SnowPro Associate: Platform Question 289
Single answer3.4 Explain how to work with unstructured data.A media company stores product manuals as PDF files in an Amazon S3 bucket. The files must be made available in Snowflake so analysts can query file metadata and data engineers can later build pipelines to process the documents. The team wants the fastest approach that lets Snowflake securely reference the files without loading their contents into relational columns first. Which action should the team take?
- A
Create an external stage that points to the S3 bucket and use DIRECTORY tables or file metadata functions to work with the files.
- B
Create a standard table with a VARCHAR column and load each PDF file into the table by using COPY INTO.
- C
Convert all PDF files to CSV outside Snowflake, then ingest them into permanent tables before Snowflake can access them.
- D
Create a materialized view on top of the S3 bucket so Snowflake can directly index the PDF document contents.
Show answer and explanation
Correct answer: A
Explanation
For unstructured data in Snowflake, the foundational pattern is to place files in an internal or external stage and then reference those files from Snowflake rather than loading them into standard relational columns first. In this scenario, because the PDF files already reside in Amazon S3, an external stage is the most appropriate solution. Snowflake supports working with staged unstructured files and can expose file-level details through features such as directory tables and metadata functions, enabling analysts and engineers to inventory and manage the files before building downstream processing. This aligns with Snowflake best practices for unstructured data: use stages to securely reference files in cloud storage, then use metadata and downstream processing as needed. Documentation topics relevant here include Snowflake stages, external stages, directory tables, and working with unstructured data.
- A. Correct.
Correct. Snowflake works with unstructured data by referencing files in internal or external stages. For files already stored in Amazon S3, creating an external stage is the appropriate first step. Snowflake can then use capabilities such as directory tables and metadata access to list and manage files without first loading the binary document contents into relational columns. This matches the requirement for a fast, secure approach that references the files in place.
- B. Incorrect.
Incorrect. COPY INTO is used to load structured or semi-structured data from staged files into tables, but that is not the right first step for PDF documents as unstructured data. Loading binary files into a VARCHAR column is not the intended pattern for working with unstructured documents in Snowflake and does not satisfy the requirement to reference files directly without first transforming them into table data.
- C. Incorrect.
Incorrect. Converting PDFs to CSV is unnecessary and impractical for this scenario. Snowflake can work with unstructured files through stages, so forcing a conversion before Snowflake can reference the files adds complexity and delays. This distractor reflects the misconception that all data must be converted into structured tabular format before Snowflake can use it.
- D. Incorrect.
Incorrect. Snowflake does not create a materialized view directly on top of an S3 bucket to index PDF file contents. Materialized views are built on Snowflake tables or certain query results, not directly on external object storage for unstructured document indexing. This option confuses Snowflake storage objects with external cloud storage integration.