Website under construction

Figure 2-53:
Typical architecture used for implementing stretch cluster replication 
This configuration makes a failover cluster tolerant not just of node failures, but entire site failures. 
When a single node in a site fails, another node in that site becomes the new source of replication. 
When all nodes in a site fail, a node in the other site becomes the source of replication. All of this 
occurs automatically, just like a normal nonstretched cluster. Stretch clustering requires a minimum of 
two nodes, and the cluster can contain up to 64 nodes. 
In Windows Server 2016, the two cluster roles recommended for replication are Hyper-V and General 
Use File Server. You should avoid configuring SOFS as a stretch cluster because Windows Server 
failover clusters are not inherently site aware, and applications will end up connecting to nodes in 
both sites and then redirecting back to the owning node where I/O writes occur. This potentially can 
lead to poor application performance. Microsoft supports the use of VM guest clusters in the for 
evaluation purposes only. 
You can manage this cluster with the Failover Cluster Manager (cluadmin.msc) through a simple 
wizard-driven interface. To create a stretched cluster, simply create a CSV and configure the General 
Use File Server role or a Hyper-V VM role. Right-click the source storage (shown in Figure 2-54 as 
Cluster Disk 3 in the cluster named np-sr-cluster.com), click Replication, and then, on the shortcut 
menu that appears, click Enable. 
Figure 2-54:
Turning on disk replication in Failover Cluster Manager 

---

51 
CHAPTER 2 | Software-defined datacenter 
In the Configure Storage Replica Wizard that opens, on the Select Destination Data Disk page, select a 
destination drive from the available storage displayed for the source drive that you want to replicate. 
In Figure 2-55, Cluster Disk 1 is selected as the destination data drive. 
Figure 2-55:
Selecting a destination disk for the Replica 
Follow the remaining wizard prompts to finish configuring your stretched cluster. When configured, 
the storage synchronously replicates between the source and destination storage on the cluster. 
When completed, replication forms the stretch cluster and Storage Replica protects the data on the 
source and destination drives. For example, Figure 2-56 shows the direction of replication after a 
subsequent failover. In this case, Cluster Disk 1 is now the source of replication and Cluster Disk 3 is 
the destination. 
Figure 2-56:
Disk information after a replica has been created 

---

52 
CHAPTER 2 | Software-defined datacenter 
Note You also can also use the Windows PowerShell Failover-Clustering and StorageReplica 
modules to create a stretch cluster.