About this interactive
Replication is redundancy for data. It copies the data and then keeps the copy up to date by sending it every change. A backup is a copy taken at one point in time, and it does not change afterwards. So a replica follows every change, good or bad, while a backup still holds the data as it was. That is why a hot site keeps its databases replicated and ready, and why you still need backups: a deletion, corruption or ransomware is replicated too.
Almost anything can be replicated: databases, files and folders, whole servers, configurations and directory information, at the machine or disk level (VMware, for example), the folder or file level (Dropbox), or the database level (SQL Server availability groups).
Synchronous replication commits a change on both copies together, so they never differ, but every write waits for the second copy, which hurts over a slow link. Asynchronous replication commits on the primary first and sends the change on afterwards, so writes are fast but the replica can be a little behind, and a change not yet sent is lost if the primary fails.
Bidirectional replication lets changes be made on either copy and replicates them both ways, as Dropbox does. Two people changing the same file in two places at once creates a conflict. File locking prevents it: while a file is open in one place it is locked in the other, and the lock is released when the file is closed.
About TechKnowSurge
TechKnowSurge builds IT and cybersecurity professionals through hands-on, concept-first training built around real understanding — not memorization. Free interactive tools, structured programs, and 25+ years of real-world experience, all in one place.
Explore free tools and programs →