Is your backend a hodgepodge of queues, event stores and cron jobs? Durable Execution to the Rescue.

Your code fails, a server reboots, and your process continues exactly where it left off. This is the promise of durable execution.

#1about 4 minutes

The hidden complexity of event-driven architectures

Event-driven systems using queues create a complex web of dependencies that makes state management, error handling, and long-running operations difficult.

#2about 2 minutes

Why traditional orchestration engines often fail developers

While orchestration centralizes logic, traditional workflow engines suffer from poor programming models that rely on visual designers or restrictive DSLs instead of real code.

#3about 2 minutes

Introducing the durable execution programming paradigm

Durable execution is a concept where application code and its state are automatically preserved across process crashes, machine failures, and deployments.

#4about 3 minutes

How state is recovered using event logging and replay

The durable execution model is implemented by logging the results of external operations and replaying the code from the beginning, feeding it the logged results to restore state.

#5about 5 minutes

The challenge of writing deterministic code for replay

To ensure reliable recovery via replay, the application code must be deterministic, requiring special handling for non-deterministic operations like random numbers, system time, and concurrency.

#6about 3 minutes

Temporal as an open source durable execution system

Temporal is an open-source platform that implements the durable execution paradigm, providing a backend service and SDKs to build resilient applications.

#7about 3 minutes

Writing a durable workflow with the Python SDK

A practical example shows how to define workflows and activities in Python using the Temporal SDK, which handles state persistence and retries automatically.

#8about 3 minutes

Common use cases for durable execution systems

Durable execution is a general-purpose abstraction applicable to a wide range of problems, including infrastructure provisioning, CI/CD pipelines, data processing, and complex business transactions.

#9about 3 minutes

Q&A on idempotency and business logic boundaries

The platform handles infrastructure failures, but developers are still responsible for business logic concerns like activity idempotency and handling data changes over time.