Leaderless replication allows any replica to handle any type of request to achieve read scalability and high availability for dis- tributed data stores. However, this entails burdensome coordina- tion overhead of replication protocols, degrading write through- put.
Leader-centric replication is not fault-tolerant by design because we lose the write operation when the Master node is down. Leaderless replication addresses this concern and ensures our system can handle Write operations even when a subset of nodes are having an outage.