Eventual Consistency

What it is

Eventual consistency means writes may not be visible everywhere immediately, but replicas should converge to the same value if no new writes occur.

Why it matters

Distributed systems often choose availability and low latency over immediate global agreement. Feeds, counters, search indexes, analytics, and replicated stores commonly accept temporary staleness.

How it works

Writes propagate asynchronously through logs, events, replication streams, or background jobs. Reads from different replicas can return different values until propagation catches up.

When to use it

Use it when the product can tolerate stale reads or delayed side effects, especially across regions or high-volume derived views.

Tradeoffs / failure modes

Users can see missing updates, out-of-order events, or stale permissions. The product needs reconciliation, conflict handling, and clear boundaries for what must be strongly consistent.

Tiny example

A tweet appears on the author’s profile instantly but reaches every follower feed over the next few seconds.

Related guides and systems

See Collaborative Editing, Feed & Timeline Generation, Webhooks, Audit Logging, Twitter, Instagram, Google Drive, and Telegram.