Browse
Integration Patterns
The standard ways two systems exchange data — sync API calls, async messaging, batch, and webhooks — and when each fits.
What it is
Integration patterns are the standard architectural shapes for connecting two systems — the choice affects coupling, latency, and failure behavior, and is usually the central technical decision in a customer integration.
Key points
- Synchronous (request/response): simplest to reason about, but couples the caller's availability and latency to the callee's — a slow or down dependency directly degrades the caller.
- Asynchronous messaging: decouples systems in time — the caller doesn't block waiting for the callee to process — at the cost of added complexity (eventual consistency, message ordering, retry/dead-letter handling).
- Batch integration: simplest operationally for high-volume, non-time-sensitive data exchange, but introduces latency equal to the batch interval.
- Webhooks: let the source system push events as they happen — good fit when the consuming system needs near-real-time updates but shouldn't have to poll.