System Design Fundamentals
Overview
Monolith vs Microservices, service boundaries, API contracts, dan scalability patterns untuk production systems.
Decision Framework
┌─────────────────────────────────────────────────────────────┐
│ SYSTEM DESIGN DECISION TREE │
└─────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────┐
│ Team Size? │
└─────────────────────────────┘
│ │
< 10 people > 20 people
│ │
▼ ▼
┌─────────┐ ┌─────────────────┐
│ Monolith│ │ Microservices │
│ Start │ │ Consider │
└─────────┘ │ carefully │
└─────────────────┘
Monolith vs Microservices
Monolith - When to Use ✅
- Team size < 10 engineers
- Early startup/MVP phase (< 6 months)
- Tightly coupled features
- Simple domain
- Need fast iteration
- Limited DevOps maturity
Microservices - When to Use ✅
- Team size > 20 engineers
- Multiple independent products
- Different scaling requirements per service
- Polyglot needs (different tech stacks)
- Need independent deployments
- Complex domain with clear boundaries
Warning Signs (Time to Migrate)
| Sign |
Indicator |
| Build time |
> 10 minutes |
| Deploy coupling |
Teams blocking each other |
| Scaling |
Different parts need different resources |
| Tech stack |
One size doesn't fit all |
Service Boundaries
Bounded Context Pattern
┌─────────────────────────────────────────────────────────────┐
│ E-COMMERCE DOMAIN │
├─────────────────┬─────────────────┬─────────────────────────┤
│ Catalog │ Order │ User │
│ Context │ Context │ Context │
├─────────────────┼─────────────────┼─────────────────────────┤
│ - Products │ - Cart │ - Profile │
│ - Categories │ - Checkout │ - Auth │
│ - Inventory │ - Payments │ - Preferences │
│ │ - Shipping │ │
└─────────────────┴─────────────────┴─────────────────────────┘
│ │ │
▼ ▼ ▼
┌──────────┐ ┌──────────┐ ┌──────────┐
│ Catalog │ │ Order │ │ User │
│ Service │ │ Service │ │ Service │
└──────────┘ └──────────┘ └──────────┘
Communication Patterns
| Pattern |
Use Case |
Pros |
Cons |
| REST |
CRUD, synchronous |
Simple, familiar |
Coupling |
| gRPC |
High perf, internal |
Fast, type-safe |
Learning curve |
| Message Queue |
Async, decoupled |
Resilient, scalable |
Complexity |
| Event Bus |
Cross-service events |
Loose coupling |
Hard to trace |
Scalability Patterns
Horizontal vs Vertical Scaling
Vertical Scaling Horizontal Scaling
┌───────────────────┐ ┌─────┐ ┌─────┐ ┌─────┐
│ Server │ │ SV1 │ │ SV2 │ │ SV3 │
│ ┌─────┐ │ → └─────┘ └─────┘ └─────┘
│ │ CPU │ + │ Load Balancer
│ │ RAM │ │
│ └─────┘ │
│ │ Pros: Infinite scale, resilient
│ Expensive │ Cons: Distributed complexity
│ Limited │
└───────────────────┘
Database Scaling
┌──────────────┐
│ Application │
└──────┬───────┘
│
┌────────────┼────────────┐
▼ ▼ ▼
┌────────┐ ┌──────────┐ ┌─────────┐
│ Read │ │ Write │ │ Cache │
│ Replica│ │ Master │ │ (Redis) │
└────────┘ └──────────┘ └─────────┘
│ │
▼ ▼
┌─────────────────────────┐
│ Sharding │
│ (when single DB fails)│
└─────────────────────────┘
Checklist
Phase 1: Discovery
Phase 2: Architecture
Phase 3: Implementation
Phase 4: Operations
Tradeoffs Summary
| Aspect |
Monolith |
Microservices |
| Deployment |
Simple |
Complex |
| Scaling |
Limited |
Infinite |
| Debugging |
Easy |
Hard |
| Team Autonomy |
Low |
High |
| Data Consistency |
Easy (ACID) |
Eventual |
| Infrastructure |
Minimal |
Extensive |
| Time to MVP |
Fast |
Slow |
Common Mistakes
| Mistake |
Solution |
| Premature microservices |
Start monolith, extract when needed |
| No clear boundaries |
Use Domain-Driven Design |
| Distributed monolith |
Actually decouple or stay monolithic |
| Forgetting operations |
Plan monitoring from day 1 |
Resources
modbender