Many organizations still grapple with concerns such as compliance and security that weigh down any aspirations to move workloads from on-premises environments.
For these organizations, hyper-converged infrastructure products provide approximate perks of the cloud, such as scalability and high availability. Azure Stack HCI is a hyperconverged infrastructure product from Microsoft that supports Windows and Linux virtual machines. What sets Azure Stack HCI apart from other Hyper-V-based HCI offerings is the operating system and hardware is specifically designed for hybrid environments and is delivered as an Azure service.
What are the Azure Stack HCI hardware requirements?
If you want to deploy Azure Stack HCI, then you have two options. You can purchase a pre-configured system from a Microsoft hardware partner or buy validated hardware and configure it yourself. While it is possible to run Azure Stack HCI on a single node, most of the integrated products require at least two nodes and support up to 16 nodes in a cluster. There are several vendors, including Dell Technologies, HPE, Hitachi and Supermicro, that sell integrated systems validated by Microsoft.
Servers that run Azure Stack HCI must be 64-bit systems, and the cluster nodes must be of the same make and model. Intel systems must use 64-bit Intel Nehalem grade while AMD systems must be AMD EPYC grade or higher with second-level address translation support. To support Intel Optane DC persistent memory, the system must use a second-generation Intel Xeon Scalable CPU.
At the minimum, Azure Stack HCI servers require 32 GB of RAM, plus an extra 4 GB for each terabyte of cached storage. However, any system intended as a virtualization host should have significantly more memory than the stated minimum.
Each node in a cluster will also require an identical storage configuration with at least one physical drive reserved for log files. The log volume must deliver a level of performance comparable to a data volume.
An Azure Stack HCI system has no special networking requirements aside from a network adapter that can be dedicated to cluster management and a switch to allow VLAN traffic. However, each host server must run a virtual machine to act as a network controller. This virtual machine will run the Azure Stack HCI operating system, which consumes between 50 and 100 GB of space.
There is the option to deploy Azure Stack HCI as a stretched cluster with servers residing at two sites. A stretched cluster requires at least two servers per site with a maximum of eight servers per site. Additionally, the servers in a stretched cluster require an additional dedicated volume for replication data.
What are the costs for Azure Stack HCI?
Azure Stack HCI is not licensed the same as Windows Server. Whereas Microsoft sells Windows Server with a perpetual license, Azure Stack HCI is licensed through an Azure subscription as a cloud service.
Like other cloud services, Azure Stack HCI's cost varies by region. As an example, the cost in the Central U.S. region is $10 per physical core per month. A 10-node cluster with 20 cores per node equals 200 cores, which brings the cost to $2,000 per month to license.
This cost covers the host service fee but does not cover the licenses for any virtual machines on the Azure Stack HCI hosts. Microsoft offers an add-on workloads license for $23.30 in the Central U.S. region per physical core per month. This optional license lets you run an unlimited number of Windows Server guests.
Organizations that own a Windows Server Datacenter license with Software Assurance get the host service fee and the Windows Server guest subscription cost waived.
Microsoft offers an Azure services pricing calculator to see how much it will cost to run Azure Stack HCI or any other Azure service. Microsoft also offers a 60-day free trial if you want to download the Azure Stack HCI software and test it in your environment.
How is Azure Stack HCI managed?
You administer Azure Stack HCI much in the same way you work with a traditional Windows Server deployment using many of the same tools.
Like Windows Server, you can manage Azure Stack HCI with the Windows Admin Center, Microsoft System Center or third-party tools.
The Azure portal is also a management option for these systems, but it has better support for Azure Stack HCI.
What are the main features of Azure Stack HCI?
Azure Stack HCI nodes run the Azure Stack HCI operating system, which is essentially a modified version of Windows Server. Azure Stack HCI enlists standard Windows Server services such as Hyper-V and Storage Spaces Direct.
Despite sharing many of the same features with Windows Server 2022, Azure Stack HCI has some capabilities that are not in the latest version of Windows Server:
- Stretched clusters for disaster recovery purposes with Storage Spaces Direct.
- Graphics processing unit acceleration for clustered workloads.
- Kernel-only reboot to reduce the time spent updating servers in a cluster.
- Simplified host networking through Network ATC.
- Storage Spaces Direct on a single server.
- Storage Spaces Direct support for thin provisioning of virtual hard disks.
- Support for dynamic processor compatibility mode for live migrations of VMs between nodes with different CPU generations.
- Cluster-aware update support for feature updates to automate maintenance for servers in the cluster.
Azure Stack HCI also supports integrated driver and firmware updates if the cluster nodes were purchased as an integrated system.
How does Azure Stack HCI differ from Azure Stack Hub?
Microsoft uses the Azure Stack branding on several offerings, such as Azure Stack Edge and Azure Stack Hub. This can lead to confusion as the company modifies product names and updates product families over time.
Microsoft positions Azure Stack HCI for organizations that need on-premises infrastructure to handle virtualized server workloads, such as SQL Server and SharePoint, whether regulatory reasons or other factors prevent a migration to the cloud.
Similar to Azure Stack HCI, Azure Stack Hub is also based on-premises with similar hardware but is designed to give organizations access to Azure services, such as IaaS or SaaS, either directly from Microsoft's cloud platform or in a disconnected state.
As an extension of Azure, Azure Stack Hub gives enterprises a way to build and maintain Azure services that run on their own hardware. Apps developed on Azure Stack Hub will work either in the on-premises infrastructure or in the Azure cloud.
Azure Stack HCI runs on one server with a maximum capacity of 16, while Azure Stack Hub requires four servers at the minimum with a maximum of 16 nodes.
You can manage and work with the software-defined technologies in Azure Stack HCI, such as Hyper-V and Storage Spaces Direct, while there is no direct administration in Azure Stack Hub.
For a look at more differences between these on-premises products, check this link.