Horizontal Scaling (Scale-Out)

Add nodes to the cluster:

Adding Nodes:

  1. Prepare new hardware (validated configuration)
  2. Join to domain
  3. Install Azure Local OS
  4. Run Add-ClusterNode cmdlet
  5. Cluster automatically rebalances

Benefits:

  • Increased capacity (compute, storage, networking)
  • Improved redundancy
  • Linear scalability
  • No downtime for addition

Limitations:

  • Maximum 16 nodes per cluster
  • All nodes should be similar hardware
  • Network bandwidth critical

Multi-Cluster Scaling

For scenarios beyond 16 nodes:

Separate Clusters:

  • Deploy multiple independent clusters
  • Manage via Azure Arc
  • Workload placement by cluster
  • Good for multi-site deployments

⚠️ Stretch Clusters Not Supported: Azure Local does not support stretch clusters (single cluster spanning two sites). For multi-site high availability, use Azure Site Recovery or Storage Replica between separate Azure Local clusters at each site.

Performance Optimization

Storage Performance Tuning

Cache Configuration:

  • Use NVMe SSDs for cache tier
  • 10-20% of capacity tier size recommended
  • Read/write cache for optimal performance

Volume Optimization:

  • Use appropriate resiliency (mirror for performance, parity for capacity)
  • Fixed provisioning for predictable performance
  • Separate volumes for different workload types

Deduplication and Compression:

  • Enable for file servers and VDI
  • Disable for databases and high-IOPS workloads
  • Monitor CPU impact

Network Performance

RDMA Tuning:

  • Verify RDMA is functioning (Get-NetAdapterRdma)
  • Enable jumbo frames (MTU 9000+)
  • Configure DCB properly (for RoCE)
  • Monitor with Get-RDMACounter

Traffic Management:

  • Use QoS for traffic prioritization
  • Separate VLANs for traffic isolation
  • Monitor switch utilization

Compute Performance

VM Placement:

  • Distribute VMs evenly across nodes
  • Consider anti-affinity for redundancy
  • Use placement rules for performance isolation

Resource Allocation:

  • Right-size VMs (don’t over-allocate)
  • Use dynamic memory wisely
  • Monitor for resource contention

Monitoring and Maintenance

Health Monitoring

Built-in Health Service:

  • Automatic fault detection
  • Predictive failure analysis
  • Storage capacity monitoring
  • Performance baseline tracking

Monitoring Tools:

  • Windows Admin Center dashboard
  • Azure Monitor (Connected Mode)
  • Performance Monitor (PerfMon)
  • Event Viewer

Key Metrics to Monitor:

  • CPU utilization (< 70% average)
  • Memory usage (< 80%)
  • Storage latency (< 5ms for SSD, < 20ms for HDD)
  • Network utilization (< 60%)
  • Storage capacity (< 80% full)

Maintenance Best Practices

Regular Tasks:

  • Apply Windows updates monthly
  • Review event logs weekly
  • Monitor capacity trends
  • Test backups regularly
  • Update firmware quarterly

Cluster-Aware Updating:

  • Schedule during maintenance windows
  • Use CAU for automatic node updates
  • Validate after each update
  • Have rollback plan

Next Steps

Continue Learning:

External Resources:

Last Updated: October 2025