Microsoft Windows Server has long supported a variety of methods for replicating data from one cluster to another....
It was previously possible to perform cluster-to-cluster replication at the SAN level. Likewise, the Hyper-V replication feature can be used to replicate a virtual machine from one clustered Hyper-V deployment to another. But with Windows Server 2016 Storage Replica, Microsoft has made cluster-to-cluster replication much more versatile.
The Storage Spaces feature, introduced in Windows Server 2012, has been extended and renamed in Windows Server 2016 to Storage Spaces Direct. It is designed for use in hyper-converged infrastructure topologies, where each cluster node includes its own local storage. Data can be replicated among the nodes in a cluster, rather than requiring nodes to connect to a Cluster Shared Volume (CSV). This approach is sometimes referred to as shared-nothing clustering.
Although Storage Spaces Direct and related features are primarily designed for storage replication within a cluster, the Windows Server 2016 Storage Replica feature can be used to facilitate cluster-to-cluster storage replication. In doing so, one cluster is treated as the source cluster, while the other is treated as the destination server. Replication flows in one direction, from the source to the destination. The Storage Replica feature exists only in the Datacenter Edition, and Storage Spaces is one of the prerequisites for it.
Cluster-to-cluster replication requirements
Cluster-to-cluster replication in Windows Server 2016 Storage Replica does not require shared-nothing clustering. It is possible to replicate a traditional cluster with a CSV. To get cluster-to-cluster replication to work properly, however, there are a few requirements that must be met.
- The source and destination clusters must both use a CSV, and the sector size must be identical for logical disks across both clusters.
- Each cluster's physical storage must be able to accommodate two virtual disks: one for the actual data and one to store the log files used by the replication process. Ideally, the virtual disks for log storage should exist on solid-state drive storage. In any case, the log virtual disks must be the same size for each cluster. The virtual disks for data must also be identical in size. Both the log and data disks must be created as GUID Partition Table disks. Master Boot Record disks are not supported.
- Because data is replicated across clusters, the network that exists between the two clusters must be secure and have enough bandwidth available to accommodate the replication process. At a minimum, the sustained data change rate must not exceed the network's bandwidth.
The actual requirements for cluster-to-cluster connectivity vary significantly depending on how replication is performed. If data is replicated synchronously, there are latency requirements, but latency is not a factor in asynchronous replication.
Configuring the cluster-to-cluster replication process
Microsoft does not provide a GUI tool for enabling cluster-to-cluster replication. The configuration process must be performed through PowerShell, but this is relatively easy.
- Provide the clusters with access to one another using the Grant-SRAccess cmdlet.
- Create a replication partnership between the two clusters using the New-SRPartnership cmdlet.
Microsoft Windows Server has long supported various methods of achieving cluster-to-cluster replication. But users can now utilize Windows Server 2016 Storage Replica, through the Datacenter Edition, to support OS-level replication of a CSV between two clusters.
Explore Windows Server 2016 disaster recovery features
Windows Server containers can advance DR
Using Storage Spaces Direct for high availability systems
Learn how containers boost Windows Server 2016