Azure Migrate Processor

What the GBRAM/Core ratio of a VM indicates

A virtual machine (VM) is configured by adding virtual Cores of vCPU and Internal Memory (GBRAM). The ratio of the GBRAM and vCores is an indication for what workloads  this VM is suited best. Some application requires more processing power, the other application needs more Internal Memory and the majority needs both.

In 80% of all configured VM’s the GBRAM/Core ratio is between a factor 2 up to 8. Where the VM’s with a factor 2 are named Compute Optimized and the ones with a factor 8 are named Memory Optimized. The ones in between, around a factor 4 are named General Purpose VM’s.

Microsoft uses this GBRAM/Core ratio to categorize these three main VM Compute categories.

For example, all newer General Purpose D-series have a GBRAM / Core ratio of 4, the Compute Optimized F-series a ratio of 2 and the Memory Optimized E-series a ratio of 8.

The main difference between the different series per category is the release date, indicated by the version number. Version 2 (v2) series are the oldest VM’s. The v4 series were released in the first half of 2020 and still in the roll-out phase to all Azure regions.

Beside the main three categories Azure is offering specialized VM’s for very specific workloads as video rendering, number crunching, scientific calculations etc. The GBRAM/Ratio’s of these series  are all above a factor 4, but these VM’s differentiate because of there processors and motherboard specifications.

Azure is offering more then 30 different VM-series, with more then 150 different VM’s to choose from to run a customers workloads best.

Here is an overview of all Azure VM Series and their GBRAM/Core ratios

In Cloudlab’s Smart Azure Calculator, among other things, automatically matches existing VM configurations with Azure VMs based on the GBRAM / Core ratio in order to be able to make accurate cost estimates.

If you are interested in a deeper dive in how the GBRAM/Core ratio is influencing the pricing see the video: https://youtu.be/hHETJftrbn8