Skip to content

System Design

"Design is not just what it looks like and feels like. Design is how it works." — Steve Jobs

System design is the process of defining the architecture, components, modules, interfaces, and data for a system to satisfy specified requirements.

  • Scalability

    Vertical vs horizontal scaling, load balancing, caching, and sharding strategies.

    Scalability Guide

  • API Design

    REST, GraphQL, gRPC patterns and best practices for service interfaces.

    API Design Guide

The Three Pillars

Every production system must balance these concerns:

System Design Pillars

Note: Click the diagram to view/edit the Excalidraw source.

Pillar Definition Key Practices
Reliability System works correctly despite faults Redundancy, failover, testing, monitoring
Scalability System handles increased load Load balancing, caching, sharding, CDN
Maintainability System is easy to operate and evolve Clean code, documentation, observability

Core Concepts

CAP Theorem

In a distributed data store, you can only guarantee two of the three:

CAP Theorem

Note: Click the diagram to view/edit the Excalidraw source.

Property Description
Consistency Every read receives the most recent write or an error
Availability Every request receives a response (not necessarily the latest)
Partition Tolerance System works despite network failures between nodes

Reality Check

Partition tolerance is mandatory in distributed systems. You choose between CP (Consistency + Partition Tolerance) and AP (Availability + Partition Tolerance).

ACID vs. BASE

ACID (RDBMS) BASE (NoSQL)
Atomicity Basically Available
Consistency Soft state
Isolation Eventual consistency
Durability
Strong consistency High availability

Back-of-the-Envelope Math

Essential numbers for capacity planning and system design interviews.

Powers of Two

Power Value Common Name
$2^{10}$ ~1,000 1 KB
$2^{20}$ ~1,000,000 1 MB
$2^{30}$ ~1,000,000,000 1 GB
$2^{40}$ ~1,000,000,000,000 1 TB

Latency Numbers

Latency Comparison

Note: Click the diagram to view/edit the Excalidraw source.

Operation Latency
L1 Cache Reference 0.5 ns
Mutex Lock/Unlock 100 ns
Main Memory Reference 100 ns
Send 2KB over 1Gbps 20,000 ns (20 us)
Read 1MB from Memory 250,000 ns (250 us)
Datacenter Round Trip 500,000 ns (500 us)
Disk Seek 10,000,000 ns (10 ms)
CA to Netherlands Round Trip 150,000,000 ns (150 ms)

System Design Interview Framework

Interview Framework

Note: Click the diagram to view/edit the Excalidraw source.

Deep Dives