What is jitter and why does it matter?

Jitter adds randomness to backoff delays. Without it, thousands of clients that failed at the same moment retry at the same moment again (a thundering herd). Full jitter sleeps a random time between 0 and the slot, equal jitter between half and the full slot, and decorrelated jitter bases each delay on the previous one, per the AWS architecture blog.

Which jitter mode should I choose?

For most distributed clients calling a shared API, full jitter is the recommended default - AWS, Google, and Stripe all recommend it. Equal jitter keeps delays closer to the computed slot, and decorrelated jitter adapts to recent history. Only use no jitter for a single client or a strict sequence like IoT device provisioning.

What base, multiplier, and cap should I use?

A common production recipe is base 0.5-1s, multiplier 2, cap 30-60s, full jitter, with 5-8 retries. Start from your API rate limits and expected outage length: a shorter base recovers faster after a blip, while a higher cap protects you from long outages without hammering the server.

What is Exponential Backoff Calculator?

Exponential backoff is the standard retry strategy for flaky networks and overloaded services: after each failed attempt you wait progressively longer before retrying, growing the delay by a multiplier up to an optional cap. This free online exponential backoff calculator computes the exact delay for every retry from your base interval, multiplier, cap, and retry count, adds jitter (full, equal, or AWS-style decorrelated) to prevent synchronized retry storms, and totals the worst-case wait. It also generates ready-to-use Python and JavaScript retry snippets matching your parameters. Everything runs 100% client-side in your browser.

Key Benefits

Prevents thundering-herd retries by visualizing how jitter spreads load
Dial in the fastest safe retry policy before writing a line of code
Know the worst-case total wait for SLAs and user-facing timeouts
Compare all four jitter strategies on identical parameters instantly
Production-ready Python and JavaScript snippets with one click

Common Use Cases

•Tuning API client retry policies for rate-limited or flaky endpoints
•Designing AWS SDK / cloud service error handling with recommended jitter
•Sizing timeouts and retry counts in microservice-to-microservice calls
•Teaching distributed-systems resilience with a concrete retry timeline
•Configuring message-queue consumers and webhook delivery retry intervals

How to Calculate an Exponential Backoff Schedule

Set the base interval and multiplier: The base is the first retry delay; the multiplier (commonly 2) grows each later delay geometrically.

Add a cap and a retry count: The cap limits the longest single delay; the retry count controls how many attempts the schedule covers.

Pick a jitter mode: Choose none, full, equal, or decorrelated jitter to see how retry delays spread across many clients.

Copy the timeline and snippets: Review the retry timeline with cumulative wait, then copy the Python or JavaScript snippet preconfigured with your parameters.

Key Features

Four jitter modes: none, full, equal, and AWS-style decorrelated

Retry-by-retry timeline with raw, capped, and cumulative delays

Optional cap with live uncapped vs capped comparison

Deterministic seed plus a Regenerate button for reproducible jitter

Copy-paste Python and JavaScript snippets generated from your parameters

Exponential Backoff Calculator

Compute exponential backoff retry schedules online. Set base interval, multiplier, cap, and jitter (full, equal, decorrelated) with a retry timeline and copy-paste Python/JS snippets. 100% client-side.

Base Interval (s)

The initial delay before the first retry, in seconds.

Multiplier

Growth factor applied after each retry. 2 doubles the delay every attempt.

Cap (s)

Maximum delay for any single retry. Leave empty for no cap.

Number of Retries

How many retry attempts to schedule (1-20).

Jitter Mode

Pure exponential backoff: every client sleeps exactly the computed delay, which can create synchronized retry thundering herds.

Total Wait Time

1m 3s

Worst-case cumulative waiting across 6 retries before the operation succeeds.

First Retry Delay

1.00 s

Max Slot (capped)

32.00 s

Retries Scheduled

Retry #	Raw (uncapped)	Slot (capped)	Sleep	Cumulative
1	1.00 s	1.00 s	1.00 s	1.00 s
2	2.00 s	2.00 s	2.00 s	3.00 s
3	4.00 s	4.00 s	4.00 s	7.00 s
4	8.00 s	8.00 s	8.00 s	15.00 s
5	16.00 s	16.00 s	16.00 s	31.00 s
6	32.00 s	32.00 s	32.00 s	1.1 min

Copy-Paste Snippets

Python

import random

def backoff_sleep(attempt, base=1.0, multiplier=2.0, cap=60.0, jitter="full", prev=None):
    """Return seconds to wait before retry `attempt` (1-based)."""
    slot = min(cap, base * multiplier ** (attempt - 1))
    if jitter == "full":
        return random.uniform(0, slot)
    if jitter == "equal":
        return slot / 2 + random.uniform(0, slot / 2)
    if jitter == "decorrelated":
        prev = base if prev is None else prev
        return min(cap, base + random.uniform(0, max(base, prev * 3) - base))
    return slot  # no jitter


# Retry loop example
for attempt in range(1, 7):
    delay = backoff_sleep(attempt)
    print(f"attempt {attempt}: sleeping {delay:.2f}s")
    # ... call the failing operation here ...

JavaScript

async function backoffSleep(attempt, {
  base = 1,
  multiplier = 2,
  cap = 60,
  jitter = "full",
  prev = null,
} = {}) {
  const slot = Math.min(cap, base * multiplier ** (attempt - 1));
  if (jitter === "full") return Math.random() * slot;
  if (jitter === "equal") return slot / 2 + Math.random() * (slot / 2);
  if (jitter === "decorrelated") {
    const p = prev === null ? base : prev;
    return Math.min(cap, base + Math.random() * (Math.max(base, p * 3) - base));
  }
  return slot; // no jitter
}

// Retry loop example
let prev = null;
for (let attempt = 1; attempt <= 6; attempt++) {
  const delay = await backoffSleep(attempt, { prev });
  prev = delay;
  console.log(`attempt ${attempt}: sleeping ${delay.toFixed(2)}s`);
  // ... call the failing operation here ...
}

About This Tool

Exponential backoff is the standard retry strategy for unreliable operations: after each failed attempt you wait longer before trying again, growing the delay by a multiplier up to an optional cap. This free online exponential backoff calculator shows the exact delay for every retry, adds optional jitter (full, equal, or AWS-style decorrelated) to prevent synchronized retry storms across clients, and totals the worst-case wait. You also get ready-to-use Python and JavaScript snippets matching your parameters. Everything runs 100% client-side in your browser.

Why Use an Exponential Backoff Calculator?

•Prevent retry storms: see how jitter spreads client retries instead of synchronizing them
•Tune the base, multiplier, and cap to balance fast recovery against API rate limits
•Know the worst-case total wait before you commit to a retry count in production
•Compare no-jitter, full, equal, and decorrelated jitter side by side on the same parameters
•Copy production-ready Python and JavaScript retry snippets - no handwritten math

Key Features

✓Four jitter modes: none, full, equal, and AWS-style decorrelated
✓Retry-by-retry timeline with raw, capped, and cumulative delays
✓Optional cap with live comparison of uncapped vs capped values
✓Deterministic seed with a Regenerate button for reproducible jitter
✓Copy-paste Python and JavaScript snippets generated from your parameters

Loading tool...

What is Exponential Backoff Calculator?

Key Benefits

Prevents thundering-herd retries by visualizing how jitter spreads load
Dial in the fastest safe retry policy before writing a line of code
Know the worst-case total wait for SLAs and user-facing timeouts
Compare all four jitter strategies on identical parameters instantly
Production-ready Python and JavaScript snippets with one click

Common Use Cases

•Tuning API client retry policies for rate-limited or flaky endpoints
•Designing AWS SDK / cloud service error handling with recommended jitter
•Sizing timeouts and retry counts in microservice-to-microservice calls
•Teaching distributed-systems resilience with a concrete retry timeline
•Configuring message-queue consumers and webhook delivery retry intervals

How to Calculate an Exponential Backoff Schedule

Set the base interval and multiplier: The base is the first retry delay; the multiplier (commonly 2) grows each later delay geometrically.
Add a cap and a retry count: The cap limits the longest single delay; the retry count controls how many attempts the schedule covers.
Pick a jitter mode: Choose none, full, equal, or decorrelated jitter to see how retry delays spread across many clients.
Copy the timeline and snippets: Review the retry timeline with cumulative wait, then copy the Python or JavaScript snippet preconfigured with your parameters.

Key Features

Four jitter modes: none, full, equal, and AWS-style decorrelated
Retry-by-retry timeline with raw, capped, and cumulative delays
Optional cap with live uncapped vs capped comparison
Deterministic seed plus a Regenerate button for reproducible jitter
Copy-paste Python and JavaScript snippets generated from your parameters

Frequently Asked Questions

Retry #

Raw (uncapped)

Slot (capped)

Sleep

Cumulative

1.00 s

2.00 s

3.00 s

4.00 s

7.00 s

8.00 s

15.00 s

16.00 s

31.00 s

32.00 s

1.1 min

import random def backoff_sleep(attempt, base=1.0, multiplier=2.0, cap=60.0, jitter="full", prev=None): """Return seconds to wait before retry `attempt` (1-based).""" slot = min(cap, base * multiplier ** (attempt - 1)) if jitter == "full": return random.uniform(0, slot) if jitter == "equal": return slot / 2 + random.uniform(0, slot / 2) if jitter == "decorrelated": prev = base if prev is None else prev return min(cap, base + random.uniform(0, max(base, prev * 3) - base)) return slot # no jitter # Retry loop example for attempt in range(1, 7): delay = backoff_sleep(attempt) print(f"attempt {attempt}: sleeping {delay:.2f}s") # ... call the failing operation here ...

async function backoffSleep(attempt, { base = 1, multiplier = 2, cap = 60, jitter = "full", prev = null, } = {}) { const slot = Math.min(cap, base * multiplier ** (attempt - 1)); if (jitter === "full") return Math.random() * slot; if (jitter === "equal") return slot / 2 + Math.random() * (slot / 2); if (jitter === "decorrelated") { const p = prev === null ? base : prev; return Math.min(cap, base + Math.random() * (Math.max(base, p * 3) - base)); } return slot; // no jitter } // Retry loop example let prev = null; for (let attempt = 1; attempt <= 6; attempt++) { const delay = await backoffSleep(attempt, { prev }); prev = delay; console.log(`attempt ${attempt}: sleeping ${delay.toFixed(2)}s`); // ... call the failing operation here ... }

About This Tool

What is Exponential Backoff Calculator?

Key Benefits

Common Use Cases

How to Calculate an Exponential Backoff Schedule

Key Features

Frequently Asked Questions

What is exponential backoff?

What is jitter and why does it matter?

Which jitter mode should I choose?

What base, multiplier, and cap should I use?

Copy-Paste Snippets

About This Tool

Why Use an Exponential Backoff Calculator?

Key Features

What is Exponential Backoff Calculator?

Key Benefits

Common Use Cases

How to Calculate an Exponential Backoff Schedule

Key Features

Frequently Asked Questions

What is exponential backoff?

What is jitter and why does it matter?

Which jitter mode should I choose?

What base, multiplier, and cap should I use?

Copy-Paste Snippets

About This Tool

Why Use an Exponential Backoff Calculator?

Key Features