It’s been more than a month since Azure Stack HCI version 22H2 became generally available. With this latest update for Microsoft’s hyper-converged infrastructure cluster solution, there are some significant changes in storage network capacity and network equipment requirements. In this article, I would like to give you a quick overview of the changes IT pros need to take into consideration to implement the latest version of Azure Stack HCI.
With the release of Azure Stack HCI version 21H2 last year, Microsoft recommended a 25 Gbps bandwidth between all storage network interfaces, even in smaller clusters. With Azure Stack HCI version 22H2, this recommendation has been removed. The minimum interface bandwidth for Azure Stack HCI storage interfaces is now at 10 Gbps.
Depending on the cluster size, it often makes sense to have a 25 Gbps bandwidth per interface, especially if you’re using flash storage with clusters above two to three nodes. For medium to larger clusters, a 10 Gbps RDMA (remote direct memory access) connection often becomes a bottleneck for storage replication.
With the new features coming to Azure Stack HCI version 22H2 such as enhanced Software Defined Networking, Microsoft introduced additional requirements for the switches you use to interconnect Azure Stack HCI nodes. Please be aware that not all of these requirements are necessary for all use cases and scenarios, but if your switch does not support required features, you could encounter some issues and Microsoft may not be able to support you with the troubleshooting.
If you want to ensure that all network switch requirements are met, you should either select a switch that Microsoft tested or use an integrated solution for Azure Stack HCI. The following vendors have worked with Microsoft to confirm that their switches support the latest Azure Stack HCI requirements:
I will focus on the new physical network requirements for Azure Stack HCI version 22H2 from now on.
According to Microsoft, a minimum of three Class of Service (CoS) priorities are required without downgrading the switch capabilities or port speed. If your device does allow ingress Quality of Service (QoS) rates to be defined, Microsoft recommends not to configure ingress rates or configure them to the exact same value as the egress (ETS) rates.
Link Layer Discovery Protocol (LLDP) allows organizations to define and encode their own custom Type-Length Values (TLVs) requirements, which are called Organizationally Specific TLVs. As Microsoft explained, all of them start with an LLDP TLV type value of 127.
The following table shows which Organizationally Specific Custom TLV (TLV Type 127) subtypes are required for Azure Stack HCI version 22H2
This is a new requirement starting with Azure Stack HCI version 22H2. As explained by Microsoft, “the maximum transmission unit is the largest size frame or packet that can be transmitted across a data link.” Starting with Azure Stack HCI version 22H2, a range of 1514 – 9174 is now required for SDN encapsulation.
This is another new requirement for Azure Stack HCI version 22H2. Border Gateway Protocol (BGP) is a routing protocol allowing the exchange of routing and reachability information between two or more networks. “Routes are automatically added to the route table of all subnets with BGP propagation enabled. This is required to enable tenant workloads with SDN and dynamic peering,” Microsoft explained.
This is the last new requirement that Microsoft implemented with Azure Stack HCI version 22H2. “The DHCP relay agent is any TCP/IP host which is used to forward requests and replies between the DHCP server and client when the server is present on a different network. It is required for PXE boot services,” Microsoft explained.
With these new requirements, some switches you may have will no longer be officially supported. If you are using them within a test or demo environment, you should be fine. In a production environment, however, you should definitely ensure that your switches support the required features.
When using a switchless design for your storage interfaces, the requirements may change too as your switch does not need to support RDMA. In most cases and with most vendors, your switches will only need an update of their firmware and/or operating system.
The network requirements changes for Azure Stack HCI version 22H2 are quite significant overall, and they may have an important impact on network equipment costs and acquisitions. As Microsoft is still updating the documentation, we should eventually get more details on which requirements are necessary for specific scenarios and deployed architectures.