Thundering Herd Challenge

When cache fails, everything breaks.

This repository serves as a long-term reference for understanding and dealing with one of the most common problems in backend systems:

The Thundering Herd Problem (Cache Stampede)

For participation details, timeline, and reward information, see the official blog post.

---

What is this challenge about?

Imagine you are building an API: GET /api/product/{id}

Your system:

• Reads data from a database (~100ms latency)
• Uses cache (in-memory, Redis, or similar)

---

The Problem

At some point, the cache expires (or misses).

Suddenly:

• Hundreds or thousands of requests hit the same key simultaneously
• All of them go directly to the database
• Database gets overwhelmed
• Latency spikes
• The system may collapse

This is known as:

The Thundering Herd Problem (a.k.a Cache Stampede)

---

Your Goal

This is not just a coding challenge.

The goal is to:

• Understand the problem deeply
• Reproduce it in a test system
• Explain it clearly in your own words
• Propose a solution with reasoning

👉 Your thought process matters more than the code itself.

---

What you need to do

1. Reproduce the problem

Build a system where:

• A cache miss causes multiple concurrent requests
• All requests hit the database for the same key

You can simulate load using tools like:

• JMeter
• k6
• Custom scripts

Show that:

• The same data is fetched multiple times
• The system degrades under load

---

2. Explain the problem

In your README, answer:

• What is the Thundering Herd problem?
• Why does it happen?
• When does it become dangerous?
• What factors make it worse?

👉 Explain in your own words, clearly and concisely.

---

3. Implement ONE solution

Choose at least one approach to mitigate the problem. Examples (optional):

• Mutex / per-key locking
• Request coalescing
• SingleFlight (or equivalent)
• Cache strategies (TTL, soft TTL, etc.)

👉 The solution doesn’t have to be perfect, but it must be well understood.

---

4. Explain your solution

Describe:

• Why your solution works
• What trade-offs it introduces
• When it might fail

---

5. Record a video (required)

Show:

• What you understood
• What you built
• Why you made your decisions

👉 No need for fancy editing. Clarity matters.

---

🔓 Freedom

• Any programming language
• Any tools or libraries
• No “correct” answer

This is an open challenge and an open-source learning resource.

---

Hint

If your solution feels too simple, increase the load or concurrency.

---

Bonus ideas (optional)

• Compare before vs after metrics
• Try multiple solutions
• Simulate multi-instance systems
• Explore distributed locking (e.g., Redis)
• Consider hot keys

---

Final Note

This README is meant as a long-term reference for anyone learning about the Thundering Herd problem.

The challenge is not just about code:

• Show how you think
• How you analyze systems
• How you explain trade-offs

Every clear, insightful explanation has value. Future readers can learn from your approach, understand the reasoning, and see different ways to solve the same problem.

Good luck. 🚀