TechPrufer
Get started
Temporal
Temporal logoTemporal · Feature trail

Task Queue Priority and Fairness

6 posts · May 2026 – Aug 2026

What this trail covers

Task Queue Priority and Fairness provide granular control over task execution order and resource distribution within Temporal. Priority allows assigning integer levels (1-5) to Workflows, Activities, and Child Workflows to ensure critical tasks are processed before lower-priority ones. Fairness uses 'Fairness Keys' and weighted round-robin to ensure equitable resource distribution among tenants or logical groups, preventing monopolization of workers. These features can be combined to support differentiated service tiers and guaranteed performance floors, and are available in Temporal Cloud and self-hosted deployments with specific configuration. Pricing for Fairness in Temporal Cloud is based on additional actions charged per hour a Namespace has the feature enabled.

All Temporal trails
  1. How LlamaIndex uses Temporal to scale reliable document orchestrationLlamaIndex implemented a custom semaphore pattern using Temporal Workflows to manage concurrency limits for their document processing pipelines. This involved creating long-lived coordinator Workflows keyed by resource names to act as semaphores. Callers signal-with-start these semaphore Workflows, which atomically acquire permits and signal back to the caller. This pattern replaced complex legacy systems built on RabbitMQ and databases, consolidating caching layers, microservices, and message metadata. The implementation demonstrated how Temporal's durable execution and signal/wait primitives can elegantly solve problems related to resource allocation, fairness, and efficient resource usage in distributed systems.
  2. Rate-Limit Downstream APIs with Task QueuesThis post details how to use separate Task Queues with rate limiting configuration to protect downstream APIs by configuring `max_task_queue_activities_per_second`, `max_activities_per_second`, and `max_concurrent_activities` to prevent "too many request" errors, failed workflows, wasted execution, cascading failures, and account suspension.
  3. Temporary Rate Limit Increases on Temporal CloudThis post details a Temporal Validated Pattern for orchestrating temporary rate limit increases on Temporal Cloud Namespaces. It describes a parent Workflow that executes an Activity to raise capacity limits, followed by an asynchronous Child Workflow with an abandon policy. The Child Workflow waits for a specified duration before reverting the capacity to its original limit. This pattern aims to automate the provisioning process, preventing workflow throttling and runaway costs associated with manual capacity adjustments.
  4. Task Queue Priority and Fairness: Your Task Queue, your wayThis post introduces and details the Generally Available features of Task Queue Priority and Fairness. It explains how Priority levels (1-5) can be assigned to Workflows, Activities, and Child Workflows to control execution order, with use cases including time-sensitive operations, SLA-based processing, emergency overrides, and AI agent workloads. It also describes Fairness, which uses 'Fairness Keys' and weighted round-robin to ensure equitable resource distribution among logical groups (e.g., tenants), preventing monopolization. The post provides code examples for assigning priority and fairness keys, explains how to enable these features in Temporal Cloud and self-hosted servers, and offers best practices for implementation and monitoring. It also notes that Fairness in Temporal Cloud incurs additional charges.
  5. Route Specialized Workloads with Task QueuesThis post introduces a Temporal Validated Pattern for routing specialized workloads using separate Task Queues. It explains how to create dedicated Worker pools for different resource requirements (e.g., GPU-intensive ML, high-memory analytics) to avoid wasted resources, inefficient usage, and dependency conflicts that arise when all Activities run on the same Worker type.