Writing system design by data flow to understand to design your system design better.

1. Scalability

1. Vertical vs. Horizontal Scaling

2. Partitioning — Divide and Conquer

3. Replication — Redundancy for Speed and Safety

• Synchronous Replication

• Asynchronous Replication

• Leader-Follower Model

1. Caching — The Shortcut to Scale

• In-Memory Caches

• Edge Caches

• Application Caches

• TTL (Time to Live)

• Write-through

• Cache-aside

1. Elasticity — Scaling as a Living System

• Auto-scaling groups

• Serverless platforms: event-driven.

• Container orchestration (Kubernetes)

• Observability and automation.

1. Load Distribution and Partitioning

• Traffic Distribution

• Data Distribution

• Task Distribution

1. Sharding — Divide and Conquer

• Throughput

• Storage

• Isolation

1. Consistent Hashing

2. Data Locality

3. Balancing the System — Avoiding Hotspots

2. Bottlenecks & Backpressures

1. Buffering — The Elastic Pause

2. Batching — Efficiency Through Grouping

3. Flow Control Strategies

• Credit-Based Flow Control: maximum number of consumer can handle.

• Rate Limiting

• Reactive Streams: producer request N messages -> consumer consume N messages.

1. Consumer Lag

• Use bounded queues

• Monitor queue depth, latency, and throughput

3. Consistency

1. CAP Theorem

2. Eventual Consistency

3. The Raft Algorithm — Consensus + The Paxos

• Leader election

• Log replication

• Safety guarantees

1. Coordination Beyond Consensus

• Distributed Locks

• Barriers

• Leases

1. Quorums — The Math Behind Consensus