What is synchronous replication?

In synchronous replication, all committed data at the primary location also exists simultaneously at the remote location. When a piece of data is being written to a disk at the primary location, it must also be written (replicated) at the secondary (remote) location, and the secondary location must acknowledge that it has received and written that piece of data. Only then, the next piece of data can be written to the primary location, replicated to the secondary location, and so on.

The result is two identical copies of data, in real time, which is ideal for rapid disaster recovery and the highest application availability. However, synchronous replication has two important limitations: cost and distance. To sustain synchronous replication, the organization needs to invest in, maintain, and staff two data centers (the primary, production site and the secondary, disaster recovery site). The organization must also maintain extremely fast and costly communication lines between the two sites. Because synchronous replication is costly, this topology is unaffordable for many organizations.

The second limitation of synchronous replication is that the primary and secondary locations must be close to each other in order to minimize latency when moving the data between the primary and DR sites. The maximum distance between the two sites is usually around 45 miles. This means that in case of a large, regional disaster, it is possible that both data centers will be hit; in this case, data will be lost, and application downtime is assured.