Replication

What it is

Replication keeps copies of data on multiple machines or regions. One node may accept writes and stream changes to followers, or multiple nodes may coordinate writes.

Why it matters

Copies let systems survive machine failure, serve reads closer to users, run backups, and move traffic during incidents.

How it works

Changes are written to a log and applied on replicas. Replication can be synchronous, where writes wait for copies, or asynchronous, where replicas catch up after the primary acknowledges.

When to use it

Use replication for databases, search indexes, object metadata, logs, and any service that needs high availability or read scaling.

Tradeoffs / failure modes

Replica lag causes stale reads. Failover can lose recent asynchronous writes. Multi-region replication adds conflict and latency questions.

Tiny example

Gmail can keep email metadata replicated so reads continue even when one storage node fails.

Related guides and systems

See Caching Strategies, Multi-Tenant Architecture, Audit Logging, Telegram, Google Drive, Gmail, and GitHub.