Azure Reservations

What are Azure Reservations?

Azure Reservations are a pricing model offered by Microsoft Azure that provide a discount on specific resources in exchange for committing to a one or three year term. The discount applies automatically to matching usage once a reservation is purchased, without needing to make any changes to existing workloads.

Azure Reservations cover a broader range of services than equivalent offerings from other cloud providers, extending beyond virtual machines to databases, storage, and other services.

What Azure Reservations cover

Reservations are available across most major Azure services. Virtual machines are the most common use case, but the model extends to databases like Azure SQL and Cosmos DB, analytics services like Synapse and Databricks, and storage. The full list has grown over time as Microsoft has expanded the program.

For most organizations, VM and database reservations are where the meaningful spend is, and where commitments have the most impact.

How they work

When you purchase a reservation, you're pre-paying for a specific resource and Azure automatically charges you the discounted rate whenever that resource runs, instead of the standard on-demand price. You pay for the reservation whether or not the underlying resources are fully utilized, so understanding your usage baseline before committing is important.

Here are a few specifics worth knowing:

Scope determines where the reservation applies. You can scope a reservation to a single subscription, a resource group, or spread it across all subscriptions within a billing account. Shared scope increases the likelihood the reservation is fully utilized.

Instance size flexibility is available for most VM reservations, meaning the discount applies across different sizes within the same instance family rather than being locked to one specific size. This makes VM reservations more forgiving than they might appear.

Payment options include paying fully upfront for the deepest discount, or monthly with no interest for the term length chosen.

Azure Reservations in action

A company runs a customer-facing application on Azure using D4s_v3 virtual machines in West Europe. The fleet has been stable for over a year with consistent utilization.

The reservation is scoped to the shared billing account level, so any matching VM usage across the organization contributes to utilization.

What to ensure you get right before committing

Azure Reservations cannot be cancelled outright, though they can be exchanged or refunded subject to certain conditions and limits. A few things worth confirming before purchasing:

• Analyze at least 30 days of usage data to identify a stable, consistent baseline.
• Choose the right scope. Shared scope almost always leads to better utilization than single-subscription scope.
• Check whether instance size flexibility is available for the resource type you are reserving. For VMs, it usually is; for databases, it varies by service.
• For database reservations, verify the DTU or vCore tier you are committing to matches actual usage, as database sizing is less straightforward than VM sizing.

Exchanging and refunding reservations

Azure offers more flexibility than AWS Reserved Instances when commitments need to change:

• Reservations can be exchanged for a different reservation of equal or greater value within the same product family.
• Reservations can be cancelled early for a refund, though Azure caps the total refund amount you can claim in a year and plans to introduce a cancellation fee in the future.
• Certain reservation types, including Azure Databricks and some database services, have more limited exchange and refund options.

This flexibility makes Azure Reservations somewhat less risky than AWS Standard Reserved Instances, where exchanges are not available once purchased.