Preview — Pro guide
You are seeing a portion of this guide. Sign in and upgrade to unlock the full article, quizzes, and interview answers.
Design Twitter/X Feed (News Feed)
End-to-end design of a social media news feed serving 200 million daily active users. The central design challenge is the fan-out problem: how to show a user their personalized timeline of tweets from all the accounts they follow, with < 200ms latency, at 180K reads/sec.
The Core Problem: Feed Generation
A Twitter-like feed must show User A the most recent tweets from all accounts User A follows, ranked by recency (and optionally by engagement). The challenge is scale: 500M total users, 200M daily active users, 50:1 read-to-write ratio (20B reads/day vs ~400M writes/day). The central trade-off is where to pay the computational cost: at write time (pre-compute each follower's feed when a tweet is posted) or at read time (compute the feed on-the-fly when the user opens their app). Neither extreme works at full scale — the answer is a hybrid.
Scale Assumptions
| Metric | Value |
|---|---|
| Total users | 500M |
| Daily active users (DAU) | 200M |
| Average follows per user | 200 |
| Tweet writes/day | ~400M (~5K/sec avg, ~15K/sec peak) |
| Timeline reads/day | ~20B (~230K/sec avg, ~700K/sec peak) |
| Read-to-write ratio | 50:1 |
| Tweet size (text + metadata) | ~1KB |
| Media (images/video) | Served separately via CDN |
| 10-year storage (tweets only) | ~19 PB |