Glossary
Data replication
Data replication is the continuous copying of data from one storage system to another, so the copy changes whenever the original does. It keeps data available when a system or site fails. Because it copies deletions, corruption and encrypted files as faithfully as good data, it does not replace backup.
Synchronous and asynchronous replication
The two modes differ in one moment: when a write is confirmed to the application. That moment decides how much data is at risk and how far apart the two systems can be.
| Mode | When a write is confirmed | Data at risk if the main site fails | Practical distance |
|---|---|---|---|
| Synchronous | After both systems have stored it | None that was confirmed | Usually within one metro area, because every write waits for the second site |
| Asynchronous | After the local system has stored it | The last seconds or minutes not yet sent | Any distance, including another region or cloud |
In asynchronous replication, the lag between the two copies becomes the recovery point objective in practice. It grows during heavy write periods, which are often the moments that matter most. Copying between regions or providers is covered under cloud storage replication.
Where replication fits beside backup
A replica holds data as it is now, in a second place. A backup holds data as it was at earlier points in time. The first protects against losing a system or a site. The second protects against losing the data itself, through deletion, corruption or ransomware. Most organizations run both, and an immutable backup adds a copy that neither an attacker nor a mistaken command can change before its retention date.
Why two copies can fail together
Two copies in the same building, on the same power feed or under the same administrator login share their failures, so a second copy only helps when its failure domain is separate. Lag that nobody monitors becomes data loss nobody measured. And a switch to the replica that has never been run is an assumption: the usual causes of a missed failover are an undersized link, older data that was never copied and applications pointed at a fixed address. Designs where both sites serve users at once are described under active-active disaster recovery.
Frequently asked questions
Is data replication a backup?
No. A replica copies deletions, corruption and encrypted files as faithfully as good data, so it does not keep the earlier versions a recovery needs.
What is the difference between replication and mirroring?
Mirroring usually means synchronous replication between two disks, arrays or volumes, with both copies kept identical at every write. Replication is the broader term and includes asynchronous copying over long distances.