Part 3: Advanced Follow-up Solutions (Google WSE)

#GoogleWSE#Series#Intermediate

Part 3: Advanced Follow-up Solutions (Google WSE) --- Google WSE Interview Series: 1. Part 1: Map & Heap Deep Dive 2. Part 2: The Reality Check Explained 3. Par

Part 3: Advanced Follow-up Solutions (Google WSE) --- Google WSE Interview Series: 1. Part 1: Map & Heap Deep Dive 2. Part 2: The Reality Check Explained 3. Part 3: Advanced Follow-up Solutions 4. Visual Explanation: Process Flow --- This document addresses the advanced follow-up questions for the Product Inventory System. --- Q1: What if multiple orders have the same price? The Solution In our implementation, we store pairs in the Min-Heap: { price, timestamp }. Even if two orders have the same price, they have different timestamps, so they occupy different entries in the heap. Coding Example Why it works: The Map this.orders still contains 2 - 100. When the "Reality Check" runs for t=1, it fails (because t=1 is gone from the Map). It then looks at the next entry t=2 and succeeds. --- Q2: What if the same timestamp is updated? The Solution When a timestamp is updated with a new price, the Map is updated instantly. A new entry is added to the Heap. The old price entry for that timestamp remains in the Heap as "garbage." Coding Example The "Reality Check" logic: 1. System looks at the heap top: {price: 50, timestamp: 10}. 2. It checks the Map: orders.get(10) is indeed 50. Success! 3. If later the price changes to $200, the $50 entry in the heap will be popped because it no longer matches the Map's truth of $200. --- Q3: How do you handle this in a Distributed System? The Challenge If you have millions of products and thousands of servers, a single memory-based object won't work. The Solution: Redis + Sharding 1. Storage: Use Redis Sorted Sets (ZSET). ZADD inventory <price <timestamp ZRANGE inventory 0 0 WITHSCORES (Gets the minimum). ZREM inventory <timestamp (Removes an order). 2. Sharding: Shard the data by ProductID. Requests for "iPhone 15" always go to Server Group A. Requests for "Pixel 8" always go to Server Group B. 3. Concurrency: Use Redis transactions or Lua scripts to ensure price updates and map updates are