Queue Concurrency in Sidequest

This document provides a detailed explanation of queue concurrency in Sidequest, its purpose, how it works, and best practices for configuring concurrency effectively.

What is queue concurrency?

Queue concurrency controls the maximum number of jobs from a single queue that can run in parallel. This allows you to prevent a queue from consuming all available worker capacity, ensuring that other queues also get a share of resources.

Concurrency is configured per queue as a positive integer and applies only to that specific queue.

How it works

Each queue in Sidequest has a dedicated pool of workers limited by its concurrency setting. When jobs are ready to run, Sidequest will not start more than concurrency jobs from a queue at the same time, even if more jobs are waiting.

The total number of jobs running across all queues is also limited by the global maxConcurrentJobs parameter.

Default behavior

If no concurrency is specified in a queue’s configuration, it defaults to 10. The minimum concurrency is 1.

Example:

queues: [
  { name: "default", concurrency: 2 },
  { name: "critical", concurrency: 5 },
  { name: "low", concurrency: 1 },
];

In this example, at most 2 jobs from the default queue, 5 jobs from critical, and 1 job from low can run concurrently, provided the maxConcurrentJobs limit (a global Sidequest Engine configuration) limit is not exceeded.

Best practices for concurrency

• 📊 Match workload characteristics: Higher concurrency for lightweight, fast jobs; lower concurrency for heavy, resource-intensive jobs.
• 🧩 Consider downstream dependencies: Avoid overloading external systems by limiting concurrency appropriately.
• 🔄 Balance with priority: Concurrency limits how many jobs from a queue can run at once, regardless of priority.
• 🧪 Monitor performance: Observe resource usage and adjust concurrency settings based on actual system behavior.