What is Interview Forge?

Interview Forge helps you practice mock interviews, follow study plans, track career goals, review learning analytics, and prepare across 36 pages.

Design Instagram (Photo & Feed) at Google — System Design Interview Guide

The problem at Google

Functional: post photos with captions, follow users, see a ranked feed of followed users' posts, like and comment.
Non-functional: feed loads fast (<200ms), posts appear within seconds, the system tolerates celebrity-scale fan-out.
Scale: ~1B users, ~100M posts/day, overwhelmingly read-heavy (feed views >> posts).
Consistency: likes/comments can be eventually consistent; feed ordering must stay monotonic (no flicker).

Out of scope (covered in detail below)

Stories, direct messages, ads/ranking ML training, explore/discovery, and spam detection — deferred and discussed in the dedicated section.

At Google, this problem is framed around algorithmic rigor and generalist problem-solving. Expect Google-specific scale numbers and a focus on production readiness and the company's engineering bar.

Capacity estimation

Resource	Estimate	Notes
Posts/day	~100M	Each with media + metadata
Feed reads/day	~billions	Read-heavy, ~100x writes
Media storage	~petabytes	Originals + thumbnails + multiple resolutions
Feed cache (Redis)	~terabytes	Pre-computed feed lists per user

Architecture overview

Post + feed architecture

+--------+  post   +-------------+  media   +-----------+  CDN  +--------+
| Client |--------->| Post Service|--------->| Object S3 |------->| Client |
+--------+          +-------------+          +-----------+        +--------+
                          |
                          v
                  +---------------+   fan-out   +-----------------+
                  | Feed Service  |------------>| Feed Cache (Redis)|
                  +---------------+             +-----------------+
                          ^
                          |
                  +---------------+
                  | Graph Service |  (who follows whom)
                  +---------------+

Media goes to object storage; the feed service builds timelines via fan-out.

The feed fan-out decision

This is the canonical feed problem and the heart of the interview. The choice is about where to do the work — at write time or read time.

Fan-out on write: when a user posts, push the post_id into each follower's pre-computed feed list (Redis sorted set). Fast reads, but celebrities with millions of followers make writes explosively expensive.
Fan-out on read: pull and merge recent posts from followed users at read time. Cheap writes, but heavier reads and harder ranking.
Hybrid (the production answer): fan-out on write for normal users, pull-on-read for celebrities — avoids write amplification for huge followings while keeping reads fast for everyone.

Unlock the full guide

Premium members get every DSA, system design, company and role interview guide — plus AI practice, mock interviews, and analytics.

Unlock with Premium