Azure SQL Database vs Managed Instance vs SQL on VM

Azure SQL Database is a fully managed PaaS for single databases; Azure SQL Managed Instance provides near-100% compatibility with on-premises SQL Server instance features; and SQL on VM is IaaS, giving you full OS access. Choose based on your need for control versus management overhead.

When should you choose SQL Server on Azure VMs?

Think of SQL Server on Azure VMs as the 'do-it-yourself' option. This is Infrastructure as a Service (IaaS), meaning Microsoft gives you the virtual hardware and the OS, but you're the boss of everything inside. You have full administrative access to the operating system, which is a non-negotiable requirement if your application depends on specific OS-level registry keys or third-party software installed on the same server.

However, with great power comes great responsibility. You are responsible for patching the OS, managing backups, and configuring high availability. If you're studying for the AZ-900, remember that this is the only option of the three that provides full OS access. It's the ideal choice for 'lift-and-shift' migrations where you can't change the application code and need the environment to mirror your on-premises setup exactly.

What makes Azure SQL Database the go-to for modern apps?

Azure SQL Database is a pure Platform as a Service (PaaS) offering. We call it 'database-as-a-service' because you don't worry about the server, the OS, or the SQL installation—you just create a database and start querying. It's designed for modern, cloud-native applications that need to scale rapidly without the headache of manual administration.

One of the biggest talking points for the AZ-900 exam is the difference between a single database and elastic pools. A single database is great for one app, but if you have 50 databases with varying usage patterns, an elastic pool allows them to share a set of resources (eDTUs or vCores). This prevents you from over-provisioning every single DB, saving you a significant amount of money while maintaining performance during peak loads.

Why is Azure SQL Managed Instance the 'middle ground'?

If Azure SQL Database is too restrictive and a VM is too much work, Azure SQL Managed Instance (MI) is your sweet spot. It's still PaaS—meaning Microsoft handles the patching and backups—but it provides nearly 100% compatibility with the SQL Server on-premises engine.

Why does this matter? Because MI supports instance-level features that the single database model doesn't, such as SQL Server Agent for scheduling jobs, Linked Servers, and Service Broker. If you have a legacy database that relies on these features but you still want to get rid of the burden of managing a VM, Managed Instance is the answer. It allows you to migrate complex on-premises workloads to the cloud with minimal code changes.

How do you decide between a Single Database and an Elastic Pool?

This is a classic exam scenario. Imagine you're managing a SaaS application where every customer gets their own database. Some customers are huge and hit the DB hard; others barely log in. If you provision each database individually for peak load, you'll waste 80% of your budget on idle resources.

By using an elastic pool, you buy a 'bucket' of resources that all databases in the pool draw from. When Customer A is quiet, Customer B can use that extra horsepower. We always recommend looking at your workload volatility; if your databases have unpredictable, staggered peaks, elastic pools are the most cost-effective route. If you have one massive, consistent workload, stick to a single database with a dedicated service tier.

Which migration path is right for your on-premises workload?

Choosing a migration path depends on how much you're willing to change. If you need to move your data in a weekend with zero code changes and require full control over the OS, go with SQL on VM. It's the fastest path to the cloud, though it leaves you with the most maintenance work.

If you want to modernize and reduce operational overhead, aim for Managed Instance. It handles the heavy lifting of the OS while keeping your instance-level features intact. For those building from scratch or willing to refactor their apps for maximum scalability and lowest cost, Azure SQL Database is the gold standard. In the real world, we often see companies start with a VM and gradually migrate toward PaaS as they modernize their architecture.

How do these concepts appear on the AZ-900 exam?

The AZ-900 doesn't expect you to be a Database Administrator, but it does expect you to match the service to the business requirement. You'll likely see questions asking which service provides 'full OS access' (VM) or which is best for 'minimizing management overhead' (SQL Database). Understanding the trade-off between control and convenience is the key to scoring high in the Azure workload domain.

To really nail these distinctions, you need a lot of repetition. That's why we provide 1,000 expert-curated Microsoft Azure Fundamentals (AZ-900) practice questions at Cert Sensei. We don't just give you the answer; we provide detailed expert reasoning for every choice and domain-level analytics so you can see exactly where your knowledge gaps are before exam day.

❓ Frequently Asked Questions