Not too long ago, companies were shouldering massive upfront costs for on-premises servers, only to hit scalability limits every time they wanted to grow. Enter cloud services. It was a game-changer, promising flexibility, predictable pay-as-you-go pricing, and no more dreaded hardware refreshes every five years.
However, managing cloud costs turned out to be far from straightforward. According to McKinsey, while cloud computing can save you money when done right, many companies saw an increase in cloud spend by 20-30% each year.
With macroeconomic pressures and everyone being asked to ‘do more with less’, IT leaders are under pressure to objectively look at their Azure cloud spend and optimise costs to free up resources for key initiatives.
Some of that cloud spending can be an indicator of ‘healthy growth’, meaning you’re spending to meet the growing demands and needs of your company. But more often than not, organisations treat Azure as a ‘set-and-forget’ platform and costs end up spiralling without regular housekeeping.
So in this post, let’s take a look at 6 practical ways to optimise your Azure cloud costs.
1. Stop paying for resources you don’t need
Sometimes the simplest of fixes can lead to a significant reduction in costs. And this might sound obvious but you’d be surprised how many companies still don’t do this: don’t pay for resources you’re not using.
A common starting point is to check for underutilised or idle Virtual Machines (VMs). The Microsoft Cost Management Dashboard and Azure Advisor often flag resources like orphaned disks, which means you have disks sitting idle that are not attached to a VM that can accumulate when housekeeping isn’t up to par.
You might have provisioned a few VMs when demand was higher but forgotten to turn it off when demand went down. In some cases, regulations require data to be retained for a while before deletion. Whatever the reason, do a quarterly audit to make sure you turn off resources when you don’t need them anymore.
2. Optimise usage with rightsizing, automation, and tier adjustments
We’ve seen instances where many workloads, including dev and test environments, stay running 24/7, over the weekends and on nights, when they’re not being used. Why don’t you shut it down so you pay only for the hours you actually use it?
Azure has built-in auto-scheduling features, so you don’t have to do this manually. You can schedule VMs to shut down at night or on weekends and restart in the morning, or only activate when someone logs in if this doesn’t compromise your service quality.
Another trick? If your workloads allow it, downgrade to a lower-tier VM (fewer cores, less RAM) or switch to a cheaper storage option, like locally redundant cool storage instead of globally redundant hot storage. Just make sure you’re okay with the trade-offs in performance.
3. Use Azure Savings Plans for longer commitments
While Pay-As-You-Go (PAYG) is an attractive option, giving you the flexibility to scale up or down as required, committing to a longer timeframe (1 or 3 years) can give more cost savings.
If your Azure environment is fairly stable and you can predict demand for the next one to three years, opting for Azure’s Savings Plan or Reserved Instances might be a good fit.
With an Azure Savings Plan, you commit to an hourly spend on eligible compute resources across all Azure regions either for 1 or 3 years. For example, if you consistently spend at least £10 every hour, but your usage comes from different resources and/or different data centre regions, then a Savings Plan would be a good option to reduce costs.
4. Azure Reserved Instances for stable workloads
Reserved Instances are another option for predictable workloads.
When you opt for reserved instances, you’re committing to using a specific type of compute instance or instance family in a specific Azure region for a set period.
For example, let’s say you commit to using a B8as_v2 VM in the UK South region. The PAYG price is $211.99. If you choose a 1-year reserved instances plan, you get a 38% discount and with a 3-year savings plan, you get a 57% discount. The discount will vary based on the type of VM you choose, with some giving even higher discounts than this example.
When should you pick Reserved Instances over a Savings Plan?
If your workloads are steady and don’t need to shift between regions or VM types, Reserved Instances are perfect. But if you need flexibility across regions or frequently adjust VM families, but know you’d spend a certain number of hours on Azure, a Savings Plan could be a better fit since it applies to eligible resources globally.
5. Tag resources for a clear view
If you want to keep costs under control, you’ve got to know exactly where the money’s going—whether that’s a department, business unit, or even individual teams.
That’s where resource tagging in Azure comes in handy. Tagging lets you assign each resource to a specific owner or department, so costs don’t just disappear into a general “IT” bucket.
With the right tags in place, you’ll be able to see exactly who’s using what and make them responsible for their own spending.
6. Rearchitecting the applications
Another culprit for not seeing optimum costs on Cloud – rehosting applications. For various reasons, you may have ‘lifted and shifted’ applications to the cloud when you migrated. This would have saved you some costs initially but will cost more in the long run.
Applications that are rehosted are often incompatible with cloud-native features such as containerised workloads, and auto-scaling, which are key to controlling costs. Or they won’t be flexible or scalable enough in the cloud based on data needs.
So once you’ve migrated and settled in, it’s good to take a look at refactoring your applications to prevent cloud services from stagnating and costs spiralling.
Optimizing Azure costs doesn’t have to be overwhelming; Often a few changes can make a noticeable difference. Just keep in mind that the cloud isn’t something you can “set and forget.” With a bit of regular housekeeping, an eye for automation, and a strategy for commitment, your Azure environment can be both cost-effective and efficient.
