Data Mesh

Domain-Oriented Ownership

AWS Implementation

resource "aws_glue_catalog_database" "domain_data" {
  name = "customer-domain"
  catalog_id = aws_glue_catalog_table.customer_data.catalog_id
  
  tags = {
    Domain = "Customer"
    Owner  = "customer-team"
    DataProduct = "true"
  }
}

resource "aws_lake_formation_permissions" "domain_access" {
  principal = aws_iam_role.domain_team.arn
  permissions = ["CREATE_TABLE", "ALTER", "DROP"]
  
  database {
    name = aws_glue_catalog_database.domain_data.name
  }
}

Data Product Implementation

Azure Data Product

resource "azurerm_synapse_workspace" "data_product" {
  name                = "customer-insights"
  resource_group_name = azurerm_resource_group.data_mesh.name
  location            = azurerm_resource_group.data_mesh.location
  storage_data_lake_gen2_filesystem_id = azurerm_storage_data_lake_gen2_filesystem.data_mesh.id
  
  identity {
    type = "SystemAssigned"
  }
  
  tags = {
    DataProduct = "CustomerInsights"
    Domain      = "Customer"
    Owner       = "customer-analytics"
  }
}

Federation Layer

GCP Implementation

resource "google_bigquery_dataset" "federated_view" {
  dataset_id = "customer_federation"
  location   = "US"
  
  access {
    role          = "READER"
    user_by_email = "[e-mail]"
  }
  
  labels = {
    environment = "production"
    federation  = "true"
  }
}

resource "google_bigquery_routine" "data_contract" {
  dataset_id = google_bigquery_dataset.federated_view.dataset_id
  routine_id = "customer_contract"
  language   = "SQL"
  
  definition_body = <<-SQL
    CREATE OR REPLACE VIEW customer_federation.customer_360 AS
    SELECT 
      c.customer_id,
      c.profile,
      o.order_history,
      p.preferences
    FROM customer_domain.profiles c
    JOIN orders_domain.history o USING (customer_id)
    JOIN preferences_domain.settings p USING (customer_id)
  SQL
}

Best Practices

1. Domain Ownership

   • Clear domain boundaries

   • Autonomous teams

   • Independent deployment

   • Self-service capabilities

2. Data Products

   • Discoverable interfaces

   • Well-defined contracts

   • Quality guarantees

   • Version management

3. Governance

   • Federated compliance

   • Automated policies

   • Access controls

   • Audit trails

4. Infrastructure

   • Scalable storage

   • Query federation

   • Cross-domain access

   • Performance monitoring