Understanding Compute Terms for Azure VMs

• The compute capacity of an Azure VM is measured in either cores or vCPUs, depending on the VM series used.
• These terms are often used inconsistently, even in Microsoft documentation, and have different implications for on-premises virtualization.
• Understanding how these terms are used when describing Azure VMs is important for customers moving workloads to Azure.

Azure VMs are typically priced and configured based on the number of physical processor cores allocated to the VM. However, technologies like Intel Hyper-Threading (HT) can make it appear that a VM has more compute resources than are physically present. Understanding how Azure VMs are typically described by Microsoft and the practices used to allocate compute resources to VMs is essential when comparing Azure VMs to on-premises virtualization technologies, and when planning migration of virtualized workloads to Azure VMs.

Cores and vCPUs within Azure VMs

Each Azure VM host server includes multiple physical processors (or CPUs), and each processor has multiple physical compute cores. How physical cores translate to VM compute resources depends on the generation of host server hardware. The differences can affect the capacity of a set of Azure VMs in unexpected ways, and require an understanding of how terms are used to describe compute resources allocated to Azure VMs.