Design an Online Bookstore/E-commerce Catalog

To design this catalog, I first clarify that we need to support millions of books with varying attributes, requiring high availability and fast read speeds. For the data model, I propose a polyglot persistence approach. We will use PostgreSQL for core transactional data like inventory levels and pricing to ensure strong consistency, which is critical for preventing overselling. Simultaneously, we will utilize a document store like DynamoDB to store flexible product metadata. This allows us to easily add new fields for different book formats or editions without altering the rigid schema of the main database. The API will expose a GET /catalog endpoint accepting query parameters for genre, price range, and sort order. To optimize performance, we will implement caching at the edge using CloudFront for popular categories. For the multi-schema scaling aspect, we can shard the product table horizontally based on the first letter of the ISBN or category ID. This distributes the write load while keeping related items together for efficient querying. Finally, we must consider eventual consistency for non-critical metrics like 'total views' to avoid locking issues during peak traffic, aligning with Amazon's customer obsession by prioritizing page load speed over real-time analytics updates.