Case Study 6 — Design Instagram

flowchart TB
    subgraph Client["Clients"]
        Mobile["📱 Mobile App"]
        Web["Web App"]
    end

    subgraph Services["Services"]
        Feed["Feed Service<br/>Fanout / generation"]
        Media["Media Service<br/>Upload + resize"]
        Social["Social Graph<br/>Follow, friends"]
        Story["Story Service<br/>24h expiry"]
        Explore["Explore/Search<br/>Discovery"]
    end

    subgraph Storage["Storage & Data"]
        S3["S3 / GCS<br/>Photos + Videos"]
        Redis["Redis<br/>Feed cache, sessions"]
        PG["PostgreSQL<br/>Users, metadata"]
        Cassandra["Cassandra<br/>Feed, activity"]
        ES["Elasticsearch<br/>Explore, search"]
    end

    Client --> Feed & Media & Social & Story & Explore
    Media --> S3
    Feed --> Redis & Cassandra
    Social --> PG
    Story --> Redis & S3
    Explore --> ES

    style Client fill:#f59e0b,color:#fff
    style Services fill:#7c3aed,color:#fff
    style Storage fill:#3b82f6,color:#fff

flowchart TB
    subgraph Fanout["Fanout on Write (Celebrity)"]
        UserCeleb["Celebrity User<br/>100M followers"] --> FeedCeleb["Don't fanout<br/>Store in celebrity's timeline<br/>Pull on read"]
        Followers["Followers"] --> Pull["Read: Merge<br/>celebrity timeline<br/>+ user feed cache"]
    end

    subgraph Fanout2["Fanout on Write (Normal)"]
        UserNormal["Normal User<br/>1,000 followers"] --> FanoutNormal["Fanout to all followers'<br/>feed caches<br/>(Redis lists)"]
        FanoutNormal --> Followers2["Each follower sees<br/>post immediately"]
    end

    style Fanout fill:#3b82f6,color:#fff
    style Fanout2 fill:#059669,color:#fff