Lecture Notes: Designing E-commerce Websites like Amazon

Introduction

• Discussion on designing an e-commerce website like Amazon.
• Similar principles apply to other e-commerce platforms like eBay, Walmart, Flipkart.
• Outline of the lecture:
  • Functional and non-functional requirements
  • Capacity estimations
  • Database design
  • APIs
  • System design for scalability
  • Assumptions in design process

Functional Requirements

• Focus on core functionalities due to time constraints (1-hour system design interview).
• Key functionalities include:
  • Product search
  • Product recommendations on homepage
  • Order placement and status viewing
  • Product reviews
• Assumptions: User profile creation, product onboarding, payment gateway available.

Non-Functional Requirements

• Low latency for recommendation and search services.
• High consistency required for order placement, payment, and status viewing.

Capacity Estimations

• User Base: 300 million monthly active users.
• Search Estimations:
  • 10 searches per user per month.
  • Total: 3 billion searches/month.
  • Translates to approximately 1,000 searches per second.
• Storage Estimations:
  • 10 million products needing 10 MB each.
  • Total: 100 terabytes of storage for product information.

Database Design

• User Database: SQL database with structured information (user ID, name, login timestamps, addresses).
• Address Database: Separate SQL database to handle multiple addresses per user.
• Product Database: NoSQL (DynamoDB/MongoDB) due to unstructured data.
• Order Database: NoSQL for flexible storage of varied order items.
• Review Database: NoSQL key-value store with product ID as key.

APIs

• GetRecommendationService: User ID input, returns product recommendations.
• SearchService: Search string and user ID, outputs deliverable products.
• AddToCartService: Inputs include user ID, product ID, quantity; outputs success status.
• PlaceOrderService: Takes user ID, order details, returns success status if inventory available.
• CheckOrderStatusService: Takes order ID, returns order status.

High-Level System Design

• Functional requirements include search service, add to cart, place order, and review services.
• Recommendation Generation: Based on search history, stored in a separate database.
• Order Processing: Sequential flow from cart to order database.
• Review Interaction: Triggered through recommendations and search.

Scaling the System

• Load Balancer: Distributes load; includes standby for redundancy.
• Multiple Instances: For each service to enhance failure resilience.
• Elasticsearch: Enhances search performance for millions of users.
• Master-Slave Database Design: For read-heavy product database.
• Data Redundancy: Multiple copies for reliability in all databases.
• Archiving: Older data like search history and past orders to maintain performance.
• Caching: For immediate access to recommendations and frequent queries.
• Message Queues: Added between services to decouple and reduce dependencies.

Conclusion

• End of lecture on scalable e-commerce system design.
• Encouragement to engage with related content for deeper understanding.