Azure Compute & Containers Fundamentals¶
Virtual Machines Basics¶
What are Azure VMs?¶
Infrastructure-as-a-Service (IaaS) offering providing virtual machines in the cloud with full control over the operating system and applications.
VM Series & Families¶
| Series | Purpose | Examples | Use Cases |
|---|---|---|---|
| General Purpose | Balanced CPU-to-memory | B, D, F, E | Web servers, small-medium databases |
| Compute Optimized | High CPU-to-memory | F | CPU-intensive applications |
| Memory Optimized | High memory-to-CPU | E, M, G | Large databases, in-memory analytics |
| Storage Optimized | High disk throughput | L | Big data, SQL, NoSQL databases |
| GPU | Graphics processing | N | Machine learning, video rendering |
| High Performance | Fastest CPU processors | H | HPC, simulations |
Popular VM Sizes¶
| Size | vCPUs | RAM | Use Case |
|---|---|---|---|
| B1s | 1 | 1 GB | Development, testing |
| D2s_v3 | 2 | 8 GB | Small production workloads |
| D4s_v3 | 4 | 16 GB | Medium applications |
| E4s_v3 | 4 | 32 GB | Memory-intensive apps |
| F4s_v2 | 4 | 8 GB | CPU-intensive workloads |
VM Pricing Options¶
| Option | Description | Discount | Use Case |
|---|---|---|---|
| Pay-as-you-go | Hourly billing | None | Variable workloads |
| Reserved Instances | 1-3 year commitment | Up to 72% | Predictable workloads |
| Spot Instances | Unused capacity | Up to 90% | Fault-tolerant workloads |
| Azure Hybrid Benefit | Bring existing licenses | Significant | Windows/SQL Server migrations |
Availability Options¶
| Option | SLA | Configuration | Use Case |
|---|---|---|---|
| Single VM | 99.9% | Premium SSD required | Simple applications |
| Availability Set | 99.95% | Multiple fault/update domains | Traditional applications |
| Availability Zone | 99.99% | Across zones | Mission-critical applications |
| Virtual Machine Scale Sets | 99.95% | Auto-scaling group | Scalable applications |
Azure Functions (Serverless)¶
What are Azure Functions?¶
Event-driven serverless compute service that runs code on-demand without managing infrastructure.
Key Characteristics¶
- Event-driven: Triggered by various Azure services
- Automatic scaling: From 0 to thousands of instances
- No server management: Azure handles infrastructure
- Pay per execution: Charge only for compute time used
- Multiple languages: C#, JavaScript, Python, Java, PowerShell
Hosting Plans¶
| Plan | Scaling | Timeout | Use Case |
|---|---|---|---|
| Consumption | Automatic | 5 minutes | Variable workloads |
| Premium | Pre-warmed | 30 minutes | Predictable performance |
| Dedicated | Manual/Auto | Unlimited | Existing App Service Plan |
Supported Triggers¶
- HTTP/HTTPS: REST APIs, webhooks
- Timer: Scheduled execution (cron)
- Blob Storage: File upload/modification
- Queue Storage: Message processing
- Event Hub: Streaming data
- Service Bus: Enterprise messaging
- Cosmos DB: Database changes
Function App Features¶
- Deployment slots: Staging and production environments
- Application settings: Environment variables and secrets
- CORS support: Cross-origin resource sharing
- Authentication: Built-in identity providers
- Monitoring: Application Insights integration
App Service (Platform-as-a-Service)¶
What is App Service?¶
Fully managed platform for building, deploying, and scaling web apps and APIs.
App Service Plans¶
| Tier | Features | Use Case |
|---|---|---|
| Free (F1) | Shared infrastructure, 1GB storage | Development, testing |
| Shared (D1) | Shared infrastructure, custom domains | Small applications |
| Basic (B1-B3) | Dedicated VMs, SSL, manual scale | Production workloads |
| Standard (S1-S3) | Auto-scale, deployment slots | Standard production |
| Premium (P1-P3) | Enhanced performance, VNet integration | High-performance apps |
| Isolated (I1-I3) | Dedicated environment, private networking | Enterprise applications |
Supported Platforms¶
- Web Apps: .NET, Java, PHP, Node.js, Python, Ruby
- API Apps: RESTful web services
- Mobile Apps: Mobile backend services
- Function Apps: Serverless functions
Key Features¶
- Auto-scaling: Scale based on metrics
- Deployment slots: Blue-green deployments
- Custom domains: Bring your own domain
- SSL certificates: Free and custom certificates
- Authentication: Social and enterprise identity providers
- Continuous deployment: Git, GitHub, Azure DevOps integration
Container Services¶
Azure Container Instances (ACI)¶
Simplest way to run containers
Key Features¶
- Serverless containers: No VM management
- Fast startup: Containers start in seconds
- Per-second billing: Pay only for running time
- Public/private: Internet-accessible or VNet-connected
- Persistent storage: Azure Files mounting
Use Cases¶
- Batch jobs: Data processing tasks
- Build agents: CI/CD pipeline runners
- Application testing: Isolated test environments
- Burst scaling: Overflow capacity for AKS
Azure Kubernetes Service (AKS)¶
Managed Kubernetes service
Key Features¶
- Managed control plane: Azure manages master nodes
- Integrated monitoring: Azure Monitor for containers
- Azure AD integration: RBAC with enterprise identity
- Virtual Node: Serverless pods with ACI
- Dev Spaces: Kubernetes development environments
Node Pool Types¶
| Type | Description | Use Case |
|---|---|---|
| System | Run system pods | Required for cluster operation |
| User | Run application workloads | Your applications |
| Spot | Use spot VMs | Cost-effective batch workloads |
Scaling Options¶
- Cluster Autoscaler: Add/remove nodes based on demand
- Horizontal Pod Autoscaler: Scale pods based on metrics
- Vertical Pod Autoscaler: Adjust pod resource requests
- Virtual Node: Burst to ACI for unlimited scale
Azure Container Registry (ACR)¶
Private container registry
Features¶
- Geo-replication: Replicate across regions
- Security scanning: Vulnerability assessment
- Content trust: Signed image validation
- Helm charts: Store Kubernetes applications
- Tasks: Build images in the cloud
Service Tiers¶
| Tier | Storage | Operations/month | Features |
|---|---|---|---|
| Basic | 10 GB | 10,000 | Manual builds |
| Standard | 100 GB | 100,000 | Webhooks, geo-replication |
| Premium | 500 GB | 500,000 | Advanced security, private endpoints |
Azure Batch¶
What is Azure Batch?¶
Cloud service for running large-scale parallel and high-performance computing (HPC) applications.
Key Components¶
- Pool: Collection of compute nodes
- Job: Logical grouping of tasks
- Task: Individual unit of work
- Application Package: Software deployed to nodes
Use Cases¶
- Financial modeling: Risk analysis, portfolio optimization
- Digital media: Video transcoding, 3D rendering
- Engineering: CAD, simulation, analysis
- Scientific computing: Research, data processing
Scaling Options¶
- Fixed: Predetermined number of nodes
- Auto-scale: Dynamic scaling based on formulas
- Low-priority VMs: Use spot instances for cost savings
Service Fabric¶
What is Service Fabric?¶
Distributed systems platform for packaging, deploying, and managing scalable microservices.
Application Models¶
- Stateless services: No local state storage
- Stateful services: Reliable local state storage
- Actor model: Virtual actor programming model
- Guest executables: Existing applications
Key Features¶
- Self-healing: Automatic failure detection and recovery
- Auto-scaling: Scale services based on demand
- Rolling upgrades: Zero-downtime deployments
- Service discovery: Built-in service location
- Health monitoring: Application and infrastructure health
Service Comparison & Decision Tree¶
When to Choose What?¶
Compute Decision Tree¶
Need full OS control? → Virtual Machines
Containerized application? → AKS or Container Instances
Event-driven function? → Azure Functions
Web application/API? → App Service
HPC/batch processing? → Azure Batch
Microservices platform? → Service Fabric
Container Orchestration¶
Simple container deployment? → Container Instances
Need Kubernetes features? → AKS
Legacy application modernization? → Service Fabric
Hybrid on-premises/cloud? → Service Fabric
Serverless Options¶
HTTP APIs/web apps? → App Service or Functions
Event processing? → Azure Functions
Long-running processes? → Container Instances
Scheduled tasks? → Azure Functions (Timer trigger)
Practical Examples¶
Web Application Architecture¶
- Frontend: App Service Web App
- API: App Service API App or Azure Functions
- Background processing: Azure Functions
- Static content: Azure Storage + CDN
- Database: Azure SQL Database
Microservices on AKS¶
- Container platform: Azure Kubernetes Service
- Service mesh: Istio or Linkerd
- Load balancing: Azure Load Balancer
- Monitoring: Azure Monitor for containers
- Registry: Azure Container Registry
Batch Processing Pipeline¶
- Orchestration: Azure Batch
- Data input: Azure Storage
- Processing: Custom applications in containers
- Results: Azure Storage or databases
- Monitoring: Azure Monitor
Event-Driven Architecture¶
- Trigger: Azure Storage blob upload
- Processing: Azure Functions
- Messaging: Service Bus or Event Hub
- State: Cosmos DB
- Notifications: Logic Apps
Best Practices¶
Virtual Machines¶
- Use appropriate VM sizes for workloads
- Implement availability sets or zones
- Use managed disks for storage
- Configure auto-shutdown for dev/test
- Monitor performance with Azure Monitor
- Use Azure Security Center recommendations
Azure Functions¶
- Keep functions small and focused
- Use appropriate hosting plan for workload
- Implement proper error handling
- Use Application Insights for monitoring
- Store secrets in Key Vault
- Design for idempotency
App Service¶
- Use deployment slots for staging
- Configure auto-scaling rules
- Implement health checks
- Use Application Insights for monitoring
- Store configuration in app settings
- Use custom domains and SSL certificates
Containers (AKS/ACI)¶
- Use multi-stage Docker builds
- Implement resource limits and requests
- Use Azure Container Registry for images
- Implement proper logging and monitoring
- Use Kubernetes RBAC for security
- Regular security scanning of images
Cost Optimization¶
General Strategies¶
- Right-sizing: Choose appropriate VM sizes
- Reserved Instances: Commit to long-term usage
- Azure Hybrid Benefit: Use existing licenses
- Spot Instances: Use for fault-tolerant workloads
- Auto-shutdown: Automatically stop dev/test VMs
Service-Specific Optimization¶
- VMs: Use B-series for variable workloads
- App Service: Choose appropriate tier for needs
- Functions: Optimize execution time and memory
- AKS: Use spot node pools for batch workloads
- Container Instances: Use for short-running tasks
Monitoring & Management¶
Azure Monitor¶
- Metrics: Performance and health data
- Logs: Detailed operational data
- Alerts: Proactive notifications
- Dashboards: Visualization and reporting
- Application Insights: Application performance monitoring
Common Metrics to Monitor¶
- VMs: CPU, memory, disk, network
- App Service: Response time, throughput, errors
- Functions: Execution count, duration, errors
- AKS: Node health, pod status, resource usage
Common Pitfalls¶
Virtual Machine Issues¶
- Not using managed disks
- Insufficient planning for high availability
- Over-provisioning resources
- Not implementing proper backup strategies
- Ignoring security recommendations
App Service Issues¶
- Wrong service plan tier selection
- Not using deployment slots
- Inadequate scaling configuration
- Missing SSL certificate setup
- Poor application architecture for scale
Container Issues¶
- Large container images affecting startup
- Not implementing resource limits
- Poor secrets management
- Inadequate health check configuration
- Missing monitoring and logging
Azure Functions Issues¶
- Cold start impact on performance
- Timeout issues with long-running processes
- Incorrect trigger configuration
- Not implementing proper retry logic
- Inappropriate hosting plan selection