Website under construction

Recommendations 
The following are highly recommended for Storage Replica testing: 

Network 

Bandwidth: ≥10 Gbps network between servers 

Latency: ≤5 ms round-trip average for synchronous replication 

Storage 

Flash (solid-state drive [SSD]) disks for the log volume(s) with at least 8 GB of free space 
Tip 
You can use the Test-SRTopology Windows PowerShell cmdlet to ensure that you meet the 
requirements and assist with recommended configuration tuning for log files. 
Scenarios 
Storage Replica was designed with two scenarios in mind: 

Stretching of a failover cluster for high availability 

Replication between servers for disaster recovery 
Stretch cluster replication 
A stretch cluster (also referred to as a multisite cluster) uses Storage Replica to connect two sets of 
asymmetric shared storage within a single failover cluster. This storage can be serially attached SCSI 
JBOD (just a bunch of drives), iSCSI target, or SAN. Cluster nodes attach to each of the two sets of 
storage, ostensibly in two physical locations, such as different buildings on the same campus or 
different metropolitan datacenters. The replicated storage can be either Cluster Shared Volumes (CSV) 
or role-assigned Physical Disk Resources (PDR). 
Figure 2-53 presents the typical architecture used for implementing stretch cluster replication using 
Storage Replica. On the left is the Redmond site, where there are two servers (SR-SRV-01 and SR-SRV-
02) and shared storage (SAN, JBOD, or iSCSI). On the right is the Bellevue site, where there are two 
more servers (SR-SRV-03 and SR-SRV-04) and more shared storage. You can use Storage Replica to 

---

50 
CHAPTER 2 | Software-defined datacenter 
combine the servers and shared storage at these two sites into a single stretched cluster by 
asymmetrically replicating storage from one site to the other.