Tips & Tricks for Provisioning Virtual Machines on Azure

Former Employee

Jun 03, 2025

At the heart of this post is Kairos IMS, an innovative Impact Management System designed to empower human-serving nonprofits and social impact organizations. Co-developed by the Urban League of Broward County and our trusted technology partner, Impactful, Kairos IMS reduces administrative burdens, enhances holistic care, and enables organizations to leverage data for increased agility and seamless service delivery. In this blog series, we’ll take a closer look at the powerful technologies that fuel Kairos IMS, from Azure services to security frameworks, offering insight into how modern infrastructure supports mission-driven impact. Click here to learn more.

Why Deploy Virtual Machines?

Last year, a small nonprofit focused on community health faced a challenge: they needed to test a new donor management system without risking their live database. With a tight budget and no room for error, they turned to Virtual Machines, and it changed everything.

Virtual Machines (VMs) offer nonprofits a cost-effective, secure way to test new tools, deploy updates, and collaborate remotely without putting critical systems at risk. By creating isolated environments, VMs add an important layer of protection: if something goes wrong in one virtual machine, the issue stays contained, leaving the rest of the system safe.

For nonprofits navigating tight budgets and big ambitions, VMs can be game changers. And the example above is just the beginning. In this guide, we’ll walk you through best practices for provisioning virtual machines and help you choose the right setup for your organization’s unique needs.

Choosing The Right Size & Type

Disclaimer: Pricing for Virtual Machines (VMs) can differ based on the region and is contingent upon availability. All listed prices are estimates and may be subject to change without prior notice. For the most accurate and up-to-date pricing, please refer to the official pricing page. The information provided here is accurate as of April 2025.

Azure offers a variety of VM series and sizes tailored to different use cases. Choosing the right one ensures you get enough performance without overspending. Here’s a quick overview of common VM types and when to use them:

B-series (Burstable VMs) – Best for development, testing, or small workloads.

- These are low-cost VMs that accumulate credits when idle and burst (use full CPU) when needed. Ideal for scenarios that don’t need full CPU constantly, such as a low-traffic website, a small database, or a dev/test server.

- B-series VMs are very economical for nonprofits on a tight budget but not suited for sustained heavy CPU usage (they’ll slow down once burst credits are used up).

D-series (General purpose) – Good for many production workloads.

- D-series VMs have a balanced mix of CPU, memory, and disk suitable for web servers, application servers, and databases in production. If you’re hosting a website or an app for your nonprofit and expect steady usage, a D-series (like D2s_v3 or D4s_v3) is a safe choice.

- They’re designed to handle most business workloads reliably.

E-series (Memory optimized) – Use for memory-intensive applications.

- E-series have a higher memory-to-CPU ratio. These are great for large relational databases, in-memory analytics, or applications like SQL Server, SharePoint, or SAP that need a lot of RAM.

- A nonprofit likely uses these only if running big databases or data analytics programs that require extra memory.

F-series (Compute optimized) – Use for CPU-intensive tasks. F-series provide a higher CPU-to-memory ratio. They are suited for workloads that need a lot of processing power but not as much RAM – for example, batch processing, scientific computations, or high-traffic application servers that perform intensive calculations. If your nonprofit does data analysis or simulations (say for research), an F-series might be fitting.

Others (specialty VMs) – Azure also has specialized families:

- N-series (GPU) – for graphics-heavy or machine learning workloads (with NVIDIA GPUs).

- H-series (High Performance Computing) – for extremely compute-intensive tasks (like simulations).

- M-series (High Memory) – for extremely large memory needs (like very large databases).

These are less common for typical nonprofit scenarios, but it’s good to know they exist if you have specialized needs (for example, an N-series VM could help a nonprofit that does AI-driven image processing or video rendering).

Cost Management Tips

Nonprofits often have limited budgets, so it’s important to optimize costs. Azure provides several ways to save money on VMs without sacrificing necessary performance:

Start/Stop VMs and Auto-shutdown: Unlike on-premises servers, you pay for cloud VMs only while they’re running. If you don’t need the VM 24/7, shut it down when not in use (and deallocate it in Azure so you’re not billed for compute). Azure even has an auto-shutdown option to turn off the VM on a schedule (e.g., every night at 7 PM). This is great for dev/test or office-hours-only scenarios. (Note: Even when a VM is stopped, you still pay for the storage of its disk, but that cost is minimal.)

Azure Spot Instances: For certain workloads, you can use Spot VMs to get huge discounts (up to 90% off) normal prices. Spot VMs allow Azure to use spare capacity – the trade-off is Azure can evict (stop) your VM if it needs that capacity back (with a 30-second warning). This is ideal for batch jobs, testing, or workloads that can handle interruptions. For example, if your nonprofit runs a large data analysis job overnight, a Spot VM could be very cost-effective. However, Spot VMs are not recommended for any critical or persistent service (like your main website) because they can be turned off unexpectedly.

Reserved Instances: If you know you’ll need a VM continuously for a long period (one year or three years), consider Azure Reserved Virtual Machine Instances. You commit to a 1-year or 3-year term for a VM and in return get a significant discount (up to ~72% vs. pay-as-you-go prices). This works well for production servers that will be always on. You pay upfront (or yearly) for the reservation, which locks in the lower rate.

Azure Hybrid Benefit: If you use Windows Server or SQL Server, you might already own licenses. Azure Hybrid Benefit allows you to apply existing licenses to Azure VMs and pay a lower rate (essentially, you’re not charged for the license again). Combining Reserved Instances + Hybrid Benefit can yield up to ~80% savings for Windows VMs. Nonprofits that have on-premises licenses or software assurance should investigate this.

Scaling Strategically: Plan your deployment to scale efficiently:

Scale Up vs Scale Out: Scaling up means increasing a VM’s size (more CPU/RAM) when needed, whereas scaling out means adding more VM instances. For example, if your nonprofit’s website is getting popular, you could scale up from a 2-core VM to a 4-core VM or scale out to two 2-core VMs behind a load balancer.

- Scaling out with multiple smaller VMs can be more cost-effective and offers redundancy (if one VM goes down, the others still serve the site). Azure has features like Virtual Machine Scale Sets that automatically add/remove VMs based on demand.

Right-Size Your VM: It’s common to start with a smaller VM and only increase if needed. Azure Monitor (discussed later) can show if your VM is under heavy load. Conversely, if a VM is mostly idle, you might downgrade to a cheaper size. Azure Advisor will recommend cost optimizations such as shutting down underused VMs or rightsizing them.

Use Free and Open-Source Software on Linux: If your workload can run on Linux, consider using Linux VMs to avoid Windows licensing costs. For instance, hosting a website with a LAMP stack (Linux, Apache, MySQL, PHP) on Ubuntu can be cheaper than running a Windows VM with IIS and MS SQL (due to license). We’ll talk more about Linux vs Windows soon.

Azure Credits and Donations: As a nonprofit, check Microsoft’s programs for nonprofits. You may receive Azure credits or grants that can cover a portion of your costs. Always use those first! Additionally, using Azure Cost Management tools, you can set up a budget to get alerted if you approach a monthly spending limit – a good safety net for tight budgets.

By combining these strategies, nonprofits can stretch their cloud budget significantly while still meeting their IT needs. Azure’s own guidance highlights that using reserved instances and existing licenses can drastically cut costs, and that spot instances offer deep discounts for flexible workloads.

Conclusion

In conclusion, you have a better understanding of how you can empower your organization to deploy Aure VMs. We covered what types of Virtual machine sizes and types to consider based on your needs. How you can weigh balancing cost and performance. You learned effective strategies on how you can save money with tips for cost management. If you would like to explore more about your options and perform a deep dive into materials, please see the links below to learn more. With careful planning, smart resource management, and Azure’s built-in tools, your organization can confidently build powerful, cost-effective cloud solutions.