Katkat
Back

Design a booking system

AirbnbSoftware Engineer · Onsite
MediumSoftware EngineerOnsiteSystem Design

Last updated

This question tests the ability to design a distributed reservation system that handles high concurrency, prevents double-booking, and ensures data consistency at scale.

Problem Statement

Design a booking system for a platform like Airbnb that allows guests to search for available listings and reserve them for specific dates. The system must support high-volume read requests (searching for listings) and high-integrity write requests (creating bookings).

Functional Requirements

  1. Users can search for available properties by location and date range.
  2. Users can create a booking for a specific property on available dates.
  3. A property cannot be double-booked for the same dates.

Non-Functional Requirements

  1. High availability for search queries.
  2. Strong consistency for booking transactions.
  3. Low latency for searching and booking.
  4. Scalability to handle millions of listings and thousands of concurrent requests.

Deliverables

  • Define the core API endpoints for search and booking.
  • Propose a data model (SQL vs. NoSQL) that balances search performance and transactional integrity.
  • Explain the concurrency control strategy to prevent race conditions during the booking process.
  • Describe the high-level architecture, including caching strategies and database sharding techniques.

Constraints

  • Assume 1 million properties and 100 million daily searches.
  • Users are distributed globally, requiring a system that is responsive regardless of geography.

Hints

Hint 1

Consider using an inverted index (like Elasticsearch) for the search component to handle complex spatial and date-based filtering efficiently.

Hint 2

For the booking transaction, evaluate whether row-level locking, optimistic concurrency control, or a distributed locking mechanism is most appropriate for your database choice.

Hint 3

Use a Read-Through or Write-Through cache (like Redis) for listing availability to reduce the load on the primary transactional database.

Follow-up questions

  • How would you handle a scenario where a user reserves a property, but the payment service fails to process the transaction?
  • If the database becomes a bottleneck during peak holiday season, what strategies would you use to shard your booking data?
  • How would you implement a 'hold' feature that temporarily reserves a property for 10 minutes while a user completes their checkout?