Oracle Data Platform Foundations
This domain tests how Oracle’s cloud data services fit together: the converged database, Autonomous Database workload choices, lakehouse access to object storage, integration services, and the security model that governs them. Expect scenario questions that ask you to choose the right service or architecture rather than recite product names.
Core rules and concepts
- Converged database: one engine handles relational, JSON, XML, spatial, graph, text, and in-database machine learning. The exam favors reducing separate silos over adding a new database per data type.
- Autonomous Database: self-driving, self-securing, self-repairing. Autonomous Transaction Processing suits OLTP; Autonomous Data Warehouse suits analytics; Autonomous JSON Database suits document-centric workloads. Automated patching, scaling, and tuning are central.
- Lakehouse pattern: keep data in Object Storage and query it through external tables or Oracle’s lakehouse features instead of forcing a full load first. This supports open formats and avoids unnecessary copies.
- Integration: GoldenGate supports real-time replication and change capture; Data Integration covers ETL/ELT; Streaming handles event streams. Match the tool to latency and transformation needs.
- Security and governance: compartments and IAM policies control access; encryption, auditing, and Data Safe address sensitive data. Governance is part of the architecture, not an afterthought.
Worked example
A retailer keeps point-of-sale transactions in Autonomous Transaction Processing and clickstream JSON files in Object Storage. It needs daily sales analytics and near-real-time fraud checks. Reason it through: leave transactions in ATP for OLTP; replicate changes with GoldenGate for low-latency fraud processing; query Object Storage through lakehouse external tables for clickstream analysis; use Autonomous Data Warehouse for aggregated reporting. Do not load all JSON into relational tables first, and do not use batch ETL where near-real-time replication is required.
Traps that cost marks
Mixing up ATP and ADW; assuming “autonomous” means no design decisions; treating GoldenGate as a batch ETL tool; loading all object data before analysis; and ignoring IAM compartments or Data Safe when the scenario mentions sensitive data.