Azure SQL — Types & Options

Overview of Azure SQL Options

Azure provides three main ways to run SQL workloads in the cloud, each targeting different migration and management needs:

Azure SQL Family
├── Azure SQL Database          → Fully managed, cloud-native (PaaS)
│   ├── SQL Databases
│   ├── Hyperscale Databases
│   └── Elastic Pools
├── Azure SQL Managed Instance  → Near 100% SQL Server compatibility (PaaS)
│   ├── SQL Managed Instances
│   └── Instance Pools
└── SQL Server (IaaS)           → Full control on Virtual Machines
    ├── SQL Server Instances
    └── SQL Server on Azure VMs

---

1. Azure SQL Database

Type: Platform as a Service (PaaS) Best For: Modern cloud-native applications

A fully managed relational database service where Azure handles all infrastructure, patching, backups, and HA automatically.

1a. SQL Databases

• The most common Azure SQL option
• Single database with its own dedicated resources
• Azure manages the OS, engine, backups, and failover
• Supports latest SQL Server features

Feature	Detail
Management	Fully managed by Azure
Compatibility	Subset of SQL Server features
Scaling	Manual or Auto-scale (serverless)
Use Case	New cloud-native apps, SaaS apps

Compute Tiers:

Tier	Description
Provisioned	Fixed resources allocated, billed per hour
Serverless	Auto-scales compute, pauses when idle, billed per use

---

1b. Hyperscale Databases

• Designed for very large databases (up to 100 TB+)
• Decouples storage and compute — scale independently
• Fast backups and restores regardless of data size
• Supports multiple read replicas for scale-out reads

Feature	Detail
Max Storage	100 TB+ (auto-grows)
Backup	Near-instantaneous (snapshot-based)
Read Scale-Out	Up to 4 named replicas
Use Case	Large OLTP workloads, data-heavy apps

1c. Elastic Pools

• A shared pool of resources (eDTUs or vCores) across multiple databases
• Databases in a pool share compute & storage but have their own data
• Cost-effective when databases have variable, unpredictable usage patterns

How It Works

Elastic Pool (e.g., 100 vCores shared)
    ├── Database A  (using 40 vCores right now)
    ├── Database B  (using 10 vCores right now)
    ├── Database C  (using 5 vCores right now)
    └── Database D  (idle)

• When A is busy, B/C/D use fewer resources — and vice versa
• You pay for the pool, not each DB individually

Feature	Detail
Resource Sharing	Yes — shared eDTUs/vCores
Cost Model	Pay per pool, not per DB
Max DBs per Pool	Up to 500
Use Case	SaaS apps with many tenants, dev/test environments

1d. SQL Logical Servers

• A logical administrative container for Azure SQL Databases
• Not a physical server — it’s a management boundary
• Used to group databases and manage:
  • Firewall rules
  • Admin login credentials
  • Azure AD authentication settings
  • Auditing and threat detection policies

SQL Logical Server (my-sql-server.database.windows.net)
    ├── Database 1
    ├── Database 2
    └── Database 3

2. Azure SQL Managed Instance

Type: Platform as a Service (PaaS) Best For: Lift-and-shift migration of on-premises SQL Server

Provides near 100% compatibility with on-premises SQL Server while still being a fully managed PaaS service.

2a. SQL Managed Instances

• A full SQL Server instance deployed and managed by Azure
• Deployed inside a Virtual Network (VNet) for isolation
• Supports SQL Server Agent, CLR, linked servers, cross-database queries
• Ideal for migrating legacy enterprise applications without code changes

Feature	Detail
Compatibility	~100% SQL Server on-premises
Deployment	Inside Azure VNet
Management	Fully managed (Azure handles patching/backups)
Use Case	Lift-and-shift from on-prem SQL Server

---

2b. Instance Pools

• A pre-provisioned pool of compute resources shared across multiple Managed Instances
• Enables deploying smaller Managed Instances (2 vCores) cost-effectively
• Resources are shared similar to Elastic Pools but for Managed Instances

Feature	Detail
Resource Sharing	Yes — shared vCores across instances
Minimum vCores	2 vCores per instance
Use Case	Multiple small managed instances, dev/test

3. SQL Server on Azure (IaaS)

Type: Infrastructure as a Service (IaaS) Best For: Full control, specific OS/SQL versions, unsupported features

You manage the VM; Azure manages only the physical hardware.

3a. SQL Server Instances

• SQL Server installed and managed by you on Azure Virtual Machines
• Full control over SQL Server version, edition, and configuration
• You handle patching, backups, HA, and OS updates

---

3b. SQL Server on Azure VMs

• Pre-built Azure Marketplace images with SQL Server pre-installed on Windows/Linux VMs
• Azure provides a SQL IaaS Agent Extension for:
  • Automated backups
  • Automated patching
  • License management (Azure Hybrid Benefit)

Feature	Detail
Control	Full (OS + SQL Server)
Compatibility	100% — any SQL Server feature
Management	Mostly manual (you own OS + SQL)
Use Case	Unsupported features, custom configs, SSRS/SSIS

---

How to Choose

Is your DB > 4 TB?
    └── YES → Hyperscale Database

Do you need 100% SQL Server compatibility?
    └── YES → SQL Managed Instance

Do you have many DBs with variable usage?
    └── YES → Elastic Pool

Do you need full OS/SQL control?
    └── YES → SQL Server on Azure VM

Default choice for new cloud apps?
    └── Azure SQL Database (Single)