Choose an Azure compute service

Azure provides many ways to host your application code. The term compute refers to the hosting model for the resources that your application runs on. This article helps you choose the right compute service for your scenario and focuses on general-purpose compute offerings.

Architecture

Use the following flowchart to select a candidate compute service.

Download a Visio file of this architecture.

The previous diagram refers to two migration strategies:

• Lift and shift: A strategy for migrating a workload to the cloud without redesigning the application or making code changes. It's also known as rehosting. We recommend that workload teams that perform migrations use a like-for-like approach and defer optimization unless their timeline and budget support redesign for cloud-native functionality. For more information, see Migrate workloads to Azure from other cloud platforms.

• Cloud optimized: A strategy for migrating to the cloud by refactoring an application to take advantage of cloud-native features and capabilities. You can also use this strategy to describe greenfield (new) workloads designed from the start to use cloud-native features.

The output from this flowchart is your starting point. Then evaluate the service to see whether it meets your needs.

This article includes several tables that can help you choose a service. The initial candidate from the flowchart might not suit your application or workload. In that case, expand your analysis to include other compute services.

If your workload consists of distinct compute components, evaluate each application's needs separately. A complete solution can include two or more compute services.

Understand the basic features

If you're unfamiliar with the Azure service that you select in the previous section, see the following overview documentation:

• Azure Virtual Machines is a service that you can use to deploy and manage virtual machines (VMs) inside an Azure virtual network.

• Azure App Service is a managed service for hosting web apps, mobile app back ends, RESTful APIs, or automated business processes.

• Azure Functions is a service that provides managed functions that run based on various trigger types for event-driven applications.

• Azure Kubernetes Service (AKS) is a managed Kubernetes service for running containerized applications. It provides direct access to the Kubernetes API and control plane.

• Azure Container Apps is a managed service built on Kubernetes, which simplifies the deployment of containerized applications in a serverless environment. It doesn't provide direct access to the underlying Kubernetes APIs. Use AKS if you require access to the Kubernetes APIs and control plane.

• Azure Container Instances is a service for running a single container or group of containers in Azure. Container Instances doesn't provide full container orchestration, but you can implement containers without the need to provision VMs or adopt a higher-level service.

• Azure Red Hat OpenShift is a fully managed OpenShift cluster for running containers in production with Kubernetes.

• Azure Batch is a managed service for running large-scale parallel and high-performance computing (HPC) applications.

• Azure VMware Solution is a managed service for running VMware workloads natively on Azure.

Understand the hosting models

For hosting models, cloud services fall into three categories:

• Infrastructure as a service (IaaS) lets you provision VMs along with the associated networking and storage components. You can then deploy any software and applications on those VMs. This model is the most similar to a traditional on-premises environment. Microsoft manages the infrastructure, and you manage the VMs.

• Platform as a service (PaaS) provides a managed hosting environment where you can deploy your application without the need to manage VMs. App Service and Container Apps are PaaS services.

• Functions as a service (FaaS) lets you deploy your code to the service, which automatically runs it. Azure Functions is a FaaS service.

  Note

  Azure Functions is an Azure serverless compute offering. To see how this service compares with other Azure serverless offerings, like Azure Logic Apps for serverless workflows, see Choose the right integration and automation services in Azure.

Cloud compute services range from IaaS to fully managed FaaS. For example, Azure VMs can automatically scale by using virtual machine scale sets. Automatic scaling capabilities are often associated with PaaS or FaaS, but capabilities vary by service rather than by compute category.

A trade-off exists between control and ease of management. IaaS provides the most control, flexibility, and portability, but you must provision, set up, and manage the VMs and network components that you create. FaaS services automatically manage nearly all aspects of running an application. PaaS provides partial management and requires some user configuration.

Notes:

¹ Higher service-level agreement (SLA) that has two or more instances.

² For Azure Functions, the Consumption and Flex Consumption plans are serverless. For an App Service plan, functions run on the VMs allocated for that plan. Choose the right service plan for Azure Functions.

³ When you use durable functions.

⁴ Recommended for production environments. Three in the system node pool and three for each user node pool.

⁵ Three primary nodes and three worker nodes.

⁶ Can scale down to zero after the job completes.

⁷ See Hosts.

For more information, see Choose an Azure container service.

Networking

Your application platform likely needs to interface with networks both as a server for your applications and as a client. For example, it might function as a client to get operating system (OS) updates. You must select a platform that supports both your east-west and north-south traffic requirements.

Notes:

¹ Requires Basic tier or higher, or App Service Environment. See virtual network integration.

² Use App Service Hybrid Connections.

³ Requires an App Service plan, Azure Functions Premium plan, or Azure Functions Flex Consumption plan.

DevOps

Notes:

¹ Options include IIS Express, Visual Studio Code, and other standard development tools based on your application stack.

² Use non-Microsoft tools like Telepresence or mirrord for local-to-cluster debugging.

Team skills and operational overhead

Scalability

Notes:

¹ See Autoscale pods.

² See Scale a cluster automatically to meet application demands on AKS.

³ See Set scaling rules in Container Apps.

⁴ See Scale clusters in a private cloud.

⁵ See VMware NSX.

Built-in multiregion capabilities

All of the application platforms that this decision guide addresses are regional. They require external routing to enable multiregion topologies, such as active-active for resiliency or active-passive for recoverability. When you deploy multiple instances of the application platform, with at least one instance in each region, an external router or load balancer can direct traffic where needed across regions.

Security

Review and understand the available security controls and visibility for each of the following services:

• Virtual Machines for Linux
• Virtual Machines for Windows
• App Service
• Azure Functions
• AKS
• Container Apps
• Container Instances
• Azure VMware Solution

Other criteria

Consider limits and cost

Use the previous comparison tables as context and evaluate the following aspects of the candidate service in more detail:

• Azure pricing and cost calculator
• Azure service availability by region
• Service limits
• SLAs

Specialty workloads

Some workloads have specific requirements and don't typically follow the general recommendations in this application platform selection guide. These workloads are usually software or database products that have characteristics that limit the service options to a narrow set of supported choices.

Contributors

Microsoft maintains this article. The following contributors wrote this article.

• Ayobami Ayodeji | Senior Program Manager
• Jelle Druyts | Principal Service Engineer
• Martin Gjoshevski | Senior Service Engineer
• Phil Huang | Senior Cloud Solution Architect
• Julie Ng | Senior Service Engineer
• Paolo Salvatori | Principal Service Engineer

To see nonpublic LinkedIn profiles, sign in to LinkedIn.

Next step

• Azure compute options

Related resources

• Choose an Azure compute option for microservices
• Technology choices for Azure solutions