Azure FinOps Essentials

In partnership with

Hi there, and welcome to this week's edition of Azure FinOps Essentials! 🎉

In this edition, we’re exploring how to optimize your non-production environments in Azure. Learn how to tailor your resources, leverage dev/test subscriptions, and apply cost-effective strategies to reduce your cloud expenses. We’ll also discuss the trade-offs you might encounter and how to balance cost savings with operational efficiency.

Dive in and start optimizing your non-prod environments today!

Cheers,
Michiel

Introduction

When it comes to cloud environments, there’s one place where your business truly adds value: production. This is where your customers interact with your services, where every action impacts your bottom line. However, rarely does code or changes go straight into production without first being tested and validated. That’s where non-production (non-prod) environments come in—staging, testing, development, and more.

While essential, non-prod environments are often costly to run, maintain, and operate. They can also slow down your time to market and, despite your best efforts, they are never a perfect replica of your production environment. In a perfect world, you might dream of a single, dynamic production environment, enhanced with feature toggles, resilient patterns, and small fault domains. You might even imagine ephemeral feature environments spun up on demand and torn down once testing is complete.

But for many, this remains an idealistic vision. So, how can you manage multiple environments effectively while still keeping costs under control?

Tailoring Non-Prod Environments to Save Costs

Non-prod environments, such as development, testing, and acceptance, often don't need to mirror the full capabilities of your production environment. Understanding this distinction can lead to significant cost savings by optimizing resources that are specific to non-prod scenarios.

1. Optimizing Options

Lets take a storage account as an example:

• Data Redundancy: In production, high data redundancy is crucial for disaster recovery and ensuring data availability. However, in non-prod environments, you can often opt for Locally Redundant Storage (LRS) instead of the more expensive Geo-Redundant Storage (GRS) or Zone-Redundant Storage (ZRS). This simple switch can lead to substantial cost reductions.

• Access Tiers: Similarly, consider downgrading the access tier of your storage accounts. Production environments may require a Hot access tier for frequent access, but non-prod environments can often rely on the Cool tier, which is more cost-effective for infrequently accessed data.

• Premium vs. Standard Storage: Ask yourself if premium storage is really necessary in non-prod environments. For many non-critical tasks, standard storage options will suffice, allowing you to save on the higher costs associated with premium storage.

2. Choosing the Right Service Tiers

There are some other resources which can benefit from different choices.

• Azure Service Bus: In non-prod environments, you may not need the full capabilities of the Premium tier (which costs around $600 per month). The Standard tier, at about $10 per month, might provide all the functionality you need for development and testing, though it lacks features like VNet support.

• API Management: Production might require the Premium version of API Management, which costs approximately $2700 per month. For non-prod environments, the Developer tier, priced under $50 per month, can be a cost-effective alternative, though it comes with limitations that are often acceptable in a non-prod context.

• Virtual Machines: When using VMs in non-prod environments, opt for smaller, less powerful instances. The difference in cost between production-grade VMs and smaller VMs can be significant, especially when scaled across multiple non-prod environments.

Scaling and Scheduling

Non-prod environments typically don’t experience the same load or demand as production environments. Leveraging this fact can unlock substantial savings through intelligent scaling and scheduling strategies.

1. Scaling Down Non-Prod Resources

• Adjust Scaling Rules: Set different scaling rules for non-prod environments. Since the demand is lower, resources can be scaled down significantly compared to production. This might involve reducing the number of VM instances, decreasing the allocated CPU/memory, or lowering the service tier where applicable.

• Use Auto-Scaling: Implement auto-scaling policies that adjust resource allocation based on actual usage. This ensures that non-prod environments only consume what they need at any given time, avoiding unnecessary costs.

2. Turn Off Resources When Not in Use

• Scheduled Shutdowns: One of the most effective cost-saving measures is to turn off non-prod resources when they’re not in use. For example, development and testing environments are often idle during nights and weekends. Implementing a schedule that automatically shuts down these resources during off-hours can cut costs dramatically.

• Scale to Zero: For certain services, consider scaling down to zero instances when no one is using the environment. This approach not only saves on compute costs but also on related services like storage and networking, which can otherwise accrue costs even when idle.

Leveraging Dev/Test Subscriptions for Cost Savings

In addition to adjusting the resources used in your non-prod environments, another effective strategy for reducing costs is to take advantage of Azure’s dev/test subscriptions. These specialized subscriptions offer significant discounts on certain Azure services, making them ideal for environments where uptime and performance aren’t as critical.

1. Understanding Dev/Test Subscriptions

Dev/test subscriptions are designed specifically for non-production environments like Sandbox, QA, UAT, and Development. These subscriptions offer discounted rates on various Azure services, helping you cut costs significantly. However, it’s important to note that these subscriptions do not come with Service Level Agreements (SLAs). This means that Microsoft doesn’t guarantee service uptime or performance, which is why these subscriptions are best suited for environments where high availability is not essential.

• No SLAs: Since dev/test subscriptions lack SLAs, they should not be used for production environments where service reliability and uptime are crucial.

• Dedicated Subscriptions: It’s recommended to use dedicated Azure subscriptions per environment (e.g., one for Dev, one for QA) to better manage resources and costs. This also helps to clearly delineate environments, ensuring that dev/test environments are separate from production.

2. Cost Savings with Dev/Test Discounts

Dev/test subscriptions offer substantial discounts on key Azure resources, which can lead to significant savings in your non-prod environments:

• Virtual Machines: Save around 50% on VMs when using a dev/test subscription. This is particularly beneficial if your non-prod environments require multiple VMs.

• SQL Server: Enjoy a 55% discount on SQL Server instances, making it more affordable to run databases in non-prod environments.

• Azure App Services: Take advantage of discounts on Azure App Services, reducing the cost of hosting and testing web applications.

These discounts can make a meaningful difference in your cloud budget, especially when scaled across multiple non-prod environments.

3. Using the Azure Price Calculator

To see the potential savings, you can use the Azure Price Calculator. This tool allows you to compare the costs of resources under standard production subscriptions versus dev/test subscriptions, giving you a clear view of how much you can save.

4. Combining with Free Services

Don’t forget to explore whether you can substitute paid services with free services in your non-prod environments. For instance, using the free tier of Azure App Services or the always-free database options could further reduce your costs, especially in less critical environments.

Conclusion

Not all environments are created equal, and it’s important to recognize that when managing your Azure resources. By taking advantage of cost-saving options such as dev/test subscriptions, adjusting resource allocations, and utilizing free services, you can make your non-prod environments significantly more cost-effective.

Remember, your cloud bill is a function of usage times rates—by tweaking both, you can achieve substantial savings. However, these optimizations may introduce some complexity into your configurations, reduce efficiency in certain areas, create differences between environments, and potentially impact the developer experience. But it’s crucial to have those discussions and explore where you can optimize.

Balancing cost efficiency with operational needs is key, so take the time to assess your current setup and see where you can make adjustments. Small changes can lead to big savings, making your non-prod environments leaner and more cost-effective without compromising on their core functionality.

The Daily Newsletter for Intellectually Curious Readers

• We scour 100+ sources daily

• Read by CEOs, scientists, business owners and more

• 3.5 million subscribers

Sign up today!

Thanks for reading this week’s edition. Share with your colleagues and make sure to subscribe to receive more weekly tips. See you next time!

P.S. I have another newsletter about GitHub, Azure, and .NET news. Subscribe as well to keep informed: