Inbox Pattern

Noun · Development

Definitions

  1. The consumer-side counterpart of the outbox pattern: incoming messages are first stored in an inbox table (with deduplication by message ID), then processed. If the same message arrives twice (at-least-once delivery), the inbox detects the duplicate and skips processing. Ensures exactly-once processing semantics on top of at-least-once delivery.

    In plain English: Storing incoming messages in a deduplication table before processing, ensuring each message is handled exactly once.

    Example: "The inbox table deduplicates by event_id -- when Kafka delivers the same OrderCreated event twice, the second insert is a no-op."

Related Terms