1Z0-184-25 Question 56
Single answerYou are designing an application on Oracle Cloud Infrastructure using Oracle Autonomous Database to store vector embeddings for a content recommendation platform. You have created an HNSW vector index on the table that holds the embeddings to enable approximate nearest neighbor (ANN) queries. However, new data is ingested daily, requiring regular index maintenance to preserve query accuracy. Which approach is recommended to efficiently keep the HNSW index up to date without incurring significant query latency?
- A
Use partition-level index rebuilds immediately after each insertion to ensure all new entries are reflected in real time.
- B
Configure a periodic incremental refresh for the HNSW vector index to batch new embeddings and update the index efficiently.
- C
Enable the default 'on commit refresh' setting for the HNSW index so that each data commit automatically rebuilds the entire index.
- D
No index maintenance is required; HNSW indexes in Oracle Autonomous Database automatically detect and index new embeddings upon insert.
Show answer and explanation
Correct answer: B
Explanation
When using an HNSW vector index in Oracle Autonomous Database, best practice involves a strategy that maintains index freshness without significantly affecting query performance. A periodic incremental refresh provides a balanced approach, batching new data and updating the index more efficiently than frequent on-demand rebuilds. For reference, Oracle� documentation on managing vector indexes (including HNSW) recommends scheduling refresh operations in periods of lower activity or using incremental refresh settings to ensure updates are processed efficiently, thereby maintaining high-performance ANN queries.
- A. Incorrect.
Option 1 is incorrect. Rebuilding the entire partition-level index immediately after every insertion can be highly inefficient and introduce unnecessary overhead, leading to performance bottlenecks as the dataset grows.
- B. Correct.
Option 2 is correct. A periodic incremental refresh batches new data and updates the index in an efficient manner. This approach ensures the index remains current while minimizing query latency and resource usage�recommended for regular data ingestion scenarios.
- C. Incorrect.
Option 3 is incorrect. An 'on commit refresh' requires rebuilding the entire index on every commit, which is not practical or efficient for large-scale or frequent data ingestion, as it will impact system performance.
- D. Incorrect.
Option 4 is incorrect. HNSW indexes do not automatically update themselves without a defined refresh process. Allowing new data to remain unindexed risks stale search results and lower accuracy.