VM sizes and families
What it is
A VM size is a specific Azure VM configuration: a number of vCPUs, memory, storage-related capabilities, network capacity, and sometimes accelerators. Sizes are grouped into families designed for broad workload types, such as general-purpose, compute-optimized, memory-optimized, storage-optimized, GPU-accelerated, or high-performance compute.
Why it exists
Workloads need different balances of CPU, memory, storage throughput, network capacity, and accelerators. Choosing a size is a capacity and cost decision; it is not just a choice of CPU count.
Key ideas
- Naming is a clue, not a contract: a size name can indicate family, vCPU count, version, CPU vendor, storage support, accelerators, or other features. Consult the specific size specification before assuming what every letter means.
- Family choice: general-purpose sizes balance CPU and memory; compute-optimized favor CPU; memory-optimized favor RAM; storage-optimized favor disk throughput or IOPS; GPU and HPC families serve specialized workloads.
- Burstable sizes: B-family sizes suit workloads with low average CPU use that occasionally burst. They are not automatically appropriate for sustained production load.
- Regional availability and quota: a size may be unavailable in a region or limited by subscription quota. Check both before designing a fleet.
- Storage is separate: the OS/data disk choice and local temporary storage behavior are separate from the VM size, though a size can constrain supported disk features and count.
Northwind example
Northwind begins its low-traffic internal reporting service with a general-purpose size after measuring CPU, memory, disk, and network needs. When month-end reports show memory pressure, the team evaluates a memory-optimized size rather than simply adding more CPU. For a stateless web tier, it considers scaling out with VM Scale Sets instead of repeatedly resizing one VM.
The three pieces (memorize!)
- vCPU count → compute capacity.
- Memory (GiB) → working data and concurrent workload capacity.
- Disk and network limits → possible storage and connection throughput.
How it fits (diagram)
Exam notes
- A resize can require the VM to be stopped/deallocated, causing downtime. Check the supported resize path for the source and target size.
- vCPU quotas are regional and can apply by VM family; capacity can also constrain deployment.
- Temporary storage is not persistent. Put business data on managed data disks or another persistent service.
- AZ-104 care about choosing an appropriate size and knowing family purpose, not memorizing every SKU.
Related
Path MOC · azure-vm · virtual-machine-scale-set · managed-disks
📘 Source: Microsoft Learn — Vm Sizes