Azure Stack Portfolio

Not every workload can run in the public cloud. Regulatory restrictions may require data to stay in a specific facility. Remote locations (oil rigs, military bases, factory floors) may have unreliable or no internet connectivity. Some industries require extreme data sovereignty. Azure Stack is a family of products that brings Azure services and management to environments where the public cloud cannot be used — while maintaining consistency with Azure APIs, tooling, and governance practices.

The Azure Stack portfolio has three distinct products: Azure Stack Hub — an on-premises integrated system that runs a subset of Azure services in your own data centre. Azure Stack HCI — a hyperconverged infrastructure (HCI) operating system for running virtualised workloads on commodity hardware with Azure management. Azure Stack Edge — a hardware appliance with GPU/FPGA for edge AI inference and data transfer to Azure, designed for locations with limited connectivity. Each targets a different primary use case.

Azure Stack Hub is an integrated system purchased from hardware partners (Dell, HPE, Lenovo) that runs a curated subset of Azure IaaS and PaaS services locally. It supports Azure VMs, AKS, Azure App Service, Azure Functions, and Azure SQL Database on-premises. Operators can build a private marketplace of approved images and services. It is licenced and billed per unit and can operate in connected (to Azure for updates and licencing) or disconnected (air-gapped) mode.

In connected mode, Azure Stack Hub receives updates automatically from Azure, supports Azure Marketplace image syndication, and uses Azure Active Directory for authentication. In disconnected mode (also called air-gapped), the system has no internet connectivity — useful for classified government networks and secure industrial environments. Disconnected mode requires a local identity provider (AD FS) and manual update processes via an update package downloaded and transferred on physical media.

Azure Stack HCI is a hyperconverged infrastructure solution that runs Windows Server Hyper-V virtualisation and Software-Defined Networking on validated commodity hardware clusters (2–16 nodes). It is billed as an Azure service per physical core per month and requires a connection to Azure for billing, monitoring, and updates (it can tolerate up to 30 days offline before licensing checks expire). Azure Stack HCI is ideal for organisations that want to modernise on-premises virtualisation while gaining Azure management, Arc integration, and AKS on-premises.

Common Azure Stack HCI use cases: VDI — host Azure Virtual Desktop session hosts on-premises for latency-sensitive users. SQL Server — run SQL on high-performance local NVMe storage with Storage Spaces Direct replication. Kubernetes on-premises — run AKS on Azure Stack HCI (AKS hybrid) for containerised workloads without going to the public cloud. Branch office consolidation — replace aging VMware or Hyper-V clusters with a modern, Azure-managed HCI solution. Integration with Arc and Azure Monitor provides cloud-style visibility.

Azure Stack Edge is a physical hardware appliance (shipped by Microsoft) for running compute at the edge — locations with limited or no cloud connectivity. Models range from the Edge Mini R (ultra-portable, rugged) to the Edge Pro 2 (rack-mounted, GPU) to the Edge Pro R (rugged rack for harsh environments). The primary use cases are AI inference at the edge (e.g. defect detection on a factory camera feed with GPU-accelerated ML models) and data pre-processing and transfer (filter and compress IoT data before sending to Azure).

Azure Stack Edge devices function as a data boundary gateway. Local data is processed (transformed, filtered, enriched) by containerised modules running on the device, and only the relevant results are sent to Azure Blob Storage or Azure IoT Hub. This reduces bandwidth requirements in remote locations where sending raw sensor data would be impractical. The device also provides a Network File System (NFS/SMB) gateway that locally caches data from Azure Files shares, enabling on-premises machines to access cloud data without constant internet connectivity.

Beyond the Azure Stack portfolio, Microsoft operates national cloud deployments of Azure for specific geographies with strict sovereignty requirements. Azure Government (US) is operated by Microsoft employees who are US citizens and is FedRAMP High certified. Microsoft Azure operated by 21Vianet (China) is a separate Azure environment operated by a local entity. Azure Germany (historical) was operated by T-Systems as a data trustee. These are not Azure Stack products but serve the same need for isolated cloud environments.

Decision tree for selecting the right product: Need Azure IaaS/PaaS APIs on-premises or in a sovereign data centre? → Azure Stack Hub. Modernising on-premises virtualisation (Hyper-V) with Azure management at lower cost? → Azure Stack HCI. Need AI inference or data pre-processing at a remote/disconnected edge location? → Azure Stack Edge. All three integrate with Azure Arc, Azure Monitor, and Microsoft Defender for Cloud once connected, providing a unified management experience from the Azure portal.

All three Azure Stack products integrate with Azure Arc to project their resources into ARM. Azure Stack HCI clusters are automatically Arc-enabled during setup. Azure Stack Hub VMs can be Arc-enabled using the standard Connected Machine agent. Azure Stack Edge IoT modules can publish telemetry to Azure IoT Hub and be managed through Arc. This integration means you can apply Azure Policy, view Defender recommendations, and monitor all Stack resources from the unified Azure portal alongside your public cloud resources.

Test your understanding of Microsoft Azure Fundamentals (AZ-900) concepts from this lesson.

In this lesson you learned: Azure Stack Hub brings Azure IaaS/PaaS APIs to on-premises or disconnected data centres, Azure Stack HCI modernises on-premises Hyper-V virtualisation with Azure management and Arc integration, and Azure Stack Edge runs GPU-accelerated AI inference and data pre-processing at remote edge locations. Next up we explore multi-cloud strategies and how Azure tools enable governance across Azure and other cloud providers.