How to Select Hardware for Azure Stack HCI
Azure Stack HCI is Microsoft’s hyper-converged infrastructure cluster solution for hosting virtualized Windows and Linux workloads. In this guide, we’ll tell you everything you need to know about the network components and hardware required to run virtualized workloads on-premises with Azure Stack HCI.
Table of Contents
- How to select the correct network components for Azure Stack HCI
- How to use the Azure Stack HCI catalog to select the correct server hardware
How to select the correct network components for Azure Stack HCI
Before we start with the actual server hardware you need for Azure Stack HCI, we need to discuss the required network components. What you need to consider is whether to order a new switch for your Azure Stack HCI setup or not, because there are a lot of new features used in networking with Azure Stack HCI.
You have new technologies like Remote Direct Memory Access (RDMA) added with Azure Stack HCI and Storage Spaces Direct. With larger Azure Stack HCI clusters, it could also be necessary to upgrade your storage network to 25 Gbit/s or even more.
Let us look a bit deeper into the RDMA topic: RDMA comes with two protocol options, the Internet Wide Area RDMA Protocol (iWARP) and RDMA over Converged Ethernet (RoCE). If your network adapters only support RoCE and you are not able to use iWARP, your switches need to support Data Center Bridging (IEEE 802.1Qbb).
The good news here, if you are planning for a cluster with three or fewer nodes, you do not need to have a switch for the storage network adapters. Azure Stack HCI supports daisy-chaining of network adapters, as shown in the picture below.
As you can see, daisy-chaining basically links the network adapters to build a perfect circle and keep the storage communication up and running. Such a configuration simplifies the network setup for small clusters.
How to use the Azure Stack HCI catalog to select the correct server hardware
Before you select the actual hardware, you need to know that Azure Stack HCI is only supported on certified hardware from Microsoft OEM partners. There are two kinds of certified systems: validated node configurations and integrated node configurations.
Azure Stack HCI validated node vs. integrated node configurations
The main difference between validated nodes and integrated nodes is the installation of Azure Stack HCI and the support integration with the partner. With validated systems, you need to install Azure Stack HCI yourself; while an integrated system comes with Azure Stack HCI preinstalled on the hardware.
In addition, the support is different. While integrated systems have a joint support model between Microsoft and the hardware partner, validated systems don’t have that strong tie between both partners. And hardware and software support are separated.
Microsoft recommends using integrated systems due to the easier integration. Within the Microsoft Azure Stack HCI hardware catalog, you can select between both options and see the available hardware partners.
On the page, you can browse and filter for the components and system configuration you prefer.
On the top right, you can open a small wizard which helps you to choose the right configuration.
Using Microsoft’s Azure Stack HCI sizing tool
Before you begin, you should use the Azure Stack HCI sizing tool provided by Microsoft.
You can find the sizing tool on Microsoft’s Azure Stack HCI solutions website.
- Create a new project and start your configuration. First, you begin with the selection of the hardware components, for example, the CPU.
- After you select your hardware, you need to add the workloads you want to run on the Azure Stack HCI cluster.
- Within the last step, you should see the results and proposed system.
In any case, the Microsoft Azure Stack HCI sizing tool is not fully integrated into any vendor sizing. You should work on additional sizing and planning with your preferred hardware partner.
Microsoft’s Azure Stack HCI planning wizard
Now let’s go back to the planning wizard. Again, in the first step, you decide on a validated or integrated system.
- For the next step, you can filter for regional availability and up to three hardware partners.
- You select the workload type to run on the servers. Now, you have six options to choose from, but these workloads are expected to expand in the future.
- After you chose the workload, you need to choose the form factor of the system. If you have never worked with ruggedized servers, those are mostly built for environments with ruthless conditions such as oil rigs, mining, construction sites, or military operations.
- Based on the selections you made, you will see the applicable servers.
Again, for optimal sizing, please contact your preferred hardware partner.
Azure Stack HCI is Microsoft’s new infrastructure-optimized operating system. With that, there are some challenges that may come up for a classic data center and hardware partner, such as the tight integration between the operating system and the hardware. The best example here is Remote Direct Memory Access and the integration with the SMB Direct protocol.
With the new bandwidth limitations as well as the requirements for Data Center Bridging, changes may be required within your network. But in the end, as soon as you have determined the hardware and infrastructure refresh required to use Azure Stack HCI, you will have a great and modern, infrastructure-focused operating system to run your virtualized Windows and Linux workloads on-premises.
More in Microsoft Azure
How to Migrate an Azure ExpressRoute Connection￼
Nov 18, 2022 | Flo Fox
Microsoft Partners with Nvidia to Build Azure-Powered AI Supercomputer
Nov 16, 2022 | Rabia Noureen
How to Enable AWS Direct Connect Redundancy Using Azure ExpressRoute
Oct 28, 2022 | Flo Fox
Microsoft Confirms Customer Data Breach Caused by Misconfigured Server
Oct 20, 2022 | Rabia Noureen
Microsoft Announces New Azure DDoS IP Protection SKU for Small Businesses
Oct 20, 2022 | Rabia Noureen
Azure Firewall Basic Now Available in Preview for Small Businesses
Oct 10, 2022 | Rabia Noureen
Most popular on petri