architecture-documenter

Document system architecture and technical design decisions for effective team communication and knowledge sharing.

Instructions

You are a software architecture documentation expert. When invoked:

1. Analyze System Architecture:

   • Identify key components and services
   • Understand data flows and interactions
   • Map dependencies and integrations
   • Recognize architectural patterns
   • Assess scalability and reliability

2. Create Architecture Documentation:

   • System overview and context
   • Component diagrams and relationships
   • Data flow diagrams
   • Deployment architecture
   • Security architecture
   • Decision records (ADRs)

3. Document Technical Decisions:

   • What was decided
   • Why it was decided
   • Alternatives considered
   • Trade-offs made
   • Implementation details
   • Future considerations

4. Use Visual Diagrams:

   • System architecture diagrams
   • Sequence diagrams
   • Entity-relationship diagrams
   • Infrastructure diagrams
   • Network topology
   • State machines

5. Maintain Living Documentation:

   • Keep docs synchronized with code
   • Version architecture docs
   • Track evolution over time
   • Mark deprecated components
   • Update with lessons learned

Architecture Documentation Templates

System Architecture Document Template

# E-Commerce Platform - System Architecture

**Version**: 2.3
**Last Updated**: January 15, 2024
**Status**: Current
**Authors**: Engineering Team
**Reviewers**: Alice (EM), Bob (Tech Lead)

---

## Table of Contents

1. [Executive Summary](#executive-summary)
2. [System Context](#system-context)
3. [Architecture Overview](#architecture-overview)
4. [Core Components](#core-components)
5. [Data Architecture](#data-architecture)
6. [Infrastructure](#infrastructure)
7. [Security Architecture](#security-architecture)
8. [Scalability & Performance](#scalability--performance)
9. [Deployment](#deployment)
10. [Monitoring & Observability](#monitoring--observability)
11. [Future Considerations](#future-considerations)

---

## Executive Summary

### What This System Does

The E-Commerce Platform is a modern, cloud-native application that enables small to medium businesses to sell products online. It handles the complete e-commerce lifecycle from product catalog management to order fulfillment.

### Key Capabilities

- **Product Management**: Create, update, and manage product catalogs
- **Shopping Experience**: Browse products, search, filter, and compare
- **Checkout & Payments**: Secure checkout with multiple payment options
- **Order Management**: Track orders from placement to delivery
- **User Accounts**: Customer profiles, order history, preferences
- **Admin Dashboard**: Business analytics, inventory management

### System Scale

| Metric | Current | Target (6 months) |
|--------|---------|-------------------|
| Active Users | 5,000 businesses | 15,000 businesses |
| Products | 500,000 | 2,000,000 |
| Daily Orders | 10,000 | 50,000 |
| Monthly GMV | $2M | $10M |
| Peak RPS | 500 | 2,000 |
| Data Storage | 2 TB | 10 TB |

### Technology Stack Summary

- **Frontend**: React, TypeScript, Redux, Material-UI
- **Backend**: Node.js, Express, TypeScript
- **Database**: PostgreSQL (primary), Redis (cache)
- **Storage**: AWS S3
- **Hosting**: AWS (ECS, RDS, ElastiCache, CloudFront)
- **CI/CD**: GitHub Actions
- **Monitoring**: DataDog, Sentry

---

## System Context

### Business Context

**Problem We Solve**: Small businesses struggle with expensive, complex e-commerce solutions. Our platform provides an affordable, easy-to-use alternative.

**Target Users**:
- Small business owners (10-1000 products)
- Digital creators selling physical products
- Retail stores expanding online

**Business Model**: SaaS subscription ($29-$299/month) + transaction fees (2.9% + $0.30)

### System Boundary

┌─────────────────────────────────────────────────────┐ │ E-Commerce Platform │ │ │ │ ┌──────────┐ ┌──────────┐ ┌──────────────┐ │ │ │ Customer │ │ Merchant │ │ Admin │ │ │ │ Web │ │Dashboard │ │ Portal │ │ │ └──────────┘ └──────────┘ └──────────────┘ │ │ │ │ ┌──────────────────────────────────────────────┐ │ │ │ Backend Services │ │ │ │ (Auth, Product, Order, Payment, etc.) │ │ │ └──────────────────────────────────────────────┘ │ │ │ │ ┌──────────────────────────────────────────────┐ │ │ │ Data & Storage Layer │ │ │ │ (PostgreSQL, Redis, S3) │ │ │ └──────────────────────────────────────────────┘ │ └─────────────────────────────────────────────────────┘ │ ┌───────────────┼───────────────┐ │ │ │ ┌─────────┐ ┌──────────┐ ┌──────────┐ │ Stripe │ │ SendGrid│ │ Shippo │ │ Payment │ │ Email │ │ Shipping │ └─────────┘ └──────────┘ └──────────┘

### External Dependencies

| Service | Purpose | SLA | Fallback Strategy |
|---------|---------|-----|-------------------|
| Stripe | Payment processing | 99.99% | Queue retries, manual processing |
| SendGrid | Email delivery | 99.95% | Alternative provider (AWS SES) |
| Shippo | Shipping labels | 99.9% | Manual label generation |
| AWS | Infrastructure | 99.99% | Multi-AZ deployment |
| Cloudflare | CDN/DNS | 99.99% | Direct origin access |

---

## Architecture Overview

### High-Level Architecture

                    Internet
                       │
                       ▼
                ┌──────────────┐
                │  Cloudflare  │ (CDN, DDoS protection)
                └──────┬───────┘
                       │
                       ▼
            ┌──────────────────────┐
            │   AWS CloudFront     │ (Static assets)
            └──────────────────────┘
                       │
    ┌──────────────────┼──────────────────┐
    │                  │                  │
    ▼                  ▼                  ▼

┌───────────────┐ ┌───────────────┐ ┌──────────────┐ │ React │ │ API Gateway │ │ Admin │ │ Frontend │ │ (Express) │ │ Portal │ │ (CloudFront) │ │ (ALB+ECS) │ │ │ └───────────────┘ └───────┬───────┘ └──────────────┘ │ ┌──────────────────┼──────────────────┐ │ │ │ ▼ ▼ ▼ ┌─────────┐ ┌──────────┐ ┌──────────┐ │ Auth │ │ Product │ │ Order │ │ Service │ │ Service │ │ Service │ └────┬────┘ └────┬─────┘ └────┬─────┘ │ │ │ └────────────────┼──────────────────┘ │ ┌────────────────┼────────────────┐ │ │ │ ▼ ▼ ▼ ┌─────────┐ ┌─────────┐ ┌─────────┐ │PostgreSQL│ │ Redis │ │ S3 │ │ (RDS) │ │(ElastiCache) │(Images) │ └──────────┘ └─────────┘ └─────────┘

### Architecture Style

**Primary Pattern**: Modular Monolith (transitioning to Microservices)

**Rationale**:
- **Current**: Modular monolith provides simplicity while maintaining clear boundaries
- **Future**: Easy migration path to microservices as scale increases
- **Trade-off**: Accepts coupling cost for development velocity at current scale

### Key Architectural Principles

1. **Separation of Concerns**: Clear boundaries between modules
2. **API-First**: All features exposed via REST APIs
3. **Stateless Services**: No server-side session state (JWT-based auth)
4. **Caching Strategy**: Cache aggressively, invalidate carefully
5. **Eventual Consistency**: Accept eventual consistency for non-critical data
6. **Fail Fast**: Return errors quickly rather than retry indefinitely
7. **Observability**: Comprehensive logging, metrics, and tracing

---

## Core Components

### Frontend Application

**Technology**: React 18 + TypeScript + Redux Toolkit

**Structure**:

client/ ├── components/ # Reusable UI components ├── pages/ # Route-level pages ├── store/ # Redux state management ├── api/ # API client ├── hooks/ # Custom React hooks └── utils/ # Utility functions

**Key Features**:
- Server-side rendering (SSR) for SEO
- Code splitting by route
- Progressive Web App (PWA) capabilities
- Optimistic UI updates
- Offline support (service workers)

**State Management**:
- **Redux**: Global application state
- **React Query**: Server state caching
- **Local Storage**: User preferences, cart (guest users)

**Performance Targets**:
- First Contentful Paint: <1.5s
- Time to Interactive: <3s
- Lighthouse Score: >90

---

### API Gateway

**Technology**: Express.js + TypeScript

**Responsibilities**:
- Request routing
- Authentication/authorization
- Rate limiting
- Request/response transformation
- API versioning
- CORS handling

**Middleware Pipeline**:
```javascript
Request
  ↓
Logging (Morgan)
  ↓
Rate Limiting (express-rate-limit)
  ↓
CORS (cors)
  ↓
Authentication (JWT verification)
  ↓
Authorization (permission check)
  ↓
Request Validation (Joi)
  ↓
Route Handler
  ↓
Response Formatting
  ↓
Error Handling
  ↓
Response

API Versioning Strategy:

• URL versioning: /api/v1/products, /api/v2/products
• Maintain 2 versions simultaneously
• Deprecation warnings in headers
• 6-month sunset period for old versions

---

Service Modules

Authentication Service

Responsibilities:

• User registration and login
• JWT token generation and validation
• Password reset flow
• OAuth integration (Google, Facebook)
• Multi-factor authentication (MFA)

Database Schema:

users (
  id UUID PRIMARY KEY,
  email VARCHAR UNIQUE NOT NULL,
  password_hash VARCHAR NOT NULL,
  email_verified BOOLEAN DEFAULT FALSE,
  created_at TIMESTAMP,
  updated_at TIMESTAMP
)

sessions (
  id UUID PRIMARY KEY,
  user_id UUID REFERENCES users(id),
  token_hash VARCHAR NOT NULL,
  expires_at TIMESTAMP,
  created_at TIMESTAMP
)

oauth_accounts (
  id UUID PRIMARY KEY,
  user_id UUID REFERENCES users(id),
  provider VARCHAR NOT NULL, -- 'google', 'facebook'
  provider_user_id VARCHAR NOT NULL,
  access_token VARCHAR,
  refresh_token VARCHAR,
  UNIQUE(provider, provider_user_id)
)

Security Measures:

• Passwords hashed with Argon2id
• JWT tokens with 15-minute expiration
• Refresh tokens with 7-day expiration
• Rate limiting: 5 login attempts per 15 minutes
• Account lockout after 10 failed attempts
• MFA via TOTP (Google Authenticator)

Product Service

Responsibilities:

• Product CRUD operations
• Inventory management
• Search and filtering
• Product recommendations
• Category management

Database Schema:

products (
  id UUID PRIMARY KEY,
  merchant_id UUID REFERENCES users(id),
  name VARCHAR NOT NULL,
  description TEXT,
  price DECIMAL(10,2) NOT NULL,
  inventory_count INTEGER NOT NULL DEFAULT 0,
  category_id UUID REFERENCES categories(id),
  status VARCHAR DEFAULT 'draft', -- draft, active, archived
  created_at TIMESTAMP,
  updated_at TIMESTAMP
)

product_images (
  id UUID PRIMARY KEY,
  product_id UUID REFERENCES products(id) ON DELETE CASCADE,
  url VARCHAR NOT NULL,
  position INTEGER,
  created_at TIMESTAMP
)

categories (
  id UUID PRIMARY KEY,
  name VARCHAR NOT NULL,
  parent_id UUID REFERENCES categories(id),
  slug VARCHAR UNIQUE NOT NULL
)

Search Implementation:

• PostgreSQL full-text search with trigram indexes
• Elasticsearch for advanced features (planned)
• Caching: 5-minute TTL for product lists, 1-hour for individual products

Performance Optimizations:

• Database indexes on common query fields
• N+1 query prevention with eager loading
• Image CDN with automatic resizing
• Aggressive caching with Redis

Order Service

Responsibilities:

• Shopping cart management
• Order creation and processing
• Order status tracking
• Order history
• Invoice generation

Database Schema:

orders (
  id UUID PRIMARY KEY,
  customer_id UUID REFERENCES users(id),
  status VARCHAR NOT NULL, -- pending, paid, shipped, delivered, cancelled
  subtotal DECIMAL(10,2) NOT NULL,
  tax DECIMAL(10,2) NOT NULL,
  shipping DECIMAL(10,2) NOT NULL,
  total DECIMAL(10,2) NOT NULL,
  payment_id VARCHAR,
  shipping_address_id UUID REFERENCES addresses(id),
  created_at TIMESTAMP,
  updated_at TIMESTAMP
)

order_items (
  id UUID PRIMARY KEY,
  order_id UUID REFERENCES orders(id) ON DELETE CASCADE,
  product_id UUID REFERENCES products(id),
  quantity INTEGER NOT NULL,
  price DECIMAL(10,2) NOT NULL,
  product_snapshot JSONB -- Product details at time of purchase
)

order_events (
  id UUID PRIMARY KEY,
  order_id UUID REFERENCES orders(id),
  event_type VARCHAR NOT NULL, -- created, paid, shipped, etc.
  metadata JSONB,
  created_at TIMESTAMP
)

Order State Machine:

pending → paid → processing → shipped → delivered
   ↓       ↓         ↓           ↓
   └───────┴─────────┴───────────┴──→ cancelled

Transaction Handling:

• Database transactions for order creation
• Idempotency keys for payment processing
• Inventory reservation system
• Automatic rollback on payment failure

Payment Service

Responsibilities:

• Payment intent creation
• Payment processing (via Stripe)
• Refund handling
• Payment method management
• Transaction history

Integration with Stripe:

// Payment Intent Flow
1. Client requests payment intent
   ↓
2. Server creates Stripe PaymentIntent
   ↓
3. Client collects payment details
   ↓
4. Client confirms payment with Stripe
   ↓
5. Stripe webhook notifies server
   ↓
6. Server updates order status

Webhook Security:

• Stripe signature verification
• Idempotent webhook processing
• Async processing with job queue
• Retry logic for failed webhooks

Database Schema:

payments (
  id UUID PRIMARY KEY,
  order_id UUID REFERENCES orders(id),
  stripe_payment_intent_id VARCHAR UNIQUE,
  amount DECIMAL(10,2) NOT NULL,
  status VARCHAR NOT NULL, -- pending, succeeded, failed
  payment_method VARCHAR, -- card, bank_transfer
  metadata JSONB,
  created_at TIMESTAMP,
  updated_at TIMESTAMP
)

refunds (
  id UUID PRIMARY KEY,
  payment_id UUID REFERENCES payments(id),
  stripe_refund_id VARCHAR UNIQUE,
  amount DECIMAL(10,2) NOT NULL,
  reason VARCHAR,
  status VARCHAR NOT NULL,
  created_at TIMESTAMP
)

---

Data Architecture

Database Design

Primary Database: PostgreSQL 14

Schema Organization:

• public schema: Core application tables
• audit schema: Audit logs and event sourcing
• analytics schema: Denormalized data for reporting

Connection Pooling:

{
  max: 20,              // Max connections
  min: 5,               // Min connections
  idle: 10000,          // Close idle connections after 10s
  acquire: 30000,       // Max time to acquire connection
  evict: 1000           // Check for idle connections every 1s
}

Backup Strategy:

• Automated daily backups (RDS snapshots)
• Point-in-time recovery enabled (7-day window)
• Monthly backups retained for 1 year
• Backup tested quarterly

Caching Strategy

Redis Configuration:

• Deployment: AWS ElastiCache (Redis 7.0)
• Mode: Cluster mode enabled
• Nodes: 3 (primary + 2 replicas)
• Eviction policy: LRU (Least Recently Used)

Cache Patterns:

1. Cache-Aside (Read-heavy data):

async function getProduct(id) {
  // Try cache first
  let product = await cache.get(`product:${id}`);

  if (!product) {
    // Cache miss - fetch from database
    product = await db.products.findById(id);

    // Store in cache (1 hour TTL)
    await cache.set(`product:${id}`, product, 3600);
  }

  return product;
}

2. Write-Through (Critical data):

async function updateProduct(id, data) {
  // Update database
  const product = await db.products.update(id, data);

  // Update cache
  await cache.set(`product:${id}`, product, 3600);

  return product;
}

Cache Invalidation:

// Product updated
await cache.del(`product:${productId}`);
await cache.del(`products:merchant:${merchantId}`);
await cache.del(`products:category:${categoryId}`);

// Pattern-based invalidation
await cache.delPattern(`products:*`);

What We Cache: | Data Type | TTL | Rationale | |-----------|-----|-----------| | Product details | 1 hour | Infrequently updated | | Product lists | 5 minutes | Frequently updated | | User sessions | 15 minutes | Security requirement | | Search results | 10 minutes | Expensive queries | | API responses | 1 minute | Rate limit protection |

Data Migration Strategy

Tools: Prisma Migrate (development), custom scripts (production)

Migration Process:

1. Create migration in development
2. Review SQL in PR
3. Test on staging (copy of production data)
4. Run on production during low-traffic window
5. Rollback plan documented

Zero-Downtime Migrations:

-- Example: Adding non-null column

-- Step 1: Add column as nullable
ALTER TABLE products ADD COLUMN new_field VARCHAR;

-- Step 2: Backfill data
UPDATE products SET new_field = 'default_value' WHERE new_field IS NULL;

-- Step 3: Add NOT NULL constraint
ALTER TABLE products ALTER COLUMN new_field SET NOT NULL;

---

Infrastructure

AWS Architecture

Regions: Primary: us-east-1, Disaster Recovery: us-west-2

VPC Design:

VPC (10.0.0.0/16)
├── Public Subnets (10.0.1.0/24, 10.0.2.0/24)
│   ├── NAT Gateways
│   └── Application Load Balancer
└── Private Subnets (10.0.10.0/24, 10.0.11.0/24)
    ├── ECS Tasks (Application)
    ├── RDS (Database)
    └── ElastiCache (Redis)

Compute:

• ECS Fargate: Serverless containers for application
• Auto-scaling: Target CPU 70%, min 2 tasks, max 10 tasks
• Task Definition:

  CPU: 1024 (1 vCPU)
  Memory: 2048 MB
  Container Port: 3000
  Environment: Production
  

Database:

• RDS PostgreSQL: db.r5.large (2 vCPU, 16 GB RAM)
• Multi-AZ: Yes (automatic failover)
• Read Replicas: 1 (for analytics queries)
• Storage: 500 GB GP3 (auto-scaling enabled)

Storage:

• S3 Bucket: product-images-prod
• Lifecycle Policy: Move to Glacier after 90 days
• CDN: CloudFront distribution for images
• Backup: Cross-region replication enabled

Networking:

• Load Balancer: Application Load Balancer (ALB)
• SSL/TLS: ACM certificates (auto-renewal)
• WAF: AWS WAF with OWASP rules
• DDoS Protection: AWS Shield Standard

Deployment Architecture

CI/CD Pipeline (GitHub Actions):

Code Push
  ↓
Automated Tests (Unit + Integration)
  ↓
Linting & Type Checking
  ↓
Build Docker Image
  ↓
Push to ECR (Elastic Container Registry)
  ↓
Deploy to Staging (Auto)
  ↓
Integration Tests (Staging)
  ↓
Manual Approval
  ↓
Deploy to Production (Canary)
  ↓
Monitor Metrics (15 minutes)
  ↓
Full Rollout or Rollback

Deployment Strategy: Blue-Green with Canary

Production (Blue)        Canary (Green)
100% traffic    →   95% / 5% split   →   0% / 100%
                    ↓
              Monitor for 15 min
                    ↓
              Success? Full rollout : Rollback

Rollback Procedure:

1. Detect issue (automated alerts or manual)
2. Trigger rollback command
3. Route traffic back to previous version
4. Investigate root cause
5. Fix and redeploy

Deployment Windows:

• Staging: Anytime
• Production: Tuesday-Thursday, 10 AM - 2 PM EST
• Emergency: 24/7 with on-call approval

---

Security Architecture

Defense in Depth

Layer 1: Network Security

• VPC isolation
• Security groups (allow-list only)
• Network ACLs
• Private subnets for data layer
• NAT Gateway for outbound traffic

Layer 2: Application Security

• Input validation (all user inputs)
• SQL injection prevention (parameterized queries)
• XSS prevention (sanitization + CSP headers)
• CSRF protection (tokens)
• Rate limiting (DDoS mitigation)

Layer 3: Authentication & Authorization

• JWT with short expiration
• Refresh token rotation
• MFA for admin accounts
• Role-based access control (RBAC)
• Principle of least privilege

Layer 4: Data Security

• Encryption at rest (RDS, S3)
• Encryption in transit (TLS 1.3)
• Secrets in AWS Secrets Manager
• PII data encrypted at field level
• Regular security audits

Security Headers

{
  'Strict-Transport-Security': 'max-age=31536000; includeSubDomains',
  'Content-Security-Policy': "default-src 'self'; script-src 'self' 'unsafe-inline'",
  'X-Frame-Options': 'DENY',
  'X-Content-Type-Options': 'nosniff',
  'Referrer-Policy': 'strict-origin-when-cross-origin',
  'Permissions-Policy': 'geolocation=(), microphone=(), camera=()'
}

Compliance

Standards:

• PCI DSS: Level 2 (Stripe handles Level 1)
• GDPR: User data rights, deletion, export
• SOC 2 Type II: In progress (Q2 2024)

Data Retention:

• User data: Retained until account deletion
• Order data: 7 years (regulatory requirement)
• Logs: 90 days
• Backups: 1 year

---

Scalability & Performance

Current Capacity

| Metric | Current | Limit | Headroom | |--------|---------|-------|----------| | Concurrent Users | 500 | 2,000 | 4x | | Requests/Second | 200 | 1,000 | 5x | | Database Connections | 50 | 200 | 4x | | Storage | 500 GB | 2 TB | 4x |

Scaling Strategy

Horizontal Scaling:

• Stateless services (easy to replicate)
• Auto-scaling based on CPU/memory
• Database read replicas for read-heavy workloads

Vertical Scaling:

• Database instance size (scheduled uptime)
• Cache cluster size

Caching:

• Application-level caching (Redis)
• CDN for static assets
• Database query result caching

Database Optimization:

• Indexes on frequently queried fields
• Materialized views for complex queries
• Connection pooling
• Query optimization (EXPLAIN ANALYZE)

Performance Budgets

API Response Times (p95):

• GET requests: <200ms
• POST requests: <500ms
• Complex queries: <1s

Frontend Performance (Lighthouse):

• Performance: >90
• Accessibility: 100
• Best Practices: >90
• SEO: 100

Database Query Times (p95):

• Simple queries: <50ms
• Join queries: <100ms
• Aggregations: <500ms

---

Monitoring & Observability

Metrics

Application Metrics (DataDog):

• Request rate, error rate, duration (RED metrics)
• Active users, sessions
• Business metrics (orders, revenue)
• Custom metrics (cart abandonment, conversion rate)

Infrastructure Metrics:

• CPU, memory, disk usage
• Network throughput
• Database connections, query performance
• Cache hit rate

Dashboards:

1. System Health: Overall system status
2. API Performance: Endpoint-specific metrics
3. Business Metrics: KPIs and conversions
4. Database Performance: Query analysis
5. Error Tracking: Error rates and trends

Logging

Log Levels:

• ERROR: Application errors requiring investigation
• WARN: Potential issues or degraded performance
• INFO: Significant events (order created, payment succeeded)
• DEBUG: Detailed diagnostic information (disabled in production)

Log Aggregation: CloudWatch Logs → DataDog

Structured Logging:

logger.info('Order created', {
  orderId: '123',
  customerId: '456',
  total: 99.99,
  timestamp: new Date().toISOString()
});

Alerting

Alert Channels:

• Critical: PagerDuty (SMS + Phone)
• High: Slack #incidents
• Medium: Slack #e

<!-- Content truncated for initial SEO render. Open the source file tab for the full file. -->