Leader-Follower Replication¶

Designating a single node to handle writes while others replicate and serve reads simplifies consistency (single writer avoids conflicts) while enabling read scalability, at the cost of write throughput limits and failover complexity.

Tension Addressed¶

Consistency vs. Availability (single writer ensures consistency; leader failure requires election)

Recognizable As¶

Primary/replica terminology; read replicas; leader election; replication lag; failover procedures; split-brain prevention; quorum writes

Canonical Sources¶

Lamport, "Paxos Made Simple" (2001); Ongaro & Ousterhout, "In Search of an Understandable Consensus Algorithm (Raft)" (2014); Kleppmann, "Designing Data-Intensive Applications" (2017)

Transfers To¶

Database Replication (primary/replica), Distributed Consensus (Paxos, Raft), Message Queues (partition leaders), Distributed Locks (lock holders)

Single-leader replication sidesteps the hardest distributed systems problem (concurrent writes to same data) by preventing it structurally. The trade-off is that the leader is a bottleneck for writes and a single point of failure requiring election protocols.

Sub-concepts: - Synchronous vs. Asynchronous Replication: Durability vs. latency trade-off - Leader Election: Consensus protocols for choosing new leader after failure