AWS has introduced AWS Durable Functions, a new feature that brings stateful, multi-step execution directly into Lambda code. The capability allows developers to run workflows with retries, checkpoints and extended suspension periods. It also removes the need for external orchestration services for many use cases.
Durable Functions can pause execution for up to one year without compute charges. This gives teams a cost-efficient option for long-running tasks. With this update, AWS aims to simplify serverless development while keeping Lambda’s event-driven model intact.
How AWS Durable Functions change Lambda development
To use AWS Durable Functions, developers add an open-source SDK to their Lambda function. After enabling durable execution, they can begin writing workflows directly in code. The SDK introduces simple constructs such as steps and waits.
A step records progress and retries failures automatically. When the function restarts, completed steps do not run again. A wait suspends execution for a set duration and resumes it later. Because Lambda ends its run during the wait, no compute charges apply.
This model lets teams build workflows inside the function body, using familiar programming tools instead of external state machines. It also reduces operational overhead, especially for applications that require predictable sequencing.
AWS Durable Functions vs. Step Functions
The launch of AWS Durable Functions immediately drew comparisons to AWS Step Functions. Step Functions remain a mature orchestration tool, offering visual diagrams, hundreds of integrations and broad operational controls. However, they require state machines defined in JSON or YAML.
Durable Functions take a different approach. All logic stays in code. Developers can manage retries, branching and long waits without leaving the Lambda environment. This lowers the barrier for teams that prefer code-based orchestration.
Cost also plays a role. Step Functions charge for every state transition. Durable Functions do not charge during suspended execution, which can significantly reduce spend for workflows with long delays.
Advanced features in AWS Durable Functions
AWS added several primitives beyond the core step and wait methods. These features expand the range of possible workflows:
- create_callback() lets functions pause until external events arrive.
- wait_for_condition() supports polling for completion of external processes.
- parallel() and map() enable concurrent execution for workloads with many independent tasks.
These operations give developers tools for building complex pipelines without adopting a separate orchestration service.
New serverless use cases enabled by AWS Durable Functions
Many early discussions highlight how AWS Durable Functions may affect AI, ML and LLM workloads. These workflows often require chained operations, long-running tasks and periods of idle waiting. Durable execution supports these patterns without container management or manual orchestration.
E-commerce operations can also benefit. Inventory checks, payment processing and shipment creation often require multiple steps that depend on each other. Durable Functions allow these steps to run sequentially with built-in state tracking. This improves reliability and reduces the need for custom workflow engines.
Addressing long-standing serverless limitations
Lambda’s 15-minute limit has complicated long workflows for years. Before today, developers needed Step Functions or custom architecture to handle extended tasks. AWS Durable Functions change this.
A single request may now span months, as long as each individual Lambda invocation stays within the runtime limit. The service achieves this through re-entry. Lambda restarts the function with its previous state restored, letting execution continue where it left off.
This design resolves a long-standing challenge in serverless development: orchestrating multi-step logic without dedicated servers or complex orchestration layers.
Competitive landscape and industry context
Microsoft Azure has offered Azure Durable Functions for several years. Because of this, comparisons across cloud providers appeared quickly. Engineers note that AWS is now providing a similar model but with Lambda’s execution environment and event-driven behavior.
Experts also suggest evaluating whether durable execution is suitable for every workflow. Tasks that require continuous compute may need a different architecture. However, workflows with clear steps, long waits or external triggers fit well within the new feature set.
Regional availability and runtime support
AWS has released AWS Durable Functions in the US East (Ohio) Region. Supported runtimes currently include JavaScript and TypeScript on Node.js 22 and 24, along with Python 3.13 and 3.14.
AWS has published pricing details and documentation on its website, along with sample projects for teams preparing to test the feature.
Conclusion: AWS Durable Functions expand Lambda’s role
The arrival of AWS Durable Functions significantly extends what developers can build with Lambda. The new feature supports reliable, stateful, multi-step workflows without external orchestration. It also introduces long suspensions, retries and checkpointing while maintaining cost efficiency. As adoption grows, Durable Functions may shift how organizations design serverless applications and long-running processes.
Read also
Join the discussion in our Facebook community.