Social timelines and activity streams (X/Twitter, Instagram, LinkedIn, Facebook News Feed) represent classic high-fanout distributed systems problems. When a regular user with 200 followers publishes a post, notifying their followers is computationally trivial. But when an account with 100,000,000 followers posts, a naive fan-out writes 100 million timeline entries simultaneously—triggering database write locks, queue backlogs, and multi-gigabyte cache churn.
This module deconstructs feed architectures: write-time versus read-time materialization, hybrid fan-out routing heuristics, Redis timeline cache management, infinite-scroll cursor pagination, and asynchronous worker queues. You will quantify the exact follower-threshold crossover where push fan-out-on-write must transition to pull fan-out-on-read, design monotonic composite cursor tokens that prevent duplicate or skipped feed items under concurrent insertions, and architect multi-tier timeline caches capable of serving millions of personalized reads per second at sub-50ms tail latency.