Website under construction

Storage Replica in Windows Server 2016 
The following are some of the key things to know concerning Storage Replica as of the Windows 
Server 2016 release: 

Network bandwidth and latency with fastest storage There are physical limitations to 
synchronous replication. Because Storage Replica implements an I/O filtering mechanism using 
logs and requiring network roundtrips, synchronous replication is likely to make application writes 
slower. By using low-latency, high-bandwidth networks as well as high-throughput drive 
subsystems for the logs, you can minimize performance overhead. 

The destination volume is not accessible while replicating When you configure replication, 
the destination volume will dismount and no longer be visible in any normal GUI tools or 
accessible to any writes by users until you remove replication, or the volume becomes the source 
due to failover. 
Block-level replication technologies are incompatible with allowing access to the destination’s 
mounted file system in a volume; NTFS and ReFS do not support users writing data to the volume 
while blocks change underneath them. 

Different implementation of asynchronous replication The Microsoft implementation of 
asynchronous replication is different than most industry implementations of asynchronous 
replication that rely on snapshot-based replication, whereby periodic differential transfers move 
to the other node and merge. In contrast, Storage Replica asynchronous replication operates just 
like synchronous replication, except that it removes the requirement for a serialized synchronous 
acknowledgment from the destination. This means that Storage Replica theoretically has a lower 
RPO as it continuously replicates. However, this also means it relies on internal application 
consistency guarantees rather than using snapshots to force consistency in application files. 
Storage Replica guarantees crash consistency in all replication modes. 
