---
title: 'How Ticket Booking Systems Handle Millions of Users | System Design Explained | Simplilearn'
source: 'https://youtube.com/watch?v=g6jQxzWTkkc'
video_id: 'g6jQxzWTkkc'
date: 2026-11-17
duration_sec: 162
channel: 'Simplilearn'
---

# How Ticket Booking Systems Handle Millions of Users | System Design Explained | Simplilearn

> Source: [How Ticket Booking Systems Handle Millions of Users | System Design Explained | Simplilearn](https://youtube.com/watch?v=g6jQxzWTkkc)

## Summary

This video explains how ticket booking systems handle millions of concurrent users without selling a single seat twice. It walks through a classic system design interview question, using the Coldplay ticket rush in India as a real-world example. The solution involves a virtual waiting room, atomic seat holds, timers, idempotent payments, and bot protection.

### Key Points

- **The Double Booking Problem** [00:17] — The naive approach of checking if a seat is free and then booking it fails under concurrency, leading to double bookings.
- **Virtual Waiting Room** [00:47] — A virtual waiting room assigns numbers and admits users in small batches to prevent server crashes.
- **Fast Seat Map Serving** [01:01] — Seat maps are served from fast in-memory caches like Redis to avoid hitting the main database for read-only operations.
- **Atomic Seat Hold** [01:15] — Checking and holding a seat must be a single atomic step to ensure only one user can secure it.
- **Hold Timer** [01:29] — Every hold has a timer (e.g., 10 minutes) to release seats if payment isn't completed.
- **Idempotent Payments** [01:43] — A unique booking ID ensures duplicate payment requests are ignored, preventing double charges.
- **Bot Protection** [02:11] — Bots are blocked by capping tickets per user and rate-limiting requests.
- **Data Sharding by Show** [02:25] — Data is sharded by show, so a rush for one event doesn't affect another.

## Transcript

tickets. That's what happened when Coldplay tickets went live in India. Now picture this. Two people tap the same seat at the exact [music] same second. Who gets it? This is a classic system design interview question. How would you
build a ticket booking system where one seat is never sold twice? The first thought is simple. Check the database. Is seat A12 free? Yes, book it. But here's where it breaks. Person A checks and it's free. Person B checks at the
same moment and it's free for them too. Both book it. Two tickets, one seat, and [music] a big fight at the stadium gate. So, here's how you'd really build it. Step one, don't let everyone in at once. [music] 1.3 crore people opening the
booking page together will crash anything. So, send them to a virtual waiting room. Everyone gets a number and the system lets people in small batches. [music] That's the six lakh people ahead of you screen. Step two, show the seat
map from fast memory like Reddus. Millions of people just looking at seats should never [music] touch the main database. Step three, the moment you tap a seat, the system holds it for you. And here's the key. Checking and holding
happen in one [music] single step. Give me the seat, but only if nobody has it. If two people try in the same millisecond, [music] only one wins. the other sees seat not available right away. [music] But here's the trap. You
held the seat then went looking for your card or your payment failed or you just closed the app. Now that seat [music] is stuck with you forever. So every hold stuck with you forever. So every hold gets a timer say [music] 10 minutes. No
payment by then the seat goes back to everyone automatically. Second trap. Your payment goes through but your network drops. You panic and tap pay again. Do you get charged twice? No, every booking carries a unique ID. If
the same ID comes in twice, the system knows it's a repeat and ignores it. Step four, only after the payment succeeds is the seat marked sold in the main database permanently. Step five, bots. One script can grab hundreds of seats in
seconds. So, you cap tickets per person. Coldplay tickets were capped at four per user, and you block [music] anyone sending too many requests. One last thing, every show is its own little world. So split the data across servers
by show. A Coldplay rush in Mumbai won't slow down someone booking a movie in Chennai. And that's how one system handles crows of fans without selling a handles crows of fans without selling a single seat twice.
