Building fault-tolerant applications with AWS Lambda durable functions

This article introduces AWS Lambda durable functions, a new capability for building fault-tolerant, multi-step applications with automatic checkpointing and failure recovery.

• Durable functions enable checkpoint progress, automatic failure recovery, and suspend execution up to one year
• Built-in durable operations like Steps, Waits, and callbacks eliminate manual state management
• Failed steps retry automatically without re-executing completed checkpoints
• Supports parallel execution, conditional waits, function invocation, and array mapping
• Available for Python and Node.js via Lambda API, CLI, CloudFormation, SAM, CDK, and SDK
• Local testing supported with AWS SAM for callbacks and execution history inspection
• Charges apply to all invocations including replays; wait periods don't incur duration charges
• Use qualified ARNs (versions/aliases) to ensure deterministic replay with consistent code

Lambda durable functions simplify building long-running workflows by automating progress tracking and error handling, allowing developers to focus on business logic.