System Design Roadmap¶

• Roadmap: https://roadmap.sh/system-design

Deep-Dive Sections¶

• 01. Capacity Estimation — QPS, storage, bandwidth, latency budgets
• 02. Trade-offs Framework — CAP, PACELC, consistency vs availability
• 03. Classic Problems — URL shortener, Twitter timeline, WhatsApp/chat, YouTube, Uber, Dropbox, Instagram, Stack Overflow, ad counter, payment system, web crawler, recommendation engine, rate limiter service, key-value store
• 04. Real-World Architectures — Google Spanner, Facebook TAO, Amazon DynamoDB, Netflix stack
• 05. Back-of-Envelope — number tables, Fermi estimation

Companion roadmaps: - Distributed Systems — consensus, replication, sharding, sagas, CRDTs - Architecture / DDD — bounded contexts, aggregates - Computer Science — OS, networking, DB internals

1. Introduction¶

• 1.1 What is System Design?
• 1.2 How to approach System Design?

2.3 Performance vs Scalability¶

2.4 Latency vs Throughput¶

2.5 Availability vs Consistency¶

• 2.5.1 CAP Theorem
• 2.5.1.1 AP - Availability + Partition Tolerance
• 2.5.1.2 CP - Consistency + Partition Tolerance

2.6 Consistency Patterns¶

• 2.6.1 Weak Consistency
• 2.6.2 Eventual Consistency
• 2.6.3 Strong Consistency

2.7 Availability Patterns¶

• 2.7.1 Fail-Over
• 2.7.1.1 Active - Active
• 2.7.1.2 Active - Passive
• 2.7.2 Replication
• 2.7.2.1 Master - Slave
• 2.7.2.2 Master - Master
• 2.7.3 Availability in Numbers
• 2.7.3.1 99.9% Availability - three 9s
• 2.7.3.2 99.99% Availability - four 9s
• 2.7.3.3 Availability in Parallel vs Sequence

3. Background Jobs¶

• 3.1 Event-Driven
• 3.2 Schedule Driven
• 3.3 Returning Results

4. Domain Name System¶

5. Content Delivery Networks¶

• 5.1 Pull CDNs
• 5.2 Push CDNs

6. Load Balancers¶

• 6.1 LB vs Reverse Proxy
• 6.2 Load Balancing Algorithms
• 6.3 Layer 7 Load Balancing
• 6.4 Layer 4 Load Balancing

6.5 Horizontal Scaling¶

7. Application Layer¶

• 7.1 Microservices
• 7.2 Service Discovery

8. Databases¶

• 8.1 Key-Value Store
• 8.2 Document Store
• 8.3 Wide Column Store
• 8.4 Graph Databases
• 8.5 NoSQL
• 8.6 RDBMS
• 8.7 Replication
• 8.8 Sharding
• 8.9 Federation
• 8.10 Denormalization
• 8.11 SQL Tuning

8.12 SQL vs NoSQL¶

9. Caching¶

• 9.1 Refresh Ahead
• 9.2 Write-behind
• 9.3 Write-through
• 9.4 Cache Aside

9.5 Strategies¶

9.6 Types of Caching¶

• 9.6.1 Client Caching
• 9.6.2 CDN Caching
• 9.6.3 Web Server Caching
• 9.6.4 Database Caching
• 9.6.5 Application Caching

10. Asynchronism¶

• 10.1 Back Pressure
• 10.2 Task Queues
• 10.3 Message Queues

11. Communication¶

• 11.1 HTTP
• 11.2 TCP
• 11.3 UDP
• 11.4 RPC
• 11.5 gRPC
• 11.6 REST
• 11.7 GraphQL

11.8 Idempotent Operations¶

12. Performance Antipatterns¶

• 12.1 Improper Instantiation
• 12.2 Monolithic Persistence
• 12.3 Noisy Neighbor
• 12.4 Synchronous I/O
• 12.5 Extraneous Fetching
• 12.6 Busy Database
• 12.7 Busy Frontend
• 12.8 Chatty I/O
• 12.9 Retry Storm
• 12.10 No Caching

13. Monitoring¶

• 13.1 Health Monitoring
• 13.2 Availability Monitoring
• 13.3 Performance Monitoring
• 13.4 Security Monitoring
• 13.5 Usage Monitoring
• 13.6 Instrumentation
• 13.7 Visualization & Alerts

14. Cloud Design Patterns¶

14.1 Design & Implementation¶

• 14.1.1 Strangler Fig
• 14.1.2 Sidecar
• 14.1.3 Static Content Hosting
• 14.1.4 Leader Election
• 14.1.5 CQRS
• 14.1.6 Pipes & Filters
• 14.1.7 Ambassador
• 14.1.8 Gateway Routing
• 14.1.9 Gateway Offloading
• 14.1.10 Gateway Aggregation
• 14.1.11 External Config Store
• 14.1.12 Compute Resource Consolidation
• 14.1.13 Backends for Frontend
• 14.1.14 Anti-Corruption Layer

14.2 Data Management¶

• 14.2.1 Valet Key
• 14.2.2 Static Content Hosting
• 14.2.3 Sharding
• 14.2.4 Materialized View
• 14.2.5 Index Table
• 14.2.6 Event Sourcing
• 14.2.7 CQRS
• 14.2.8 Cache-Aside

14.3 Messaging¶

• 14.3.1 Sequential Convoy
• 14.3.2 Scheduling Agent Supervisor
• 14.3.3 Queue-based Load Leveling
• 14.3.4 Publisher/Subscriber
• 14.3.5 Priority Queue
• 14.3.6 Pipes and Filters
• 14.3.7 Competing Consumers
• 14.3.8 Choreography
• 14.3.9 Claim Check
• 14.3.10 Async Request Reply

15. Reliability Patterns¶

15.1 Availability¶

• 15.1.1 Deployment Stamps
• 15.1.2 Geodes
• 15.1.3 Throttling
• 15.1.4 Health Endpoint Monitoring
• 15.1.5 Queue-Based Load Leveling

15.2 High Availability¶

• 15.2.1 Deployment Stamps
• 15.2.2 Geodes
• 15.2.3 Bulkhead
• 15.2.4 Health Endpoint Monitoring
• 15.2.5 Circuit Breaker

15.3 Resiliency¶

• 15.3.1 Bulkhead
• 15.3.2 Circuit Breaker
• 15.3.3 Compensating Transaction
• 15.3.4 Health Endpoint Monitoring
• 15.3.5 Leader Election
• 15.3.6 Queue-Based Load Leveling
• 15.3.7 Retry
• 15.3.8 Scheduler Agent Supervisor

15.4 Security¶

• 15.4.1 Federated Identity
• 15.4.2 Gatekeeper
• 15.4.3 Valet Key