System Design & Architecture

Dokumentasi lengkap arsitektur sistem MStore Backend dari high-level design hingga implementation details.

💡 Working with Diagrams:

Zoom: Cmd/Ctrl + Scroll atau klik kanan → “Open Image in New Tab”
Download: Copy code diagram → paste ke Mermaid Live Editor → Download PNG/SVG
Full Guide: Lihat Mermaid Guide untuk tutorial lengkap

🎯 Architecture Principles

MStore Backend dibangun dengan prinsip:

✅ Clean Architecture: Separation of concerns (handler → service → repository)
✅ Domain-Driven Design: Business logic organized by domain
✅ Multi-Tenant: Merchant-scoped data isolation
✅ Microservices-Ready: Modular design untuk future decomposition
✅ Event-Driven: Message broker integration (NATS, RabbitMQ, Kafka)
✅ API-First: RESTful API dengan OpenAPI/Swagger documentation
✅ Observability: LGTM stack (Loki, Grafana, Tempo, Mimir)

🏗️ High-Level Architecture

📦 Clean Architecture Layers

1. Handler Layer (Presentation)

Responsibility: HTTP request/response handling, validation, authentication

// internal/domains/transaction/transaction_handler.go
type TransactionHandler struct {
    service TransactionService
}

func (h *TransactionHandler) CreateTransaction(c *fiber.Ctx) error {
    // 1. Parse & validate request
    var req CreateTransactionRequest
    if err := c.BodyParser(&req); err != nil {
        return response.BadRequest(c, "invalid request")
    }
    
    // 2. Get user context
    userID := c.Locals("user_id").(string)
    
    // 3. Call service
    tx, err := h.service.CreateTransaction(c.Context(), req, userID)
    if err != nil {
        return response.InternalError(c, err.Error())
    }
    
    // 4. Return response
    return response.Success(c, tx)
}

2. Service Layer (Business Logic)

Responsibility: Business rules, orchestration, transaction management

// internal/domains/transaction/transaction_service.go
type TransactionService interface {
    CreateTransaction(ctx context.Context, req CreateTransactionRequest, userID string) (*Transaction, error)
    VerifyPayment(ctx context.Context, txID int64, req VerifyPaymentRequest) (*Transaction, error)
    SyncTransaction(ctx context.Context, txID int64) (*Transaction, error)
}

type transactionService struct {
    repo           TransactionRepository
    inventoryRepo  InventoryRepository
    accountingRepo AccountingRepository
    paymentClient  PaymentClient
}

func (s *transactionService) CreateTransaction(ctx context.Context, req CreateTransactionRequest, userID string) (*Transaction, error) {
    // 1. Validate business rules
    if err := s.validateTransaction(req); err != nil {
        return nil, err
    }
    
    // 2. Begin transaction
    return s.repo.Transaction(ctx, func(tx *gorm.DB) (*Transaction, error) {
        // 3. Create transaction
        transaction := &Transaction{
            BranchCode:    req.BranchCode,
            CashierID:     userID,
            Status:        "draft",
            Items:         req.Items,
            GrandTotal:    calculateTotal(req.Items),
        }
        
        if err := s.repo.Create(ctx, transaction); err != nil {
            return nil, err
        }
        
        // 4. Generate transaction code
        transaction.TransactionCode = s.generateTransactionCode(transaction)
        
        return transaction, nil
    })
}

3. Repository Layer (Data Access)

Responsibility: Database operations, query building

// internal/domains/transaction/transaction_repository.go
type TransactionRepository interface {
    Create(ctx context.Context, tx *Transaction) error
    GetByID(ctx context.Context, id int64) (*Transaction, error)
    Update(ctx context.Context, id int64, updates map[string]interface{}) error
    Transaction(ctx context.Context, fn func(*gorm.DB) error) error
}

type transactionRepository struct {
    db *gorm.DB
}

func (r *transactionRepository) Create(ctx context.Context, tx *Transaction) error {
    return r.db.WithContext(ctx).Create(tx).Error
}

func (r *transactionRepository) GetByID(ctx context.Context, id int64) (*Transaction, error) {
    var tx Transaction
    err := r.db.WithContext(ctx).
        Preload("Items").
        Preload("Items.Product").
        First(&tx, id).Error
    return &tx, err
}

🔀 Domain Organization

internal/domains/
├── auth/                   # Authentication & Authorization
│   ├── auth_handler.go
│   ├── auth_service.go
│   ├── auth_repository.go
│   ├── auth_payload.go
│   └── auth_response.go
├── transaction/            # POS Transactions
│   ├── transaction_handler.go
│   ├── transaction_service.go
│   ├── transaction_repository.go
│   ├── transaction_state.go      # State machine
│   └── transaction_journal_mapping.go
├── inventory/              # Inventory & Warehouse
│   ├── inventory_handler.go
│   ├── inventory_service.go
│   └── inventory_repository.go
├── payments/               # Payment Gateway Integration
│   ├── payment_service.go
│   ├── xendit/
│   │   ├── xendit_service.go
│   │   └── xendit_payload.go
│   └── midtrans/
├── pos/                    # POS Operations
│   ├── pos_handler.go
│   ├── pos_service.go
│   └── offline/            # Offline-first support
├── approvals/              # Multi-level Approval
│   ├── approval_handler.go
│   ├── approval_service.go
│   └── approval_repository.go
├── accounting/             # Journal & Financial
│   ├── accounting_service.go
│   └── journal_repository.go
├── product/                # Product & Bundle
├── promotions/             # Promotions & Price Lists
├── merchant/               # Merchant Management
├── branches/               # Branch Management
├── user/                   # User Management
└── index.go                # Domain registry

🔄 Request Flow

Typical API Request Flow

🗄️ Database Architecture

Multi-Database Strategy

MySQL (Primary - Transactional)

Transactions
Inventory & Warehouse
Approvals
Accounting & Journals
Master data (Products, Branches, Users)

MongoDB (Secondary - Flexible Schema)

POS offline transactions (temporary)
Audit logs
Analytics data
Unstructured documents

Redis (Cache & Session)

Session storage
API rate limiting
Real-time data cache
Pub/Sub for real-time updates

Multi-Tenant Data Isolation

-- Every table has merchant_id for isolation
CREATE TABLE transactions (
    id BIGINT UNSIGNED PRIMARY KEY,
    merchant_id BIGINT UNSIGNED NOT NULL,
    branch_id BIGINT UNSIGNED,
    transaction_code VARCHAR(255),
    -- ... other columns
    
    INDEX idx_merchant (merchant_id),
    FOREIGN KEY fk_merchant (merchant_id) 
        REFERENCES merchants(id)
);

-- Application-level filtering
SELECT * FROM transactions 
WHERE merchant_id = ? 
  AND branch_id = ?
  AND deleted_at IS NULL;

🔐 Security Architecture

Authentication Flow

Authorization (RBAC)

// Middleware for role-based access
func RequireRole(roles ...string) fiber.Handler {
    return func(c *fiber.Ctx) error {
        userRole := c.Locals("user_role").(string)
        
        for _, role := range roles {
            if userRole == role {
                return c.Next()
            }
        }
        
        return response.Forbidden(c, "insufficient permissions")
    }
}

// Usage in routes
app.Post("/api/v1/transactions/:id/void", 
    middleware.Auth(),
    middleware.RequireRole("MANAGER", "ADMIN"),
    handler.VoidTransaction,
)

📡 Event-Driven Architecture

Message Broker Integration

// Publish event after transaction created
func (s *transactionService) CreateTransaction(ctx context.Context, req CreateTransactionRequest) (*Transaction, error) {
    // ... create transaction
    
    // Publish event
    event := TransactionCreatedEvent{
        TransactionID:   tx.ID,
        TransactionCode: tx.TransactionCode,
        MerchantID:      tx.MerchantID,
        BranchID:        tx.BranchID,
        Amount:          tx.GrandTotal,
        CreatedAt:       tx.CreatedAt,
    }
    
    if err := s.messageBroker.Publish("transaction.created", event); err != nil {
        // Log error but don't fail transaction
        log.Printf("failed to publish event: %v", err)
    }
    
    return tx, nil
}

// Subscribe to events
func (s *inventoryService) SubscribeToTransactionEvents() {
    s.messageBroker.Subscribe("transaction.created", func(msg Message) {
        var event TransactionCreatedEvent
        json.Unmarshal(msg.Data, &event)
        
        // Process inventory movement
        s.ProcessTransactionInventory(context.Background(), event)
    })
}

🔍 Observability Architecture

LGTM Stack Integration

Loki (Logs)

// Structured logging with Zap
logger.Info("transaction created",
    zap.String("transaction_code", code),
    zap.Int64("merchant_id", merchantID),
    zap.Float64("amount", amount),
)

Tempo (Traces)

// OpenTelemetry tracing
ctx, span := tracer.Start(ctx, "CreateTransaction")
defer span.End()

span.SetAttributes(
    attribute.String("transaction.code", code),
    attribute.Int64("transaction.amount", amount),
)

Mimir (Metrics)

// Prometheus metrics
counter, _ := meter.Int64Counter("transactions.created")
counter.Add(ctx, 1, 
    metric.WithAttributes(
        attribute.String("branch", branchCode),
        attribute.String("payment_method", "CASH"),
    ),
)

Grafana (Dashboards)

Transaction volume & revenue
API latency & error rate
Database query performance
Cache hit rate
Message broker throughput

🚀 Deployment Architecture

Development Environment

# docker-compose.yml
version: '3.8'

services:
  app:
    build: .
    ports:
      - "8080:8080"
    environment:
      - ENV=development
      - DB_HOST=mysql
      - REDIS_HOST=redis
    depends_on:
      - mysql
      - redis
      - nats
  
  mysql:
    image: mysql:8.0
    environment:
      MYSQL_ROOT_PASSWORD: root
      MYSQL_DATABASE: mstore
    volumes:
      - mysql_data:/var/lib/mysql
  
  redis:
    image: redis:7-alpine
    ports:
      - "6379:6379"
  
  nats:
    image: nats:latest
    ports:
      - "4222:4222"

Production Architecture

📊 Performance Characteristics

Target Metrics

Metric	Target	Notes
API Latency (P95)	< 100ms	Excluding external API calls
API Latency (P99)	< 500ms	Including database queries
Throughput	10k req/s	Per instance
Database Query	< 50ms	P95 for simple queries
Cache Hit Rate	> 80%	Redis cache
Availability	99.9%	~8.7 hours downtime/year
Error Rate	< 0.1%	4xx + 5xx errors

Scalability Strategy

Horizontal Scaling

Stateless application servers
Load balancer distribution
Database read replicas
Redis cluster for cache

Vertical Scaling

Database server resources
Cache memory allocation
Message broker capacity

Caching Strategy

L1: In-memory cache (per instance)
L2: Redis cluster (shared)
Cache invalidation via events

💡 Best Practices

DO ✅

Follow clean architecture layers
Use dependency injection
Implement proper error handling
Log with structured logging (Zap)
Use context for cancellation & timeout
Implement circuit breaker for external APIs
Use database transactions for atomic operations
Implement idempotency for critical operations
Monitor with OpenTelemetry
Document APIs with Swagger

DON’T ❌

Jangan skip input validation
Jangan hardcode credentials
Jangan ignore context cancellation
Jangan block goroutines indefinitely
Jangan skip error logging
Jangan expose internal errors ke client
Jangan skip database indexes
Jangan ignore connection pooling

🔄 Migration Strategy

Microservices Decomposition (Future)

Decomposition Strategy:

Start with bounded contexts (domains)
Extract stateless services first
Implement API gateway
Use event bus for inter-service communication
Gradual migration per domain

Tech Stack

Library dan tools yang digunakan

Service Template

Template untuk membuat service baru

API Styleguide

Panduan API design & conventions

Database Schema

ERD lengkap database

Observability

LGTM stack setup & monitoring

Need Help? Contact backend team atau check GitHub Issues

Overview

Components

System Design & Architecture

System Design & Architecture

🎯 Architecture Principles

🏗️ High-Level Architecture

📦 Clean Architecture Layers

1. Handler Layer (Presentation)

2. Service Layer (Business Logic)

3. Repository Layer (Data Access)

🔀 Domain Organization

🔄 Request Flow

Typical API Request Flow

🗄️ Database Architecture

Multi-Database Strategy

Multi-Tenant Data Isolation

🔐 Security Architecture

Authentication Flow

Authorization (RBAC)

📡 Event-Driven Architecture

Message Broker Integration

🔍 Observability Architecture

LGTM Stack Integration

🚀 Deployment Architecture

Development Environment

Production Architecture

📊 Performance Characteristics

Target Metrics

Scalability Strategy

💡 Best Practices

DO ✅

DON’T ❌

🔄 Migration Strategy

Microservices Decomposition (Future)

Tech Stack

Service Template

API Styleguide

Database Schema

Observability

Overview

Components

​System Design & Architecture

​🎯 Architecture Principles

​🏗️ High-Level Architecture

​📦 Clean Architecture Layers

​1. Handler Layer (Presentation)

​2. Service Layer (Business Logic)

​3. Repository Layer (Data Access)

​🔀 Domain Organization

​🔄 Request Flow

​Typical API Request Flow

​🗄️ Database Architecture

​Multi-Database Strategy

​Multi-Tenant Data Isolation

​🔐 Security Architecture

​Authentication Flow

​Authorization (RBAC)

​📡 Event-Driven Architecture

​Message Broker Integration

​🔍 Observability Architecture

​LGTM Stack Integration

​🚀 Deployment Architecture

​Development Environment

​Production Architecture

​📊 Performance Characteristics

​Target Metrics

​Scalability Strategy

​💡 Best Practices

​DO ✅

​DON’T ❌

​🔄 Migration Strategy

​Microservices Decomposition (Future)

​📚 Related Documentation

Tech Stack

Service Template

API Styleguide

Database Schema

Observability

System Design & Architecture

🎯 Architecture Principles

🏗️ High-Level Architecture

📦 Clean Architecture Layers

1. Handler Layer (Presentation)

2. Service Layer (Business Logic)

3. Repository Layer (Data Access)

🔀 Domain Organization

🔄 Request Flow

Typical API Request Flow

🗄️ Database Architecture

Multi-Database Strategy

Multi-Tenant Data Isolation

🔐 Security Architecture

Authentication Flow

Authorization (RBAC)

📡 Event-Driven Architecture

Message Broker Integration

🔍 Observability Architecture

LGTM Stack Integration

🚀 Deployment Architecture

Development Environment

Production Architecture

📊 Performance Characteristics

Target Metrics

Scalability Strategy

💡 Best Practices

DO ✅

DON’T ❌

🔄 Migration Strategy

Microservices Decomposition (Future)

📚 Related Documentation