1Z0-184-25 Question 124
Single answerA gaming company needs to build a high-performance Retrieve-Augment-Generate (RAG) application on Oracle Cloud Infrastructure (OCI) to handle game support tickets. They plan to convert the tickets into numeric embeddings and store them so the application can quickly retrieve relevant information for each incoming query, then pass it to a large language model for generating responses. Which solution provides the best design for storing and querying embeddings in a real-time RAG workflow on OCI?
- A
Store embeddings in Oracle Autonomous Database 23c with vector search enabled, and use Oracle Cloud Infrastructure Data Flow to periodically update the embedding index
- B
Write embeddings to Oracle Object Storage and rely on the built-in search capabilities of Object Storage to find relevant tickets
- C
Export ticket data as CSV and store it in Oracle Logging for advanced text analysis and real-time semantic searches
- D
Use Oracle Functions to process each query on-the-fly, pulling raw ticket data from the autonomous database without storing embedding indexes
Show answer and explanation
Correct answer: A
Explanation
For RAG workflows, a specialized data store that can handle vector embeddings with efficient indexing is key to low-latency semantic search. Oracle Autonomous Database 23c introduces support for vector queries, enabling you to store embeddings in JSON columns and quickly retrieve relevant data. Periodic indexing using OCI Data Flow ensures newly embedded content is searchable in real-time. Refer to Oracle's OCI documentation on Autonomous Database 23c enhancements and best practices for storing and querying vector embeddings.
- A. Correct.
Correct. Oracle Autonomous Database 23c supports storing vector embeddings (via JSON columns or similar structures) and provides a native vector search feature. Periodically updating the index using Oracle Cloud Infrastructure Data Flow ensures newly ingested tickets get indexed and remain searchable in near real-time, which is critical for an RAG application.
- B. Incorrect.
Incorrect. Object Storage is excellent for cost-effective data archiving but does not offer built-in vector or semantic search. Storing embeddings in Object Storage would require a separate indexing mechanism outside of Object Storage, complicating the real-time retrieval process.
- C. Incorrect.
Incorrect. Oracle Logging collects, indexes, and analyzes log events, not vector embeddings or semantic structures suitable for RAG. It is typically used for operational observability and compliance rather than advanced text retrieval.
- D. Incorrect.
Incorrect. Processing raw ticket data for every query without storing embeddings would dramatically increase query latency and system overhead. A core principle of RAG architectures is to store precomputed embeddings for rapid semantic retrieval.