1Z0-184-25 Question 29
Single answerYou have a table in Oracle Database 23c running on OCI that stores recommendation engine embeddings in a vector column named VECTOR_EMBEDDINGS. You want to add a new vector index on VECTOR_EMBEDDINGS to speed up similarity searches, but you also need to minimize application downtime. Which approach should you use?
- A
Use an online index creation statement on the VECTOR_EMBEDDINGS column to build the index while the table remains fully accessible.
- B
Export all rows to an external storage location, drop and recreate the table with a new index, then import the data back into the table.
- C
Disable DML operations on the table, create the index offline, and then enable DML again once the index is built.
- D
Add a new column with the index pre-created on it, copy the data from VECTOR_EMBEDDINGS, then drop the original column.
Show answer and explanation
Correct answer: A
Explanation
Online index creation is a well-documented feature in Oracle Database (including Oracle Database 23c) that allows you to build or rebuild indexes without blocking queries and DML statements. This capability is particularly useful when working with vector columns for machine learning tasks, as it reduces downtime and improves availability. Reference: Oracle Database 23c SQL Language Reference on 'Creating Indexes Online.'
- A. Correct.
Correct. Oracle Database 23c supports online index creation to minimize downtime, even for vector indexes. You can build the index on VECTOR_EMBEDDINGS in a separate session without blocking queries or data manipulation on the table.
- B. Incorrect.
Incorrect. Exporting and reimporting data involves more downtime and is more error-prone, as the table must be dropped and recreated. This is not an optimal solution for minimizing application interruption.
- C. Incorrect.
Incorrect. Disabling DML operations (offline index build) still causes downtime for any concurrent application usage. Although it might be simpler to implement, it sacrifices continuous availability during the index build.
- D. Incorrect.
Incorrect. While adding a new column is sometimes used for schema evolution, it is unnecessarily complex here and still requires data migration steps, which can lead to longer maintenance windows.