SQLite is all you need for durable workflows
The article, published on obeli.sk, challenges the assumption that durable workflows require heavyweight infrastructure like Kafka or Temporal. It presents a pattern where SQLite serves as both the state store and the coordination mechanism for workflow execution. By leveraging SQLite's WAL mode and row-level locking, the author shows how to implement retries, idempotency, and failure recovery without external dependencies. The key insight is that for many applications—especially those with moderate throughput—SQLite's performance and reliability are sufficient. The post includes code examples in Go, demonstrating a simple workflow engine that uses SQLite as the source of truth for workflow state. It claims this approach simplifies deployment (single binary + SQLite file) and reduces costs. The author acknowledges limitations: not suitable for high-throughput or multi-region scenarios, but argues it covers a wide range of real-world needs.
Simplifies durable workflow implementation, reducing infrastructure complexity and cost for many applications.