3.2.1. SQL on VM (IaaS) vs. Managed Instance vs. SQL Database (PaaS)

💡 First Principle: IaaS gives you a virtual machine where you manage everything above the hypervisor—operating system, patches, backups, high availability. PaaS removes infrastructure management so you focus only on data. The question isn't which is "better" but which matches your requirements and what you're willing to manage.

Scenario: Your IT team currently patches Windows servers every month, manages SQL backups to tape drives, and handles storage provisioning manually. They want to reduce this burden but have varying requirements across applications.

SQL Server on Azure VM (IaaS)

• What You Get: A full Windows/Linux VM with SQL Server installed
• What You Manage: OS patching, SQL patching, backups, high availability, storage
• What Azure Manages: Physical hardware, hypervisor, networking
• Use Case:
  • Need OS-level access (install custom software, configure registry)
  • Legacy app requires specific SQL Server version (e.g., SQL 2008)
  • "Lift and shift" migration with minimal changes
• Trade-off: Maximum control / Highest management overhead

Azure SQL Managed Instance (PaaS)

• What You Get: A fully managed SQL instance with near 100% compatibility with on-premises SQL Server
• What You Manage: Database configuration, performance tuning
• What Azure Manages: OS, patching, backups, high availability, storage
• Use Case:
  • "Lift and shift" migration requiring SQL Server Agent, CLR, Service Broker, or cross-database queries
  • Need VNet integration for hybrid connectivity
  • Enterprise features without VM management
• Trade-off: High compatibility / Higher cost than SQL Database

Azure SQL Database (PaaS)

• What You Get: A fully managed, serverless database engine
• What You Manage: Database schema, queries, performance tuning
• What Azure Manages: Everything else (OS, patching, HA, scaling, backups)
• Use Case:
  • New cloud-native applications
  • SaaS applications needing elastic scaling
  • Simple, low-maintenance database needs
• Trade-off: Lowest control / Maximum convenience (zero patching, auto-backups)

Visual: Azure SQL Decision Matrix

Comparative Table: Azure SQL Options

⚠️ Exam Trap: Choosing SQL Database for a "lift and shift" migration without checking feature compatibility is a common mistake. If the app uses SQL Agent, it will break on SQL Database. Always check feature comparison.