System Design - Book Price Aggregator and Purchase Platform

In Databricks system design interviews, the Book Seller Platform (or Book Price Aggregator) is a frequent scenario-based question. It tests your ability to design a resilient, high-scale system that orchestrates multiple external calls and handles complex transaction workflows. Hello Interview +3

Problem Statement Overview

You are tasked with designing a web service that enables customers to automatically purchase the cheapest available copy of a specific book from multiple third-party online sellers. YouTube +1 1

Core Functional Requirements

• Search & Aggregate: When a customer requests a book (providing title/ISBN), the system must contact multiple book sellers simultaneously to fetch current prices.
• Automatic Purchase: If a seller offers the book at or below a customer's specified maximum price, the system must automatically execute the purchase from the cheapest seller.
• Feedback Loop: If a purchase is successful, inform the customer of the seller and the final price. If no seller meets the price limit, report the lowest price found so the customer can decide whether to try again later.
• Seller Integration: The system must handle heterogeneous protocols and diverse API formats for various registered sellers. Hello Interview +5

Technical Constraints & Key Challenges

• Scalability: The system should scale to handle high traffic, often cited as 10,000 queries per second (QPS).
• Latency Management: You must manage "tail latency" (slow responses from certain sellers) to ensure the overall user experience remains fast.
• Reliability & Consistency: Idempotency: Ensure that a customer is not charged twice for the same request. Distributed Transactions: Manage the multi-step workflow of "Search -> Compare -> Pay -> Confirm" across external systems.
• Inventory Volatility: Address the challenge of rapidly changing off-platform inventory where a book might be sold out by the time you attempt the purchase. Hello Interview +4

Common Interview Focus Areas

Interviewers at Databricks often use this problem to dive deep into:

• Fan-out Pattern: How you handle calling many external APIs in parallel.
• Asynchronous Processing: Using message queues (like Kafka) to handle the purchase workflow without blocking the user.
• Payment Safety: Securely handling shipping and payment details through third-party integrations. Hello Interview +5

Suggested Next Step

Would you like to explore a high-level architecture for this system (e.g., using microservices or serverless functions), or should we focus on a specific deep dive like handling tail latency or payment idempotency?

[0] - Design a Book Seller Platform | Hello Interview [1] - Design an Online Bookstore Service - Most Frequently Asked ...