System Design

System Design Crash Course

A fast-track, 30-module system design course - foundations, six database deep dives, caching, the distributed-systems stack, streaming, storage & compute, reliability and observability. Built by a Netflix/AWS-led team.

Intermediate1912222 lessons16 free lessonssystem-designhlddistributed-systemsdatabasescaching

System Design Crash Course

A complete, fast-track tour of system design as a discipline - the full vocabulary and the production depth, in 30 focused modules. Authored by Brandon Carter (Amazon L7 Principal, S3 distributed storage) with guest lessons from Nikhil Joshi (Google/Spanner), Emeka Okafor (Netflix/Kafka), and Deepika Nair (Meta/AI Infrastructure).

You'll move through the foundations (what HLD really is, the five design dimensions and their trade-offs, back-of-envelope estimation, the browser-to-server path, API styles, reverse proxies and load balancing, the storage hierarchy, indexing, transactions), six database deep dives (PostgreSQL, MySQL, MongoDB, DynamoDB, Cassandra, Elasticsearch), the caching layer (strategies, Redis & Memcached), the distributed-systems backbone (replication, partitioning and consistent hashing, consistency models, distributed transactions, coordination services), async communication (message queues, Kafka event streaming, Flink & Spark processing), storage and compute (S3, AWS Lambda, Docker & Kubernetes), and the operational layer (reliability patterns and observability with Prometheus).

The first module is free to preview. The full course is a one-time purchase with lifetime access.

What you will learn

• What HLD actually is — vs LLD, vs design docs — and the five design dimensions and their trade-offs
• Back-of-envelope estimation, the browser-to-server path, API styles, reverse proxies and load balancing
• The storage hierarchy, indexing strategies, and transactions / isolation / concurrency control
• Six database deep dives: PostgreSQL, MySQL, MongoDB, DynamoDB, Cassandra, Elasticsearch
• Caching strategies plus Redis & Memcached in practice
• The distributed-systems backbone: replication, partitioning & consistent hashing, consistency models, distributed transactions, coordination services
• Async communication and processing: message queues (RabbitMQ, SQS), Kafka, Flink & Spark
• Storage & compute (S3, Lambda, Docker & Kubernetes) and operations (reliability patterns, observability with Prometheus)

Prerequisites

• Comfort with basic programming and web fundamentals

Course outline

Module 1: What System Design (HLD) Is

What System Design (HLD) Is — 7 topics.

• The Discipline of HLD: What It Is, What It Isn't 10 min read
• HLD vs LLD vs Design Docs vs Architecture 11 min read
• The Five-Component Mental Model 13 min read
• Reading Existing Systems 11 min read
• Failure as a Design Dimension 12 min read
• How HLD Evolves Over a System's Lifetime 12 min read Pro lesson
• Module Reflection from Brandon 12 min read Pro lesson

Module 2: The Five Dimensions and Their Trade-offs

The Five Dimensions and Their Trade-offs — 7 topics.

• Latency, Precisely (P50, P95, P99, P999) 10 min read
• Throughput vs Latency vs Concurrency 10 min read
• Availability: Nines, SLAs, and What 99.9% Actually Allows 10 min read
• Consistency: The Spectrum (Preview) 11 min read
• Durability: Write Durability, Replication, and What "Lost" Means 12 min read
• Trade-offs in Practice: CAP and PACELC 10 min read Pro lesson
• How Senior Engineers Reason About Trade-offs 11 min read Pro lesson

Module 3: Back-of-Envelope Estimation

Back-of-Envelope Estimation — 7 topics.

• The Latency Numbers to Memorize 10 min read
• The Throughput Numbers to Memorize 10 min read Pro lesson
• Storage Estimation Patterns 11 min read Pro lesson
• Traffic Estimation Patterns 11 min read Pro lesson
• Fermi Reasoning Under Uncertainty 11 min read Pro lesson
• Worked Example: Sizing a Social Media Feed 11 min read Pro lesson
• Worked Example: Sizing a Cache Layer 12 min read Pro lesson

Module 4: From Browser to Server

From Browser to Server — 8 topics.

• DNS Resolution End-to-End 10 min read
• TCP Handshake and Connection Management 11 min read Pro lesson
• TLS 1.2 vs TLS 1.3 Handshakes 10 min read Pro lesson
• HTTP/1.1: The Baseline and Its Limits 9 min read Pro lesson
• HTTP/2: Multiplexing, Server Push, Header Compression 10 min read Pro lesson
• HTTP/3 over QUIC 10 min read Pro lesson
• Connection Pooling, Keep-Alive, Head-of-Line Blocking 9 min read
• Where Latency Comes From: The Breakdown 10 min read Pro lesson

Module 5: API Styles

API Styles — 7 topics.

• REST as a Discipline (Not "JSON over HTTP") 11 min read
• gRPC and Protobuf: When Binary Protocols Win 10 min read Pro lesson
• GraphQL: The Actual Trade-offs Beyond the Marketing 11 min read Pro lesson
• WebSocket for Bidirectional Communication 10 min read Pro lesson
• Server-Sent Events for Server Push 9 min read Pro lesson
• Webhooks and Async Callbacks 10 min read Pro lesson
• Choosing Between Styles for a Given System 11 min read Pro lesson

Module 6: Reverse Proxies & Load Balancing (Nginx)

Reverse Proxies & Load Balancing (Nginx) — 7 topics.

• What a Reverse Proxy Is and Why 11 min read
• Layer 4 vs Layer 7 Load Balancing 12 min read Pro lesson
• Load Balancing Algorithms 12 min read Pro lesson
• Sticky Sessions and When They're a Smell 10 min read Pro lesson
• Nginx Configuration Patterns 10 min read Pro lesson
• TLS Termination and Connection Management 10 min read Pro lesson
• Comparison: Nginx vs HAProxy vs Envoy vs ALB 11 min read Pro lesson

Module 7: The Storage Hierarchy

The Storage Hierarchy — 7 topics.

• The Full Hierarchy: Registers to Cold Storage 12 min read
• CPU Caches and Why Application Developers Should Care 11 min read Pro lesson
• RAM Characteristics and Limits 11 min read Pro lesson
• SSDs: How They Actually Work 11 min read Pro lesson
• HDDs: When They're Still Right 10 min read Pro lesson
• Network-Attached Storage (EBS, NFS, S3) 10 min read Pro lesson
• IOPS Economics and the Cost Model 10 min read Pro lesson

Module 8: Indexing Strategies

Indexing Strategies — 7 topics.

• Life Without Indexes 10 min read Pro lesson
• B-trees and B+ trees in Depth 11 min read Pro lesson
• LSM-trees: Optimized for Writes 10 min read Pro lesson
• Hash Indexes 8 min read Pro lesson
• Inverted Indexes (Preview of Elasticsearch) 10 min read Pro lesson
• Bloom Filters and Other Probabilistic Structures 10 min read Pro lesson
• When Indexes Hurt: Write Amplification, Planner Mistakes 9 min read Pro lesson

Module 9: Transactions, Isolation, Concurrency Control

Transactions, Isolation, Concurrency Control — 7 topics.

• ACID Precisely Defined 10 min read Pro lesson
• Isolation Levels: Read Uncommitted to Serializable 10 min read Pro lesson
• Phantom Reads, Lost Updates, Write Skew 11 min read Pro lesson
• MVCC Explained 10 min read Pro lesson
• Locking-Based Concurrency 11 min read Pro lesson
• Optimistic Concurrency 9 min read Pro lesson
• Where Each Isolation Level Breaks 11 min read Pro lesson

Module 10: PostgreSQL Deep Dive

PostgreSQL Deep Dive — 8 topics.

• Why PostgreSQL (The Default-Good Answer) 9 min read Pro lesson
• MVCC Implementation in Postgres 9 min read Pro lesson
• Index Types: B-tree, Hash, GIN, GiST, BRIN 10 min read Pro lesson
• Partial and Expression Indexes 8 min read Pro lesson
• EXPLAIN and EXPLAIN ANALYZE 9 min read Pro lesson
• Replication: Streaming and Logical 9 min read Pro lesson
• Scaling Patterns: Read Replicas, Partitioning, Citus 9 min read Pro lesson
• When PostgreSQL Is the Wrong Answer 9 min read Pro lesson

Module 11: MySQL Deep Dive

MySQL Deep Dive — 7 topics.

• When MySQL Over PostgreSQL 9 min read Pro lesson
• InnoDB Architecture 10 min read Pro lesson
• The Buffer Pool 10 min read Pro lesson
• Clustered Indexes and the Data Layout Difference 11 min read Pro lesson
• Replication: Async, Semi-Sync, Group, GTIDs 10 min read Pro lesson
• MySQL at Scale: Vitess and Sharding 10 min read Pro lesson
• Honest PostgreSQL vs MySQL Comparison 8 min read Pro lesson

Module 12: MongoDB Deep Dive

MongoDB Deep Dive — 7 topics.

• When Document Models Fit 10 min read Pro lesson
• BSON and the Schema-less Myth 8 min read Pro lesson
• Replica Sets and Elections 10 min read Pro lesson
• Sharding: Hashed vs Ranged 10 min read Pro lesson
• Aggregation Pipeline 9 min read Pro lesson
• Indexing Patterns for Documents 9 min read Pro lesson
• Honest Take: When MongoDB Is Wrong 9 min read Pro lesson

Module 13: DynamoDB Deep Dive

DynamoDB Deep Dive — 7 topics.

• The DynamoDB Model 9 min read Pro lesson
• Partition Key as the Central Decision 10 min read Pro lesson
• GSIs vs LSIs 10 min read Pro lesson
• On-Demand vs Provisioned Capacity 9 min read Pro lesson
• Hot Partitions and How to Avoid Them 10 min read Pro lesson
• Single-Table Design 11 min read Pro lesson
• DynamoDB Streams and Event-Driven Patterns 10 min read Pro lesson

Module 14: Cassandra Deep Dive

Cassandra Deep Dive — 7 topics.

• When Cassandra Over DynamoDB 9 min read Pro lesson
• Ring Topology and Consistent Hashing 11 min read Pro lesson
• Tunable Consistency (ONE, QUORUM, ALL) 10 min read Pro lesson
• Partition Keys and Clustering Columns 11 min read Pro lesson
• Compaction Strategies 13 min read Pro lesson
• Anti-patterns: Queue, Joins, Ad-hoc Queries 11 min read Pro lesson
• Real-World Deployments: Netflix, Discord, Apple 10 min read Pro lesson

Module 15: Elasticsearch Deep Dive

Elasticsearch Deep Dive — 7 topics.

• Inverted Indexes in Detail 10 min read Pro lesson
• Segments and Shards 10 min read Pro lesson
• Scoring: TF-IDF and BM25 10 min read Pro lesson
• Refresh Interval and Near-Real-Time 8 min read Pro lesson
• Aggregations 10 min read Pro lesson
• When Elasticsearch Is the Wrong Tool 11 min read Pro lesson
• Alternatives: Meilisearch, Typesense, Postgres FTS 9 min read Pro lesson

Module 16: Caching Strategies

Caching Strategies — 7 topics.

• Caching as Denormalization 11 min read Pro lesson
• Cache-Aside, Write-Through, Write-Back, Write-Around 11 min read Pro lesson
• TTL Strategies 10 min read Pro lesson
• Cache Invalidation 10 min read Pro lesson
• Cache Stampedes and Prevention 11 min read Pro lesson
• CDN as a Cache 11 min read Pro lesson
• When NOT to Cache 11 min read Pro lesson

Module 17: Redis & Memcached

Redis & Memcached — 8 topics.

• Redis vs Memcached: the High-Level Comparison 9 min read Pro lesson
• Redis Data Structures 12 min read Pro lesson
• Redis Persistence: RDB and AOF 10 min read Pro lesson
• Redis Replication and Sentinel 10 min read Pro lesson
• Redis Cluster 10 min read Pro lesson
• Memcached: When Simpler Is Better 9 min read Pro lesson
• Redis as a Primary Database 9 min read Pro lesson
• The Decision Framework, Summed Up 8 min read Pro lesson

Module 18: Replication

Replication — 8 topics.

• Why We Replicate 10 min read Pro lesson
• Leader-Follower (Single-Leader) Replication 10 min read Pro lesson
• Multi-Leader Replication 11 min read Pro lesson
• Leaderless Replication (Dynamo-Style) 12 min read Pro lesson
• Synchronous vs Asynchronous Replication 9 min read Pro lesson
• Replication Lag and the Bugs It Causes 10 min read Pro lesson
• Read-Your-Writes, Monotonic Reads, Consistent Prefix Reads 12 min read Pro lesson
• Guest: Nikhil on Google Spanner at Scale 13 min read Pro lesson

Module 19: Partitioning & Consistent Hashing

Partitioning & Consistent Hashing — 7 topics.

• Why Partition 9 min read Pro lesson
• Range-Based Partitioning 10 min read Pro lesson
• Hash-Based Partitioning 10 min read Pro lesson
• Consistent Hashing in Detail 12 min read Pro lesson
• Virtual Nodes 10 min read Pro lesson
• Rebalancing When Topology Changes 11 min read Pro lesson
• Hot Spots and Mitigation 12 min read Pro lesson

Module 20: Consistency Models

Consistency Models — 7 topics.

• Why Consistency Is Hard 7 min read Pro lesson
• CAP Theorem Properly Explained 9 min read Pro lesson
• PACELC: The More Useful Framing 8 min read Pro lesson
• Linearizability 7 min read Pro lesson
• Sequential and Causal Consistency 8 min read Pro lesson
• Eventual Consistency in Practice 7 min read Pro lesson
• Real-World Choices: AP vs CP 9 min read Pro lesson

Module 21: Distributed Transactions

Distributed Transactions — 7 topics.

• Why Distributed Transactions Are Hard 8 min read Pro lesson
• Two-Phase Commit (2PC) 8 min read Pro lesson
• Why 2PC Mostly Sucks 8 min read Pro lesson
• Saga Pattern (Orchestration vs Choreography) 9 min read Pro lesson
• Outbox Pattern for Reliable Event Publishing 8 min read Pro lesson
• Idempotency Keys as the Practical Answer 8 min read Pro lesson
• Real-World Patterns: Stripe's Idempotency Model 9 min read Pro lesson

Module 22: Coordination Services

Coordination Services — 7 topics.

• What Coordination Is and Why It's Hard 8 min read Pro lesson
• ZooKeeper's Data Model: Znodes and Watches 9 min read Pro lesson
• Ephemeral and Sequential Nodes 8 min read Pro lesson
• ZAB Protocol Overview 9 min read Pro lesson
• Leader Election Patterns 8 min read Pro lesson
• Distributed Locks (and Why They're a Smell) 9 min read Pro lesson
• etcd vs ZooKeeper, and Kubernetes's Choice 10 min read Pro lesson

Module 23: Message Queues (RabbitMQ & SQS)

Message Queues (RabbitMQ & SQS) — 8 topics.

• The Queue Model 8 min read Pro lesson
• RabbitMQ: Exchanges, Queues, Bindings 8 min read Pro lesson
• RabbitMQ: Exchange Types (Direct, Fanout, Topic, Headers) 6 min read Pro lesson
• Dead-Letter Queues and Poison Messages 7 min read Pro lesson
• SQS: Standard vs FIFO 6 min read Pro lesson
• SQS: Visibility Timeout and Long Polling 7 min read Pro lesson
• RabbitMQ vs SQS: When Each Fits 6 min read Pro lesson
• When Queues Are Wrong (Use Streams Instead) 9 min read Pro lesson

Module 24: Event Streaming (Kafka)

Event Streaming (Kafka) — 8 topics.

• The Log Abstraction 7 min read Pro lesson
• Topics, Partitions, Offsets 7 min read Pro lesson
• Producers and the Partitioning Strategy 6 min read Pro lesson
• Consumer Groups and Rebalancing 6 min read Pro lesson
• Exactly-Once Semantics 6 min read Pro lesson
• Log Compaction 5 min read Pro lesson
• Kafka Connect and Schema Registry 6 min read Pro lesson
• Guest: Emeka on Netflix's 8B-Events/Day Pipeline 9 min read Pro lesson

Module 25: Stream & Batch Processing (Flink & Spark)

Stream & Batch Processing (Flink & Spark) — 8 topics.

• Batch vs Stream: The Conceptual Difference 7 min read Pro lesson
• Spark: RDDs, DataFrames, Catalyst 6 min read Pro lesson
• Spark Streaming and Structured Streaming 5 min read Pro lesson
• Flink: The Streaming-First Model 6 min read Pro lesson
• Event Time vs Processing Time 6 min read Pro lesson
• Windowing and Watermarks 7 min read Pro lesson
• Exactly-Once Across the Pipeline 6 min read Pro lesson
• Lambda vs Kappa Architectures 8 min read Pro lesson

Module 26: Object Storage (S3 Deep Dive)

Object Storage (S3 Deep Dive) — 7 topics.

• The Object Storage Abstraction 6 min read Pro lesson
• Buckets, Keys, Prefixes 5 min read Pro lesson
• S3 Consistency Model (and the December 2020 Change) 6 min read Pro lesson
• Storage Classes and Lifecycle Policies 6 min read Pro lesson
• Pre-Signed URLs and Access Patterns 6 min read Pro lesson
• Multipart Upload 5 min read Pro lesson
• S3 as a Data Lake Substrate 8 min read Pro lesson

Module 27: AWS Lambda Deep Dive

AWS Lambda Deep Dive — 7 topics.

• The Lambda Execution Model 6 min read Pro lesson
• Cold Starts: Causes and Mitigations 6 min read Pro lesson
• Concurrency Limits and Quotas 5 min read Pro lesson
• Lambda + API Gateway 4 min read Pro lesson
• Lambda + SQS / Kinesis 4 min read Pro lesson
• When Serverless Wins 4 min read Pro lesson
• When Serverless Loses 7 min read Pro lesson

Module 28: Containers & Orchestration (Docker & Kubernetes)

Containers & Orchestration (Docker & Kubernetes) — 9 topics.

• Why Containers (vs VMs, vs Bare Processes) 5 min read Pro lesson
• Docker: Images, Layers, Union Filesystem 5 min read Pro lesson
• Docker: Networking and Storage 2 min read Pro lesson
• Why Orchestration 3 min read Pro lesson
• Kubernetes: Pods, Services, Deployments 5 min read Pro lesson
• Kubernetes: StatefulSets, DaemonSets, Jobs 2 min read Pro lesson
• Kubernetes Control Plane Internals 4 min read Pro lesson
• When Kubernetes Is Overkill 3 min read Pro lesson
• Guest: Deepika on Kubernetes for Distributed ML Training 7 min read Pro lesson

Module 29: Reliability Patterns

Reliability Patterns — 8 topics.

• Idempotency at the API Level 4 min read Pro lesson
• Retry Strategies and Exponential Backoff 4 min read Pro lesson
• The Thundering Herd Problem 3 min read Pro lesson
• Circuit Breakers 4 min read Pro lesson
• Rate Limiting: Token Bucket, Leaky Bucket, Sliding Window 5 min read Pro lesson
• Backpressure 4 min read Pro lesson
• Bulkheads 3 min read Pro lesson
• Graceful Degradation 5 min read Pro lesson

Module 30: Observability (Prometheus)

Observability (Prometheus) — 9 topics.

• Metrics, Logs, Traces: The Three Pillars 4 min read Pro lesson
• SLIs, SLOs, SLAs 4 min read Pro lesson
• The RED Method (Rate, Errors, Duration) 2 min read Pro lesson
• The USE Method (Utilization, Saturation, Errors) 3 min read Pro lesson
• Prometheus: Pull Model and Time-Series Storage 3 min read Pro lesson
• PromQL 2 min read Pro lesson
• Alertmanager 3 min read Pro lesson
• OpenTelemetry as the Future of Instrumentation 3 min read Pro lesson
• Course Closing: What's Next 5 min read Pro lesson

Browse all courses