데이터 메시

# 데이터 프로덕트 정의 예시
# data-products/orders/dataproduct.yaml

apiVersion: datamesh.io/v1
kind: DataProduct
metadata:
  name: orders-data-product
  domain: order-management
  owner: order-team@company.com
  version: "2.1.0"

spec:
  description: |
    주문 도메인의 핵심 데이터 프로덕트.
    모든 주문 트랜잭션 및 집계 메트릭 제공.

  # 출력 포트 정의 (데이터 접근 방법)
  outputPorts:
    - name: orders-events
      type: streaming
      format: avro
      location: kafka://orders.events
      description: "실시간 주문 이벤트 스트림"
      sla:
        latency: "p99 < 100ms"
        availability: "99.9%"

    - name: orders-snapshot
      type: batch
      format: delta
      location: s3://data-products/orders/snapshot/
      description: "주문 스냅샷 테이블 (일별 업데이트)"
      schedule: "0 2 * * *"
      schema:
        fields:
          - name: order_id
            type: string
            description: "주문 고유 식별자"
            pii: false
          - name: customer_id
            type: string
            description: "고객 ID (암호화됨)"
            pii: true
          - name: total_amount
            type: decimal(18,2)
            description: "주문 총액"
          - name: status
            type: string
            enum: [pending, confirmed, shipped, delivered]
          - name: created_at
            type: timestamp

    - name: orders-metrics
      type: api
      format: rest
      location: https://api.internal/orders/metrics
      description: "주문 집계 메트릭 API"

  # 입력 소스 (의존성)
  inputPorts:
    - name: raw-orders
      type: database
      source: postgres://order-service-db
    - name: customer-data
      type: data_product
      source: customer-domain/customer-data-product

  # 데이터 품질 규칙
  quality:
    tests:
      - name: order_id_unique
        type: unique
        column: order_id
      - name: no_null_amounts
        type: not_null
        column: total_amount
      - name: valid_status
        type: accepted_values
        column: status
        values: [pending, confirmed, shipped, delivered]
      - name: freshness_check
        type: freshness
        column: created_at
        max_age: "24h"

  # 접근 제어 (연합 거버넌스)
  access:
    public: false
    classification: confidential
    teams:
      - name: analytics-team
        permissions: [read]
      - name: ml-team
        permissions: [read]
      - name: order-team
        permissions: [read, write, admin]

  # 문서 및 지원
  documentation:
    readme: docs/README.md
    wiki: https://wiki.company.com/orders-data-product
    slack: "#orders-data-support"

---
# 셀프 서비스 플랫폼: 인프라 자동 프로비저닝
# platform/terraform/data-product-infra.tf

resource "datamesh_data_product" "orders" {
  name   = "orders-data-product"
  domain = "order-management"
  owner  = "order-team@company.com"

  compute {
    type    = "spark"
    version = "3.5"
    size    = "medium"
    autoscaling {
      min_workers = 2
      max_workers = 10
    }
  }

  storage {
    type     = "delta"
    location = "s3://data-products/orders/"
    retention_days = 90
  }

  catalog {
    type      = "unity_catalog"
    namespace = "orders_domain"
    register_to_global = true
  }

  monitoring {
    enable_lineage = true
    alert_channels = ["slack:#orders-alerts"]
  }
}