At the annual Build conference, Microsoft announced the public preview of Azure Compute Fleet, a new service that streamlines the provisioning and management of Azure compute capacity across different virtual machine (VM) types, availability zones, and pricing models to achieve desired scale, performance, and cost.
Compute Fleet enables users to access Azure's quick compute capacity in a specific region by launching a combination of Standard and Spot VMs at the best price and highest availability. Different pricing models, including Reserved Instances, Savings Plan, Spot, and Pay-As-You-Go (PYG) options, can help users achieve better price-performance ratios.
Microsoft's Azure principal product manager, Rajeesh Ramachandran, wrote:
Tell us what you need, capacity and instance-wise, for Standard and Spot VM, and Compute Fleet will create both Standard and Spot VMs from a customized SKU list tailored to your workload requirements with the capability to maintain the Spot VM target capacity in the fleet.
Overview of Azure Compute Fleet (Source: Tech Community blog post)
Users can deploy up to 10,000 VMs with a single API, using Spot and Standard VM types together. Aidin Finn, an Azure MVP, tweeted:
Be handy for those times when I need 10k VMs with a single command.
Deploying Compute Fleet with a predefined fleet allocation strategy allows users to optimize their mix of VMs for capacity and cost efficiency or a balance of both. This will help alleviate concerns about the complexity of determining optimal VM pricing, available capacity, managing Spot evictions (instances that are closed down because another customer reserved them), and SKU availability.
Both AWS and Google Cloud Platform (GCP) provide services like Azure Compute Fleet for managing large deployments of virtual machines and instances across different environments. AWS offers AWS Systems Manager Fleet Manager and Amazon EC2 Fleet. The former provides a centralized interface to manage servers across AWS and on-premises environments. At the same time, the latter allows users to launch and manage a collection of EC2 instances across multiple instance types, availability zones, and pricing models through a single API call.
On the other hand, GCP offers Google Cloud Deployment Manager to automate the creation and management of Google Cloud resources.
Currently, Azure Compute Fleet is available in the following regions: West US, West US2, East US, and East US2. The documentation pages provide more details on Azure Compute Fleet.