Figure 4.11 Storage Replica synchronous replication workings
The process is basically the same for asynchronous, except that the write is
acknowledged to the application as soon as it is persisted to the local source log file,
and then as soon as possible the write is replicated to the destination server, where it
is written to the replica log file. Then the log file data is persisted to the data disk at
some point in the future. Note that even though this is asynchronous, Storage Replica
does not batch up the I/Os and sends them as fast as it can, typically sub-second. Even
when running in asynchronous mode, data loss would be minimal, if any, during an
unplanned disaster.
Like Storage Spaces Direct, Storage Replica is easy to get up and running, whatever the
method you use for management, and like Storage Spaces Direct, this feature is
available only in the Datacenter SKU. To check whether a configuration is suitable for
Storage Replica, run the Test-SRTopology cmdlet with the source and destination
server along with the disks that will be used for both data and logs. This will check
every aspect of the environment, including ports, bandwidth, and IOPS. It outputs an
easy-to-read HTML report that includes how long initial synchronization will take.
To set up Storage Replica using PowerShell, use the New-SRPartnership cmdlet with
the same parameters as Test-SRTopology. For example:
New-SRPartnership -SourceComputerName -SourceVolumeName <drive>: ‐SourceLogVolumeName <drive>:
-DestinationComputerName -DestinationVolumeName <drive>: ‐DestinationLogVolumeName <drive>:
-LogSizeInBytes <size, 8GB by default> `
[-ReplicationMode <Asynchronous, Synchronous is the default>]
Storage Spaces Direct and Storage Replica Together
Storage Spaces Direct requires low latency, and it is based around nodes being in the