---
last_verified: 2026-03-14
service: Autonomous Database Dedicated (ADB-D)
category: database
oci_color: "#AA643B"
---

what: Autonomous Database on dedicated Exadata infrastructure. Provides isolation,
  higher resource limits, and more control while retaining autonomous operations
  (auto-patching, auto-tuning, auto-scaling).

when_to_use:
  - Regulatory/compliance requirement for dedicated infrastructure
  - Need for container database (CDB) level control
  - Sustained workloads >128 OCPUs per database
  - Multiple databases that benefit from shared Exadata infrastructure
  - Organizations that want autonomous features but need infrastructure isolation
  - Need to control maintenance windows precisely

when_NOT_to_use:
  - Small/medium workloads (ADB-S is simpler and cheaper)
  - Teams that don't need infrastructure isolation
  - Budget-constrained projects (dedicated infrastructure has higher base cost)
  - Rapid prototyping or dev/test (ADB-S provisions faster)

limits:
  max_ocpu_per_vm_cluster: 400
  max_adb_per_vm_cluster: 200
  max_storage_per_vm_cluster_tb: 192
  min_exadata_infrastructure: "Quarter rack (base)"
  max_vm_clusters_per_infra: 8

features:
  auto_scaling: true
  auto_indexing: true
  auto_partitioning: true
  autonomous_data_guard: true
  cross_region_adg: true
  private_endpoint: true
  customer_managed_keys: true
  fleet_management: true
  maintenance_window_control: true
  resource_pool_management: true
  os_access: false
  rac: false  # Not traditional RAC, but Exadata-managed HA

ha_dr:
  local_ha: "Exadata-managed, automatic failover within infrastructure"
  autonomous_data_guard:
    same_region: "Synchronous, automatic failover"
    cross_region: "Asynchronous, manual or automatic failover"

sizing_rules:
  minimum_infrastructure: "Quarter rack equivalent"
  ocpu_allocation: "Shared across all ADBs in the VM cluster"
  storage_allocation: "Shared Exadata storage pool"

pricing:
  model: dedicated_infrastructure
  infrastructure_base: "Exadata infrastructure (fixed cost per shape)"
  ocpu_payg: "Per OCPU per hour on top of infrastructure"
  byol_discount: "~50% on OCPU costs"
  notes:
    - "Infrastructure cost is fixed regardless of utilization"
    - "OCPU billing is for databases provisioned on the infrastructure"
    - "Higher base cost but better economics at scale (>128 OCPUs sustained)"

gotchas:
  - id: ADB-D-001
    issue: "Provisioning infrastructure takes hours, not minutes"
    impact: "Cannot spin up quickly like ADB-S"
    recommendation: "Plan infrastructure provisioning well ahead of database needs"
  - id: ADB-D-002
    issue: "Minimum infrastructure commitment"
    impact: "Even at low utilization, you pay for the base infrastructure"
    recommendation: "Only choose ADB-D when sustained usage justifies the base cost"
  - id: ADB-D-003
    issue: "Maintenance windows affect all databases on the infrastructure"
    impact: "Need to coordinate maintenance across all tenants"
    recommendation: "Use maintenance window control to schedule during off-peak"