Edit

Find your Azure Local deployment type

Azure Local supports several deployment types across two connectivity modes. This article guides you through three decisions - connectivity, scale, and architecture - and recommends a documentation path based on your answers.

Treat this article as your start here page. When you identify your deployment type, go to the corresponding Connected or Disconnected section in the table of contents.

Overview

Azure Local provides a consistent on-premises experience for critical workloads and Arc services. It supports a wide spectrum of scale points and use cases, from a single machine up to hundreds of machines. You can deploy it in different ways depending on your use case and needs.

The following table summarizes the Azure Local deployment types covered by this guide:

Deployment type Description What's new
Hyperconverged Clusters of 1-16 machines using hyperconverged storage. You can also attach an external SAN for more storage capacity. Hyperconverged what's new
Disaggregated Disaggregated deployments range from a single machine footprint to a maximum of 64 machines that use SAN storage. Hyperconverged what's new
Multi-rack Integrated racks of compute, storage, and networking that expand up to hundreds of machines. Multi-rack what's new
Small form factor Compact deployments for space and power-constrained environments such as retail, branch, and edge sites. Small form factor overview

Disconnected operations is a connectivity mode, not a separate deployment type. You can run hyperconverged and disaggregated deployments in either connected or disconnected mode. Multi-rack and small form factor deployments are supported in connected mode only.

For an in-depth overview of each deployment type, see Azure Local scalability and deployment types.

Step 1: Connected or disconnected?

Pick the connectivity mode that matches your organization's sovereignty, regulatory, and operational requirements.

Choose If your environment...
Connected Has reliable outbound connectivity to Azure. The control plane runs in an Azure cloud region. This deployment mode is standard and supports the broadest set of Azure services.
Disconnected Has strict sovereignty, regulatory, or air-gap requirements that prevent connectivity to the Azure public cloud. The control plane runs locally on a dedicated appliance with a subset of Azure capabilities.

Step 2: What scale do you need?

Estimate the scale point you need today, plus headroom for growth.

Choose If you need...
Single machine to small cluster (1-3 machines) A small footprint for branch, edge, or evaluation. Switchless storage configurations are available at this scale.
Standard cluster (4-16 machines) A typical datacenter or large branch deployment. Requires a physical switch for storage network traffic.
Large cluster (up to 64 machines) Compute and storage scaled independently by using SAN storage.
Rack-scale (hundreds of machines) Hyperscale workloads that need integrated racks with built-in fault tolerance and dedicated network fabric.
Edge / compact Space and power-constrained sites where a compact small form factor appliance is preferred over a server cluster.

Use the Azure Local sizer tool to refine your scale estimate based on workload profile.

Step 3: Which deployment architecture fits?

Match your scale and storage strategy to a deployment architecture.

Choose If you want...
Hyperconverged Compute and storage on the same machines. Simplest operational model and the most common choice. Supports cluster sizes of up to 16 machines for standard clusters and up to 8 machines for rack aware clusters.
Disaggregated Compute and storage scaled independently. Uses SAN storage and supports up to 64 machines. Best when storage requirements grow at a different rate than compute.
Multi-rack Prescriptive, preintegrated racks of compute, storage, and networking. Required for hundreds-of-machines scale.
Small form factor A compact appliance with zero-touch provisioning for edge, branch, and retail sites.

Use your answers from steps 1-3 to find your starting point in the table of contents:

Your result Start here
Connected + Hyperconverged Connected environments > Hyperconverged and disaggregated > Install and deploy (hyperconverged)
Connected + Disaggregated Connected environments > Hyperconverged and disaggregated > Install and deploy (disaggregated)
Connected + Multi-rack Connected environments > Multi-rack
Connected + Small form factor Connected environments > Small form factor
Disconnected + Hyperconverged Disconnected environments > Hyperconverged and disaggregated > Install and deploy (hyperconverged)
Disconnected + Disaggregated Disconnected environments > Hyperconverged and disaggregated > Install and deploy (disaggregated)

For every deployment type, you start with About, then work through Plan, Prepare, Deploy, and (where applicable) Enable. After setup, you find ongoing management and monitoring tasks in the Configure and manage node within your deployment section. For hyperconverged and disaggregated, a single shared Configure and manage node serves both models.

Workload availability by deployment type

Workload availability varies by deployment type. For example, multi-rack and small form factor deployments might support fewer workloads than hyperconverged and disaggregated deployments.

Connectivity mode also affects availability. For the workloads and services available in disconnected mode, see Supported services.

Use the following table to determine which workloads are supported for each deployment type.

Workload Hyperconverged Disaggregated Multi-rack Small form factor
Azure Local VMs Yes Yes Yes No
Azure Kubernetes Service (AKS) Yes Yes Yes Yes
SQL Server Yes Yes No No
Azure Virtual Desktop Yes Yes No No
Microsoft 365 Local Yes Yes No No
GitHub Enterprise Local Yes Yes No No
AI workloads Yes Yes No Yes (preview)
Foundry Local Yes Yes No Yes (preview)
Azure IoT Operations Yes Yes No No

For the latest support information, see the workload overview article linked in the table.

Next steps

  • Last updated on