Temporal Patterns

info

This page is part of the Temporal Knowledge Hub.

note

Add and remove Temporal Workflow design patterns relevant for your organization.

Parallel Activity

What it does: Execute multiple Activities concurrently. Why use it: Improve Workflow performance when Activities are independent and don't need sequential execution. Code samples: Python, Java

Custom Search Attributes

What it does: Adds custom key-value metadata to Workflow executions. Why use it: Enables advanced filtering, sorting, and visibility of Workflows in the Web UI and CLI based on business-specific data. Code samples: Python, Java

Child Workflow

What it does: Spawns a new Workflow execution from within a parent Workflow. Why use it: Partition work into smaller chunks, encapsulates Activities into observable components, and model business entities with different lifecycles. Code samples: Python, Java

Continue as new

What it does: Atomically completes the current Workflow execution and starts a new one with the same Workflow ID. Why use it: Prevents "Event History Limit Exceeded" errors and other Workflow Execution limits by clearing the history. Code samples: Python, Java

Exception handling

What it does: Implements logic to catch and respond to Activity or Workflow failures. Why use it: Ensures system resilience by defining fallback logic, compensation transactions, or specific retry policies when errors occur. Code samples: Python, Java

Cancellation

What it does: Sends a request to gracefully terminate a running Workflow or specific scope. Why use it: Stops unnecessary processing and cleans up resources when a result is no longer needed or a user explicitly stops the process. Code samples: Python, Java

Async Activity completion

What it does: Enables the Activity Function to return without the Activity Execution completing. Why use it: Essential for long-running external processes that can heartbeat and inform Temporal of its completion. Code samples: Python, Java

Local Activity

What it does: Executes short-lived Activity logic within the same process as the Workflow Worker. Why use it: Reduces latency and history size for short, high-throughput operations that do not require global durability guarantees. Code samples: Python, Java

Batch Processing (Sliding Window)

What it does: Processes a large stream of items in controlled, concurrent chunks. Why use it: Manages concurrency and throughput limits while efficiently processing high volumes of data without overwhelming downstream services. Code samples: Python, Java

Custom Metrics

What it does: Emits application-specific telemetry (counters, gauges, timers) from Workflows and Activities. Why use it: Provides observability into business-level KPIs and specific Workflow performance characteristics beyond default system metrics. Code samples: Python, Java

Encryption

What it does: Encrypts Workflow and Activity payloads client-side using a custom Data Converter. Why use it: Ensures sensitive data remains secure and opaque to the Temporal Server, satisfying strict compliance and privacy requirements. Code samples: Python, Java

Polling

What it does: Periodically checks the state of an external system from within an Activity. Why use it: Provides reliable integration with external APIs or systems that do not provide webhooks or asynchronous event notifications. Code samples: Python, Java

Worker routing

What it does: Dynamically routes Activities to specific Task Queues monitored by designated Workers. Why use it: Targets tasks to specific hosts or environments; required for file-system affinity, local caching strategies, or hardware-specific (e.g., GPU) operations. Code samples: Python, Java

Saga

What it does: Manages long-running, distributed transactions by executing a sequence of steps. If a step fails, it triggers "compensating actions" (undo operations) in reverse order to revert the changes made by previous steps. Why use it: Ensures data consistency across microservices (e.g., booking a flight, hotel, and car) without locking resources for long periods. It handles partial failures gracefully by rolling back the system to a known consistent state. Code samples: Java

Early Return

What it does: Uses "Update with Start" to begin a Workflow execution and synchronously return a result to the client (e.g., validation success) while continuing to process longer-running tasks (e.g., database updates, external API calls) in the background. Why use it: Drastically reduces end-user latency in interactive applications. Users receive immediate feedback (like an "Order Received" confirmation) without waiting for the entire process to complete. Code samples: Java

Example Temporal Applications

See Temporal Code Exchange for example Temporal applications.