Domain 1: Base Database Service - VM (BaseDB) (20%)
Domain 1 of the 1Z0-1093-25 Oracle AI Cloud Database Services 2025 Professional exam covers the Oracle Base Database Service running on Virtual Machine (VM) DB systems. This domain represents 20% of the exam — approximately 10 questions out of 50 (90 minutes, 68% passing score). BaseDB is the foundational OCI database service and the exam tests detailed knowledge of provisioning, lifecycle management, backup/recovery, patching, Data Guard, PDB management, and networking.
The exam syllabus defines six objective areas for this domain:
- Service description and capabilities
- Deployment and configuration procedures
- Database lifecycle management operations
- Backup and recovery strategies
- User-managed component patching and upgrades
- Monitoring tools and administrative interfaces
1. Service Overview and Architecture
What Base Database Service Is
Oracle Base Database Service (BaseDB) is an OCI-managed database service that runs Oracle Database on virtual machines. You get full administrative access (root and oracle user SSH access) to the VM, Oracle Database, and Grid Infrastructure. Oracle manages the underlying infrastructure (hypervisor, networking, physical hardware), while you manage the database software, patching decisions, and data. (BaseDB FAQ)
BaseDB provisions a DB system — one or two VMs with attached storage, Grid Infrastructure, and an Oracle Database instance. Each DB system contains:
- One or two VM nodes (single-node or 2-node RAC)
- One Database Home per DB system
- One Container Database (CDB) per Database Home
- One or more Pluggable Databases (PDBs) within the CDB
Exam trap: BaseDB supports exactly one CDB per VM DB system. You cannot create multiple CDBs. You can have multiple PDBs within that single CDB.
Supported Database Versions and Editions
| Edition | Single-Node | 2-Node RAC | Notes |
|---|---|---|---|
| Standard Edition | Yes | No | No RAC support |
| Enterprise Edition (EE) | Yes | Yes (EEP required) | Base enterprise features |
| Enterprise Edition - High Performance (EE-HP) | Yes | Yes (EEP required) | Adds partitioning, OLAP, Advanced Security, etc. |
| Enterprise Edition - Extreme Performance (EE-EP) | Yes | Yes | Required for RAC. Includes Active Data Guard. |
| Enterprise Edition - Developer | Yes | No | Restricted: Ampere A1 only, no Data Guard, no cross-region backup |
Supported Oracle Database versions: (BaseDB About)
- Oracle AI Database 26ai
- Oracle Database 21c
- Oracle Database 19c
Exam trap: You cannot change the database edition after provisioning. If you provision Standard Edition, you cannot upgrade to Enterprise Edition in place. You must create a new DB system.
Exam trap: Enterprise Edition - Extreme Performance is required for 2-node RAC. Not Enterprise Edition, not EE-HP — only EE-EP. This is a frequently tested point.
Licensing Models
Two licensing models are available and can be changed after provisioning: (BaseDB FAQ)
| Model | Description |
|---|---|
| License Included | Oracle Database license is included in the hourly/monthly price |
| Bring Your Own License (BYOL) | Use existing on-premises Oracle Database licenses; lower cloud price |
Billing is per-second with a one-minute minimum. You can stop a VM DB system to stop compute billing while retaining the database and storage.
2. VM Shapes and Compute
Flexible Shapes
BaseDB uses OCI Compute shapes. The current generation shapes are all flexible — you choose the number of OCPUs (or ECPUs) within the shape's range. (BaseDB VM DB Systems)
| Shape | Processor | OCPU Range | Memory per OCPU | Max Network BW | Max Theoretical IOPS | DB Versions |
|---|---|---|---|---|---|---|
| VM.Standard.x86 | Intel | 4-256 ECPUs (increments of 4) | 8 GB per 4 ECPUs | — | — | 26ai only |
| VM.Standard.E5.Flex | AMD EPYC (4th Gen) | 1-64 OCPUs | 16 GB | 40 Gbps | 640K | 26ai, 19c |
| VM.Standard.E4.Flex | AMD EPYC (3rd Gen) | 1-64 OCPUs | 16 GB | 40 Gbps | 640K | 26ai, 21c, 19c |
| VM.Standard3.Flex | Intel Xeon (X9) | 1-32 OCPUs | 16 GB | 32 Gbps | 512K | 26ai, 21c, 19c |
| VM.Standard.A1.Flex | Ampere (ARM) | 1-57 OCPUs | 8 GB | — | — | 26ai, 19c |
| VM.Standard2.x (X7) | Intel (legacy) | Fixed: 1/2/4/8/16/24 | ~15 GB | — | — | 26ai, 21c, 19c |
Exam trap: The Ampere A1 shape has only 8 GB per OCPU (not 16 GB like AMD/Intel shapes). It also uses LVM only (no ASM), supports single-node only, and cannot migrate to or from Intel/AMD shapes. Data Guard between Ampere and Intel/AMD is not supported.
Exam trap: The new VM.Standard.x86 shape uses ECPUs (not OCPUs), requires Oracle AI Database 26ai only, supports single-node only with LVM, and has no fault domain support.
Exam trap: For 2-node RAC, the minimum is 2 OCPUs per node across all shapes.
Legacy Standard2 Shapes
The VM.Standard2 series are fixed shapes (not flexible). VM.Standard2.1 with 1 OCPU supports single-node only — it cannot run RAC.
Shape Migration Paths
Not all shape transitions are supported. Key migration rules: (Shape Change)
| From | To | Supported? |
|---|---|---|
| Intel X7 (Standard2) | AMD E4 | Yes |
| Intel X7 (Standard2) | AMD E5 | Yes |
| AMD E4 | AMD E5 | Yes |
| Intel X9 (Standard3) | AMD E4/E5 | No |
| AMD E4/E5 | Intel X9 | No |
| Ampere A1 | Any Intel/AMD | No |
| Any Intel/AMD | Ampere A1 | No |
| AMD E5 | Any other shape | No (currently one-way) |
Exam trap: Shape migration from AMD E5 to any other shape is not supported. Migration to AMD E5 is one-way. Also, you cannot roll back from AMD to Intel after migration.
Shape Change Behavior
- Single-node: Shape change requires a restart (downtime).
- 2-node RAC: Shape change occurs in a rolling fashion with zero downtime.
- Shape change does not affect storage allocation.
- Prerequisites: DB system must be Available, database must use SPFILE (not PFILE),
SGA_TARGETmust be nonzero, Cluster Ready Services must be running.
3. Storage Configuration
Storage Types
| Storage Type | Use Case | Available With |
|---|---|---|
| Block Volume (Balanced) | Standard workloads; 10 IOPS per GB | All flexible shapes |
| Block Volume (Higher Performance) | Performance-sensitive workloads; 20 IOPS per GB | All flexible shapes |
Storage range: 256 GB to 80 TB data storage, plus up to 20 TB recovery storage. Total attachment up to 100 TB. (BaseDB FAQ)
Storage Management Software
| Feature | ASM | LVM (Fast Provisioning) |
|---|---|---|
| Single-node | Yes | Yes |
| 2-node RAC | Yes | No |
| VM.Standard.x86 shape | No | Yes (only) |
| Ampere A1 shape | No | Yes (only) |
| Database Software Images | Yes | No |
Exam trap: LVM (fast provisioning) is available for single-node systems only. RAC always requires ASM. You cannot change the storage management software after provisioning.
Storage Scaling
| Operation | Downtime? | Notes |
|---|---|---|
| Scale storage up | No downtime | Can increase data and recovery storage independently |
| Scale storage down | Requires migration | Must create new DB system and migrate data |
Total storage = Available data storage + Recovery area storage + ~200 GB reserved for DB software.
Oracle recommends keeping recovery storage at 20% of total storage or higher.
Exam trap: Storage can be scaled up online without downtime. Storage cannot be scaled down in place — you must migrate to a new DB system. This is a key difference from Exadata, which supports elastic storage scaling in both directions.
Storage Performance (Higher Performance Option)
With higher performance storage on AMD E4/E5 shapes:
- Single-node: up to 640K IOPS and 5,120 MB/s throughput
- 2-node RAC: up to 1,000K IOPS with doubled throughput
4. Database Lifecycle Management
Core Operations
| Operation | Description | Downtime? |
|---|---|---|
| Start | Brings a stopped DB system online | N/A (was already stopped) |
| Stop | Shuts down the DB system; stops compute billing | Yes |
| Reboot | Restarts the DB system | Yes |
| Terminate | Permanently deletes the DB system | Irreversible |
| Clone | Creates a duplicate DB system | Source remains available |
| Scale Storage | Increases data or recovery storage | No |
| Change Shape | Modifies OCPU count | Single-node: yes; RAC: rolling (no downtime) |
Exam trap: When you stop a VM DB system, compute billing stops but storage billing continues. The database and all data remain intact.
Exam trap: Terminate is permanent and irreversible. You are prompted to choose what happens to backups (retain or delete). Always back up before terminating.
Management Interfaces
BaseDB can be managed through five interfaces: (BaseDB FAQ)
| Interface | Use Case |
|---|---|
| OCI Console | Web-based GUI for all operations |
| OCI CLI | Command-line automation |
| OCI REST API / SDKs | Programmatic access (Python, Java, Go, etc.) |
| Terraform (OCI Provider) | Infrastructure as Code |
| DBCLI | On-host CLI for database-specific operations (patching, backup config) |
| Enterprise Manager | Advanced monitoring and management |
| OCI Database Management | Cloud-native monitoring service |
5. Backup and Recovery
Backup Destinations
Three backup destinations are available: (BaseDB Backup Overview)
| Destination | Durability | Availability | Recovery Speed | Notes |
|---|---|---|---|---|
| Autonomous Recovery Service | High | High | High | Recommended. Based on ZDLRA technology. Real-time protection. Same cost as Object Storage. |
| OCI Object Storage | High | High | Medium | Secure, scalable. Requires Swift endpoint connectivity (service gateway recommended). |
| Local Storage (FRA) | Low | Medium | High | Fast recovery but if DB system fails, backup is also lost. |
Exam trap: Automatic backups are not enabled by default. You must explicitly enable them during or after provisioning.
Exam trap: Databases in security zone compartments must have automatic backups enabled. This is enforced by policy.
Retention Periods
| Destination | Available Retention Periods | Default |
|---|---|---|
| Autonomous Recovery Service | Bronze (14d), Silver (35d), Gold (65d), Platinum (95d), Custom (14-95d) | Silver (35 days) |
| Object Storage | 7, 15, 30, 45, 60 days | 30 days |
Automatic Backup Schedule (Object Storage)
| Backup Type | Frequency | Notes |
|---|---|---|
| Level 0 (full) | Weekly (on specified weekend day) | Displayed as "incremental" type in Console |
| Level 1 (incremental) | Daily for 6 days after Level 0 | |
| Archived redo logs | Minimum every 60 minutes | No OCID assigned |
Backup window: Default is 00:00 - 06:00 (6 hours) in the DB system's region time zone. You can choose from 12 custom 2-hour windows on even-numbered hours. Backup jobs are not guaranteed to complete within the scheduling window.
On-Demand and Standalone Backups
- On-demand full backup: Created manually at any time (unless database is a standby in Data Guard).
- Standalone backup: Remains in Object Storage even after DB system termination. Can be used to create a new DB system.
Long-Term Retention (LTR) Backups
- Duration: 90 to 3,650 days (or 1-10 years)
- Storage: Recovery Service only
- In-place restore: Not supported (must restore to new DB system)
- Immutable once created
Recovery Options
| Method | Description |
|---|---|
| Restore to Latest | Recovers to last known good state; minimal data loss |
| Restore to Timestamp | Point-in-time recovery to a specific time |
| Restore to SCN | Recovers to a specific System Change Number |
Cross-region backup and restore is supported for both Object Storage and Autonomous Recovery Service destinations.
Exam trap: You can create a new DB system from a backup (out-of-place restore). You can also do in-place restore of an existing database. These are distinct operations.
Exam trap: Backups are encrypted with the same TDE master key used for the database. All BaseDB databases are provisioned with TDE enabled by default.
Backup with Data Guard
- Primary and standby must use the same backup destination when both are configured.
- Can configure automatic backups on primary only, standby only, or both.
- After switchover: if auto-backup was on primary, it continues on the new standby.
- After failover: if auto-backup was on standby, it continues on the new primary.
Key Backup Restrictions
- PDB-level backup is not supported. Backups occur at the CDB level and include all PDBs.
- Block storage volumes cannot be attached for network backups.
- Cannot create backups while patching, Data Guard operations, or other availability-impacting operations are running.
- The
/u01directory must have sufficient free space. - RMAN backup settings must not be manually modified (use Console/API or DBCLI).
6. Patching and Updating
Types of Updates
BaseDB has three categories of updates: (BaseDB Update DB System)
| Update Type | Scope | Versions Available |
|---|---|---|
| Grid Infrastructure (GI) update | Patches the DB system infrastructure layer | 2 most recent (N, N-1) |
| Database update | Patches the Oracle Database software | 4 most recent (N through N-3) |
| OS update | Updates the operating system | Via direct host access (not Console) |
Exam trap: OS patching is not supported through the OCI Console or APIs. You must SSH to the host and patch the OS manually. This is unique to BaseDB — Exadata handles OS patching automatically.
Exam trap: Always update the DB system (GI) before updating the database within it. Oracle explicitly recommends this sequence.
Update Procedure
- Precheck: Run a precheck to validate prerequisites before applying the update. Oracle recommends this step.
- Apply: Apply the update through the Console, CLI, or API.
- Verify: Check the Update History tab for status.
Prerequisites for successful updates:
/u01must have minimum 15 GB free space- Oracle Clusterware must be running
- All DB system nodes must be running
- DB system needs Object Storage connectivity (service gateway recommended)
Interim Updates (One-Off Patches)
- Applied manually using OPatch tool via SSH
- Automatically rolled back when applying new quarterly updates (since April 2022 for 12.1/12.2, July 2022 for 19c)
- Oracle recommends using custom database software images instead of applying one-off patches directly
OJVM Updates
- Must be applied manually using the OPatch tool
- Cannot be applied through the standard Console/API update procedures
Exam trap: Updates applied via DBCLI or command-line tools do not appear in the Console's Update History. Only Console-applied updates are shown.
Database Version Upgrades
Distinct from patching. You can upgrade the Oracle Database version (e.g., 19c to 26ai) as a separate operation. The database edition cannot be changed during an upgrade.
7. Oracle Data Guard
Architecture and Requirements
Data Guard for BaseDB creates a physical standby database as a transactionally consistent copy of the primary. (BaseDB Data Guard)
Key constraints:
- Maximum one standby per primary database
- Both databases must have identical versions and editions
- For 26ai: standby can be a higher minor version than primary
- Both DB systems must be in the same compartment
- Same region: must use the same VCN
- Cross-region: VCNs must be peered; databases must be in the same realm
Edition Requirements for Data Guard
| Edition | Data Guard | Active Data Guard |
|---|---|---|
| Standard Edition | Not supported | Not supported |
| Enterprise Edition | Yes | No (disabled by default) |
| Enterprise Edition - High Performance | Yes | No (disabled by default) |
| Enterprise Edition - Extreme Performance | Yes | Yes (ADG included) |
| Enterprise Edition - Developer | Not supported | Not supported |
Exam trap: Active Data Guard (read-only standby) requires Enterprise Edition - Extreme Performance. EE and EE-HP support Data Guard but not Active Data Guard.
Exam trap: Standard Edition and Developer Edition do not support Data Guard at all.
Protection Modes and Transport Types
| Protection Mode | Transport Type | Description |
|---|---|---|
| Maximum Performance | ASYNC | Default. No primary performance impact. Minimal data loss possible. |
| Maximum Availability | SYNC | Zero data loss under normal operation. Falls back to async if standby unavailable. |
Exam trap: BaseDB Data Guard supports Maximum Performance (ASYNC) and Maximum Availability (SYNC) only. Maximum Protection mode is not listed as a supported option in the OCI Console. Fast-Start Failover (FSFO) is also not supported through the OCI feature — manual configuration on the host is required.
Networking Requirements
- TCP port 1521 must be open between primary and standby subnets
- Security list rules must be stateful (default)
- If cross-region, VCN peering must be configured
- OCI Vault for TDE keys is required for cross-region Data Guard
Operations
| Operation | Description |
|---|---|
| Switchover | Planned role reversal. Primary becomes standby; standby becomes primary. No data loss. |
| Failover | Unplanned. Standby becomes primary. Possible data loss (depending on protection mode). |
| Reinstate | Re-enables the old primary as a standby after failover. |
Fault Domain Placement
- Oracle recommends placing standby in a different availability domain than primary.
- If same AD: place in a different fault domain.
- 2-node RAC constraint: with 3 fault domains per AD and 4 total nodes (2 primary + 2 standby), only one standby node can be in a non-overlapping fault domain.
Known Cross-Region Limitations
After a cross-region switchover:
- Cannot rotate Database KMS key on the new primary
- Cannot create PDB on the new primary
8. Pluggable Database (PDB) Management
Supported Operations
PDB management is performed through the OCI Console, CLI, or API: (BaseDB PDB Management)
| Operation | Description | Key Details |
|---|---|---|
| Create | Add a new PDB to the CDB | Created one at a time; no effect on existing PDBs |
| Start | Opens PDB in read-write mode | |
| Stop | Closes PDB (mount mode) | |
| Delete | Removes PDB | |
| Local Clone | Copy within the same CDB | |
| Remote Clone | Copy to a different CDB | Same AD, same or later DB version, across compartments/DB systems/VCNs |
| Refreshable Clone | Remote clone that can be periodically refreshed from source | Source cannot be deleted until clone is disconnected |
| Relocate | Move PDB from one CDB to another | Same AD, source PDB is removed |
| In-place Restore | Restore PDB within same CDB | To last good state or specific timestamp; one PDB at a time |
| Convert | Convert refreshable clone to regular PDB | Disconnects from source permanently |
PDB Open Modes
- Read-Write: Normal operation
- Read-Only: Read queries only
- Mounted: PDB is closed but metadata is accessible
Key PDB Facts for the Exam
- PDB names: max 30 alphanumeric characters, must start with a letter
- PDBs created via SQL (not Console/API) are not immediately visible in the Console. OCI syncs every 6 hours.
- Backup is at the CDB level — individual PDB backups are not supported.
- Oracle recommends taking a backup immediately after creating or cloning a PDB, because the PDB is not recoverable until the next daily auto-backup completes.
- Refreshable clone can only be refreshed while in mount mode.
- When cloning from 19c to 26ai, the cloned PDB is automatically upgraded to 26ai.
- Refreshable clones on standby databases in Data Guard are not supported — must create on primary.
Exam trap: If you create a PDB using SQL*Plus directly on the host (not through OCI Console), it will not appear in the Console for up to 6 hours. Always use the OCI Console or API for PDB management when possible.
9. Networking Requirements
VCN and Subnet Configuration
Every BaseDB DB system must be placed in a VCN subnet. Requirements:
| Component | Requirement |
|---|---|
| VCN | Must exist before DB system creation |
| Subnet | DB system is placed in a subnet; can be public or private |
| Service Gateway | Recommended for backups, patching, and managed updates (access to Object Storage and Identity endpoints) |
| Security Lists / NSGs | Control inbound/outbound traffic to DB system |
| DNS | Hostname and domain are assigned during provisioning |
Security Rules for 2-Node RAC
For multi-node RAC DB systems, the following rules must be configured:
- Port 22 (SSH) must be open for both ingress and egress between nodes
- Security rules must be stateful (the default)
Security Rules for Data Guard
- Port 1521 (SQLNet) must be open between primary and standby subnets for both ingress and egress
Private Subnet Considerations
For DB systems in private subnets with only a service gateway:
- Service gateway must allow access to all Oracle Services (not just Object Storage) to enable patching and updates
- If Identity and Object Storage endpoints are accessible by other means, this is not required
Exam trap: A common exam scenario tests whether you know that a service gateway is required for DB systems in private subnets to perform backups to Object Storage and to receive managed updates. Without it, these operations fail.
Custom IP Addresses
- Available for single-node and cloned systems only
- Not available for multi-node RAC DB systems
10. BaseDB vs. Exadata Database Service — Key Differences
This comparison is critical for the exam, since both Domain 1 (BaseDB) and Domain 2 (ExaDB) are tested.
| Feature | Base Database Service (BaseDB) | Exadata Database Service (ExaDB) |
|---|---|---|
| Infrastructure | Standard OCI VMs with block storage | Dedicated Exadata hardware (compute + storage servers) |
| CDBs per system | 1 CDB per DB system | Multiple CDBs per VM cluster |
| Max nodes | 1 or 2 (single or RAC) | Up to 32 compute nodes |
| Storage type | Block Volume (balanced or higher performance) | Exadata Smart Storage (local high-performance NVMe + PMEM) |
| Storage scaling | Up only (down requires migration) | Elastic scaling up and down |
| Compute scaling | Shape change (restart for single-node; rolling for RAC) | Online OCPU scaling without restart |
| OS patching | Manual (SSH to host) | Managed through Console/API |
| Automatic maintenance | DB system and database updates via Console | Infrastructure maintenance windows, automated patching |
| Performance | Up to 640K IOPS (single-node, higher perf storage) | Millions of IOPS, Exadata Smart Scan, storage indexes |
| Data Guard | 1 standby per primary | 1 standby per primary (same constraint) |
| Cost model | Per OCPU/hour + storage | Quarter-rack minimum or Exascale (pay-per-use ECPU) |
| Best for | Dev/test, small-medium production, cost-sensitive workloads | Mission-critical, high-performance, consolidation workloads |
Exam trap: BaseDB storage can only scale up. Exadata storage can scale up and down elastically. If a question asks about reducing storage, the answer for BaseDB is always "migrate to a new DB system."
Exam trap: BaseDB supports exactly one CDB per DB system. Exadata supports multiple CDBs per VM cluster. This is a fundamental architectural difference.
Quick-Reference: Common Exam Traps Summary
| Topic | Trap |
|---|---|
| RAC edition | 2-node RAC requires Enterprise Edition - Extreme Performance |
| Active Data Guard | Requires EE-EP; EE and EE-HP do not include it |
| Edition change | Cannot change edition after provisioning |
| Storage scaling | Up = online, no downtime. Down = not supported in place. |
| Compute scaling | Single-node = downtime. RAC = rolling, no downtime. |
| Auto backups | Not enabled by default |
| Backup retention (ObjStore) | Default 30 days; options: 7, 15, 30, 45, 60 |
| Backup retention (Recovery Svc) | Default Silver (35 days); Bronze 14, Gold 65, Platinum 95 |
| OS patching | Not via Console/API — manual SSH only |
| Update sequence | GI (DB system) first, then database |
| PDB via SQL | Not visible in Console for up to 6 hours |
| LVM | Single-node only; no RAC |
| Ampere A1 | 8 GB/OCPU (not 16); LVM only; no cross-shape migration |
| Service gateway | Required for private subnet backups and updates |
| Shape migration | AMD E5 is one-way; no rollback to Intel |
| Data Guard editions | Standard Edition and Developer Edition: no Data Guard |
| Backup level | CDB-level only; no individual PDB backup |
| FSFO | Not supported through OCI feature (manual only) |
References
- Oracle Base Database Service Documentation
- BaseDB VM DB Systems
- BaseDB FAQ
- Backup and Recovery Overview
- Update a DB System
- Update a Database
- Change the Shape of a DB System
- Scale the DB System
- About Pluggable Databases
- Oracle Data Guard on Base Database Service
- 1Z0-1093-25 Exam Page
- 1Z0-1093-25 Exam Syllabus