🏗️ System Design

RESHADED framework + core building blocks

RESHADED Framework (45-60 min)

Requirements

Functional + non-functional. Users, scale, latency SLA

▼

Estimation

DAU → QPS, storage, bandwidth. Back-of-envelope math

▼

Storage Schema

SQL vs NoSQL. Data model, partition key, indexes

▼

High-Level Design

Block diagram: clients, LB, services, DB, cache, queue

▼

API Design

REST endpoints, auth, pagination, rate limiting

▼

Detailed Design

Deep dive 1-2 critical components. Data flow

▼

Evaluate

Single points of failure, bottlenecks, 10× scaling

▼

Distinctive

Monitoring, CI/CD, gradual rollout, A/B testing

12 Core Concepts

Load Balancing

Round Robin
Least Connections
Consistent Hashing
L4 vs L7

Caching

Cache-Aside
Write-Through
Write-Behind
LRU/LFU
Redis / Memcached

Database Scaling

Read Replicas
Sharding
Denormalization
Connection Pooling

Message Queues

Point-to-Point
Pub/Sub
At-most/least/exactly-once
Kafka, SQS, RabbitMQ

CAP Theorem

CP: Strong consistency (Zookeeper)
AP: High availability (Cassandra)
Pick 2 of 3 in partition

Consistent Hashing

Hash ring
Virtual nodes
Minimal redistribution
Used in: DynamoDB, Cassandra

Rate Limiting

Token Bucket
Sliding Window Log
Sliding Window Counter
Fixed Window

CDN

Pull vs Push
Edge caching
CloudFront / Cloudflare
Static + dynamic content

Back-of-Envelope Numbers

Key Sizes

1 KB 1,024 bytes A short text

1 MB 1,048,576 bytes A photo

1 GB ~1 billion bytes A movie

QPS DAU × actions / 86,400 Peak = 2-3× avg

Storage/yr QPS × size × 86400 × 365 Plan 3-5 years

Memory 1M entries × 1KB = 1GB Fits in RAM

Latency Numbers

L1 cache ref 1 ns

L2 cache ref 4 ns

RAM ref 100 ns

SSD read (4KB) 150 μs

HDD seek 10 ms

Same DC round trip 0.5 ms

Cross-region 150 ms