Getting started — slowdep docs

Setup

Four steps to realistic latency

1

Install

Add slowdep to your project. Zero dependencies, ships TypeScript types.

npm install slowdep

2

Import

Import withLatency using ESM or CommonJS — both work out of the box.

import { withLatency } from 'slowdep'

const { withLatency } = require('slowdep')

3

Wrap a function with a preset

Pass your async function and a preset name. The wrapped function has the same signature as the original.

// Your original async function
async function queryUser(id) {
  return db.query('SELECT * FROM users WHERE id = $1', [id])
}

// Wrapped with Postgres-realistic latency
const slowQueryUser = withLatency(queryUser, 'postgres')

4

Use the wrapped function

Call the wrapped function exactly as you called the original. It returns the same result — just with realistic latency added before execution.

// Use exactly like the original
const user = await slowQueryUser(42)
console.log(user.name) // same result, realistic timing

How it works

What just happened?

When you call slowQueryUser(42), slowdep samples a latency value from a lognormal distribution parameterized by the 'postgres' preset (p50: 5ms, p99: 200ms). It waits that many milliseconds, then calls your original queryUser(42) function and returns its result.

The lognormal distribution is key. Unlike a flat setTimeout, it produces realistic variance: most calls are fast (near 5ms), some take longer (20�50ms), and occasionally one takes 100�200ms. That long tail is where timeout misconfiguration and missing retry logic hide.

If the preset's errorRate fires (postgres: 0.1% of calls), the wrapper rejects with a transient error before calling your function — giving your retry and fallback code something real to handle.

Production tip

Only wrap in development and test environments. Use process.env.NODE_ENV !== 'production' ? withLatency(fn, preset) : fn to ensure zero overhead in production. See Production usage for the full pattern.

Next steps

Where to go from here

concepts

Core concepts

Why lognormal latency matters and how slowdep models real-world dependency behavior.

reference

Built-in presets

All 10 presets with their p50/p99 values and the services they model.

API

withLatency

Full API reference for wrapping a single async function.

API

withLatencyAll

Wrap every async method on a client object at once.