Choosing Azure VMs to Support Your Workloads

 In Microsoft Azure, Virtual Machines

We’ve talked before about how to choose Azure virtual machines (VMs) that fit your workload and your budget. Selecting the right instance type remains one of the most important decisions you can make when migrating to cloud.

To start, take a look at the page where Microsoft describes the available Azure VM types. Choices include:

  • General purpose VMs. These are size B, Dsv3, Dv3, DSv2, Dv2, Av2, and DC. As general-purpose instances, they balance memory and CPU. Microsoft recommends them for testing and development instances, small databases, and web servers that don’t get a lot of traffic, but this will serve any workload that’s non-critical and has fairly low levels of demand. Once your workload starts having performance needs that require functionality like load balancing, you’ll want to start looking at other instance types.
  • Compute optimized VMs. Azure calls these instances size Fsv2, Fs, and F. As the name suggests, these VMs are designed to ensure good CPU performance and boost the amount of CPU relative to memory. They’ll support web servers that get a bit more traffic, routine batch processes, application servers, and network appliances.
  • Memory optimized VMs. The Esv3, Ev3, M, GS, G, DSv2, Dv2 and instances are the flip side of the compute optimized VMs, offering high amounts of memory compared to CPU. As a result, they’re well suited for applications that use a lot of memory, including caches, databases, and analytics.
  • Storage optimized VMs. The Ls instance type is designed for applications that heavily access disk for data (whether reading it or writing it) rather than storing it in memory. Instances of this size are a good fit for Big Data and any larger database application.
  • GPU. When you’re all about the image, consider instance types NV, NVv2, NC, NCv2, NCv3, and ND. Not intended for general purpose computing, these instance types are built to do the high speed, complex processing needed for intensive graphics calculations needed for video editing. They can also be used to support the inferencing needed by machine learning applications.
  • High performance compute. The H series offers Azure’s top-of-the-line CPUs. High throughput network interfaces help these VMs provide performance needed to support applications like scientific computing.

Choosing the right-size instance requires considering some additional factors besides the number of CPUs and amount of memory:

  • You also need to set up the right size virtual OS disk and add the appropriate number of data disks; different size Azure machines can support different numbers of data disks.
  • You need to decide between either Azure Standard Storage or Premium Storage. As the name suggests, Premium Storage provides better performance, so it should be used for workloads with high I/O demands. Not all Azure sizes support this Premium Storage.
  • You need to review the maximum network interfaces supported by the Azure instance to ensure your choice will support applications needing to interface with multiple virtual networks, if required.
  • You need to consider the region where your Azure instance will reside. Some instance types may not be available in all regions; instance pricing can vary by region as well.

Because matching instance types to workloads is complicated, there are online calculators to help you make the decision. You can also review the Azure Marketplace to find validated virtual machine images preconfigured to support certain workloads.

Another way to make sure you pick the right instance for your workloads? Talk to Prescient Solutions.  As Microsoft Partners, we offer our Azure expertise through IT consulting and managed services in Chicago and Schaumburg. Call us to learn more about using Azure in ways that benefit your budget, your IT operations, and your business.

Recent Posts
*/ 7 Reasons Municipal AgenciesMicrosoft Azure