Scaling And Sharding

Database Scaling & Sharding

Core Concepts

As data grows beyond the capacity of a single machine, databases must scale.

1. Vertical vs Horizontal Scaling

• Vertical (Scale-Up): Adding more CPU/RAM to a single server. Easy, but has a hard physical limit.
• Horizontal (Scale-Out): Adding more servers. Requires partitioning data (Sharding).

2. Sharding & Consistent Hashing

Sharding splits a large table across multiple databases based on a Shard Key (e.g., user_id).

• Modulo Hashing: hash(user_id) % N. Fails catastrophically if N (number of servers) changes, requiring massive data reshuffling.
• Consistent Hashing: Maps data and servers onto a logical ring. If a server is added/removed, only the data adjacent to that server is remapped, minimizing disruption.

Consistent Hashing Ring Map

%%{init: {"theme": "default", "flowchart": {"useMaxWidth": true}}}%%
flowchart TD
    subgraph Ring ["Consistent Hash Ring"]
        A["Node A (Hash 0-100)"]
        B["Node B (Hash 101-200)"]
        C["Node C (Hash 201-300)"]
    end
    
    subgraph Router ["API Gateway / DB Router"]
        D["Request: user_id = 45"]
        E["Request: user_id = 150"]
    end
    
    D -->|"Hash(45) -> Route to A"| A
    E -->|"Hash(150) -> Route to B"| B
    
    A -.->|"Replication"| C