Underneath the hood of the Windows Server 2019 cluster sets feature are a number of technologies working in concert to boost flexibility of important workloads and make life easier for administrators.
A cluster set is a combination of elements that work together to connect multiple clusters across different locations for less friction during the migration process and better overall VM performance due to enhanced resource sharing. Before testing or adopting Windows Server 2019 cluster sets, learn about the main elements that make up this new clustering arrangement.
Multiple components form the fabric of a cluster set
Member clusters. Each individual cluster is called a member, and a cluster set is a fabric of member clusters. Administrators configure and provision member clusters to run workloads based on VMs and storage technologies, such as Storage Spaces Direct. VMs can move freely among member clusters for tasks, including workload balancing and regular maintenance. Member clusters do not handle management tasks at the greater fabric level.
Management cluster. This cluster is distinct from the member clusters that host virtualized workloads. As the name implies, the purpose of this cluster is to operate the management functionality of the cluster set. When the administrator builds the management cluster, it creates a unified storage namespace that spans the cluster set atop the Infrastructure Scale-Out File Server (SOFS), a new role in Windows Server 2019. A management cluster adds to the overall resiliency to ensure continuous availability and support for cluster sets.
Cluster Set Namespace SOFS. Storage poses some unique challenges for extended clusters, which require a specialized storage server to deliver flexibility and performance from server message block (SMB) storage resources. A Cluster Set Namespace SOFS gives SMB workloads access to SMB shares hosted on member clusters. The SOFS has almost no impact on storage I/O because each client node caches SMB activity and updates as needed.
Cluster set components. A cluster set consists of a cluster set master (CS-Master) and a cluster set worker (CS-Worker). A CS-Master handles communication among member clusters. Microsoft designed the CS-Master for high availability (HA) in the face of failures in management cluster nodes and member clusters. Each member cluster has a CS-Worker, which interacts with the common CS-Master to assist in certain operations, such as determining the placement of a VM in the cluster.
Fault domains and availability sets. An administrator shapes the fault domains, or the collection of hardware and applications that could fail together, in a Windows Server 2019 cluster set. On the flip side of fault domains are availability sets, which the administrator creates to produce redundancy of clustered workloads across fault domains. This practice keeps VMs distributed on two or more availability sets that are not in the same fault domain to maintain HA.