---
title: Introduction
description: Control how many tasks run at once and how fast they start.
---

# Introduction

When you call an API that limits how many requests you can make (e.g. 100 per minute), or when you don’t want to flood a service with too many parallel calls, you need something to **queue** and **pace** your work.

**rate-queue** does that: you give it async jobs, and it runs them according to rules you set.

**Production = multiple Node instances.** In production, Node usually runs on more than one process (containers, PM2 cluster, several servers). Each process has its own memory, so a limit that lives only in one process doesn’t apply to the others. For a **single shared limit** across all instances, use the **Redis-backed** limiter (`DistributedRateLimiter`). The in-memory limiter (`RateLimiter`) is for a single process (e.g. local dev or one instance).

- **Concurrency** — e.g. “at most 5 requests at a time” (across all instances with Redis).
- **Rate** — e.g. “start at most one every 100ms” or “max 60 per minute”.
- **Token bucket** — e.g. “10 tokens, refill 10 every minute”.

Next: [Installation](/installation) → [Quick start](/quick-start).