Temporal

Temporal is an open source durable execution platform that enables developers to build reliable, fault-tolerant workflows and distributed applications without manually handling retries, timeouts, and state persistence. Created by the founders of Cadence (a similar system originally built at Uber), Temporal addresses one of the most challenging problems in distributed systems: ensuring that complex, multi-step processes complete successfully even when individual components fail.

The core concept in Temporal is the Workflow. A Temporal Workflow is a code function, written in a supported programming language, that defines the sequence of steps to accomplish a task. Unlike regular function calls that fail if the process crashes, a Temporal Workflow is durable: its execution state is automatically persisted after every step, so it can be resumed from exactly where it left off after any kind of failure including process crashes, network outages, server reboots, and cloud provider incidents. From the developer's perspective, the code runs to completion as if failures never happen.

Activities are the building blocks of Temporal Workflows. Each Activity represents a single, potentially fallible operation such as calling an external API, writing to a database, sending an email, or processing a file. Temporal automatically retries failed activities based on a configurable retry policy, with exponential backoff and maximum attempt limits. Activities are executed by Worker processes that pull activity tasks from the Temporal server's task queues, enabling horizontal scaling of execution capacity.

The programming model in Temporal is natural and language-native. Developers write Workflows and Activities as regular code in TypeScript/JavaScript, Python, Go, Java, .NET, PHP, or Ruby. There are no special YAML definitions or external workflow DSLs to learn. The SDK handles all the complexity of durability, retries, and state serialization transparently, allowing developers to express complex orchestration logic using standard programming constructs like loops, conditionals, recursion, and async/await.

Signals and Queries provide mechanisms for external interaction with running workflows. Signals are asynchronous messages that can change the state of a running workflow, useful for human approvals, external triggers, and pause/resume scenarios. Queries allow external code to read the current state of a running workflow synchronously, enabling status dashboards, polling systems, and debugging.

Child Workflows enable decomposing complex workflows into smaller, independently executed sub-workflows. Timers in Temporal can sleep for arbitrary durations, from milliseconds to years, without consuming resources or requiring external scheduler infrastructure. These timers are durable and survive process restarts, making long-running business processes like subscription renewals, trial period tracking, and regulatory holds trivially easy to implement.

Temporal Cloud is the managed cloud service that provides a production-ready Temporal deployment with horizontal scaling, multi-region namespaces, security, and SLAs. Temporal Open Source can be self-hosted on any infrastructure using Kubernetes or Docker. The Temporal Web UI and CLI provide visibility into running and completed workflow executions for debugging and monitoring.