# Overview A data replication approach where one node (i.e., replica) is designated as the leader (i.e., master, primary). This is where clients send their writes. The other replicas act as followers. The leader sends the writes to the followers for replications. The specifics on how this process works depends on [[Database Replication#Strategies for Completing a Replication]] and [[Database Replication#Strategies for Communicating Changes to Replicate]]. # Key Considerations If a leader fails, then a new leader must be selected. This can be done manually, through a controller node, or through a [[Consensus]] algorithm. ## Handling Leader Failures #flashcard - Need to be sure that the leader is actually down - is it just a network issue? - Losing writes that are only on the leader - The leader returning and thinking it is still the leader (i.e., split brain) <!--ID: 1751507776617--> # Pros - Improved durability thanks to more copies of the data - Increased read throughput # Cons # Use Cases # Related Topics - [[Active-passive Replication]]