Case Study: Simplifying VM Reservations for Real Azure Savings

Problem

An enterprise Azure environment had a large virtual machine estate with inconsistent sizing and scattered reservation coverage. Azure Advisor showed savings opportunities, but the recommendations needed validation against real workload behaviour.

The risk was buying commitments against a messy estate and locking in spend before the VM families, shutdown patterns, and operating requirements were clear.

Context

The environment included approximately 45 different VM SKUs. That amount of variation made reservation planning harder, reduced flexibility, and made it difficult to explain which workloads were suitable for commitment.

The agency had a large number of Azure virtual machines provisioned across the tenant. Over time, the environment had grown organically, with virtual machines deployed using many different SKUs and sizes.

Non-production workloads also needed attention. Some could be shut down outside business hours or reviewed before any commitment purchase was made.

What was accomplished

Our team reviewed compute spend, analysed usage patterns, checked Azure Advisor recommendations, and separated stable workloads from workloads that needed flexibility.

Key decisions and trade-offs

The main trade-off was commitment versus flexibility. Buying reservations for everything would have looked good in a savings forecast, but it could have created waste if workloads changed.

The challenge was not simply to buy reservations. Buying reservations without first understanding the environment could have locked the agency into the wrong commitment.

The practical approach was to standardise the estate, reserve the stable baseline, and keep some pay-as-you-go headroom for change.

Result

The review identified a six-figure saving opportunity by combining VM family rationalisation, reservation planning, and non-production optimisation.

The saving was not achieved by simply purchasing reservations. It was achieved by preparing the environment so reservations could be used effectively.

The estate moved from a scattered SKU pattern toward a cleaner model where commitment decisions could be explained, governed, and repeated.

Conclusion

Reservations work best after the VM estate has been cleaned up enough for the commitment to make sense.

Many organisations treat reservations as a finance or procurement decision. In reality, reservations are most effective when they are supported by engineering discipline.

A focused Azure cost optimisation review can turn Advisor recommendations into a practical buying plan instead of a risky commitment exercise.