Oracle Data Platform Foundations Associate cheat sheet

Oracle's data platform spans four layers: infrastructure (OCI regions, compute, storage), database services (Autonomous Database, Base Database Service, Exadata), data management and integration (GoldenGate, Data Integration, Streaming), and analytics plus AI (OCI Data Science, Analytics Cloud, MySQL HeatWave).

Autonomous Database is self-driving, self-securing, and self-repairing: it automates provisioning, tuning, patching, scaling, and backups; the customer still chooses workload type (Transaction Processing, Data Warehouse, JSON, APEX, or Lakehouse) at creation and cannot change it afterward.

Autonomous Database Serverless bills on ECPU or OCPU plus storage consumed, with auto-scaling optional; Dedicated deployment gives isolated Exadata infrastructure with a fixed minimum of ECPUs or OCPUs and is chosen when workload isolation or specific compliance controls are required.

Base Database Service is a VM or bare-metal DB system where the customer administers the database (patching windows, backups, Data Guard configuration); it suits workloads needing OS-level access, custom parameters, or database versions and options not exposed by Autonomous Database.

Exadata Database Service (dedicated and cloud@customer) provides Exadata smart scan, storage indexes, and RDMA over RoCE between database and storage servers; Exascale Infrastructure is a separate, elastically scalable Exadata architecture used for Exadata Database Service on Dedicated Infrastructure.

Oracle Database 23ai adds AI Vector Search with vector data types and vector indexes for similarity search, JSON Relational Duality views, and schema-level annotations, letting one database serve transactional, analytical, document, and vector workloads.

Multitenant architecture: a container database (CDB) hosts one or more pluggable databases (PDBs); PDBs are moved, cloned, and plugged in or out, and each PDB is a portable set of data files plus metadata.

Data Guard provides standby databases for disaster recovery (Maximum Protection, Maximum Availability, Maximum Performance protection modes); Active Data Guard opens the standby read-only for reporting and offloads queries from the primary.

GoldenGate performs real-time, log-based replication across heterogeneous sources and targets; it reads redo or transaction logs rather than issuing queries against source tables, so impact on the source is minimal.

Autonomous Database workload type is fixed at creation: choosing Transaction Processing versus Data Warehouse changes default indexing, parallelism, and optimization behavior; migrating between types requires creating a new database and moving data.

Scaling rule: Autonomous Database auto-scaling expands compute up to three times the base ECPU or OCPU count when demand spikes, and billing follows actual consumption; manual scaling changes the base and is not instantaneous for all deployment types.

Backup rule: Autonomous Database retains automatic backups for 60 days by default; you can create manual backups, restore to a point in time within the retention window, or clone from a backup.

Cost decision rule: use Autonomous Database Serverless for variable or unpredictable workloads, Dedicated when isolation or fixed capacity is mandated, and Base Database Service when you must control the operating system or database configuration.

OCI Object Storage is the standard landing zone for data lake files; Autonomous Database accesses it through credentials and external tables or the DBMS_CLOUD package, and can query files such as CSV, JSON, and Parquet directly.

Partitioning improves manageability and can improve performance: range, list, hash, interval, and composite strategies are available, and partition pruning lets the optimizer skip partitions that cannot satisfy the predicate.

Oracle Analytics Cloud consumes data from Autonomous Database, Object Storage, and external sources; it provides semantic models, dashboards, and augmented analytics, and is distinct from the database's own SQL analytics features.

MySQL HeatWave runs OLTP and analytics in one MySQL-compatible service; HeatWave accelerates analytical queries in memory without extracting data to a separate warehouse, and HeatWave Lakehouse queries data in Object Storage.

Identity rule: OCI IAM policies grant access to compartment-scoped resources using policy statements of the form allow group X to verb resource-type in compartment Y; database-internal users and roles are managed separately inside the database.