primaryreplica

A primary replica is the writable node in a replicated data system that holds the authoritative copy of the data and processes all write operations. In a typical configuration, one primary replica receives updates from clients and propagates those changes to one or more secondary replicas, which hold copies of the same data.

Operation and replication methods vary by system. In synchronous replication, commits on the primary are coordinated with one or more secondaries, ensuring that changes are durably applied before a transaction is considered committed. In asynchronous replication, the primary may acknowledge a commit before the corresponding updates are applied on secondaries, trading some risk of data loss for lower write latency. Replication can occur through log shipping, streaming replication, or other mechanisms, and secondary replicas may be configured as read-only to offload query workloads.

High availability and disaster recovery are common motivations for using a primary-secondary setup. If the primary replica fails, an automated or manual failover to a secondary replica can restore service. Read scalability is another benefit, as secondaries can serve read-heavy workloads or reporting queries without impacting the primary’s write performance. Backups can sometimes be performed on secondary replicas, depending on the system’s capabilities and policy.

Caveats include replication lag, especially in asynchronous configurations, potential data divergence during network failures, and the complexity of maintaining consistency across replicas. Applications may need to handle failover events, connection routing, and read/write split logic. Terminology and exact behavior differ across systems, but the core idea remains: a primary replica handles writes, while secondary replicas maintain up-to-date copies for availability, recovery, and read scaling.