Snowflake SnowPro Advanced: Data Engineer (DEA-C02) Sample Questions:

1. A financial institution needs to tokenize sensitive customer data (credit card numbers) stored in a Snowflake table named 'CUSTOMER_DATA before it's consumed by a downstream reporting application. The institution uses an external tokenization service accessible via a REST API. Which of the following approaches is the MOST secure and scalable way to implement tokenization during data loading, minimizing exposure of the raw credit card data within Snowflake?

A) Utilize Snowflake's Snowpipe to ingest the data directly. Inside a COPY INTO statement, use an external function to call the tokenization service during the ingestion process to tokenize the data before it's loaded into the target table.
B) Use Snowflake's Data Sharing feature to securely share the raw data with the downstream application, instructing them to perform the tokenization within their own environment.
C) Use a Snowflake UDF (User-Defined Function) written in Java that calls the external tokenization API directly. Create a masking policy that utilizes the UDF and applies it to the credit card number column.
D) Load the raw data into a staging table, then create a Snowflake Task that executes a stored procedure. The stored procedure calls the external tokenization API using 'SYSTEM$EXTERNAL_FUNCTION_REQUEST' for each row and updates the original table with the tokenized values.
E) Load the raw data directly into the 'CUSTOMER DATA' table. Create a masking policy that utilizes a UDF that calls the external tokenization API directly to tokenize the credit card number values on read.

2. You have a Snowflake stage pointing to an external cloud storage location containing numerous Parquet files. A directory table is created on top of it. Over time, some files are deleted or moved from the external location. You notice discrepancies between the directory table's metadata and the actual files present in the storage location. Choose the option that best describes how Snowflake handles these discrepancies and the actions you should take.

A) Snowflake requires you to drop and recreate the directory table periodically to synchronize the metadata with the external storage. Using 'ALTER DIRECTORY TABLE REFRESH' will not remove deleted files from the directory table's metadata. However, these invalid files wont be shown in select unless explicitly used.
B) Snowflake automatically detects deleted files and marks them as 'invalid' in the directory table. Queries will automatically exclude these invalid files.
C) Snowflake automatically updates the directory table in real-time, reflecting the changes immediately. No action is needed.
D) Snowflake does not automatically detect these changes. You must manually refresh the directory table using 'ALTER DIRECTORY TABLE ... REFRESH' to synchronize the metadata. Snowflake does not provide an automated cleanup of metadata associated with removed files.
E) Snowflake does not track file deletions. If a file is deleted from cloud storage after being added to a directory table, Snowflake continues to reference the deleted file, potentially causing errors during data loading. Run 'VALIDATE on the directory table.

3. A data engineering team has implemented a continuous data pipeline that loads data into a Snowflake table named 'SALES DATA' They notice that the pipeline intermittently experiences performance degradation, particularly during peak business hours. The team wants to implement alerts to proactively identify and address these performance issues. Which of the following approaches would be MOST effective for monitoring the pipeline and triggering alerts based on specific performance metrics related to data loading?

A) Enable Snowflake's query acceleration service. This service automatically analyzes query performance and identifies opportunities for optimization, removing the need for manual monitoring and alerting. Use Snowflake's resource monitors to track credit usage.
B) Implement a data streaming service that monitors the 'SALES_DATX table in real-time. The streaming service should track the number of rows inserted per minute and trigger an alert if the insertion rate drops below a predefined threshold. No Snowflake object or Alert required.
C) Create a Snowflake Task that periodically queries the 'QUERY_HISTORY view, calculates the average load duration for 'SALES_DATX, and triggers an alert if the duration exceeds a predefined threshold. Use a Stored Procedure to handle the alert logic and send notifications.
D) Create a Snowflake Alert based on a metric in the Account Usage views (e.g., that identifies when load durations for the data warehouse associated with the data pipeline exceed a specified threshold. Configure a Notification Integration to route alerts to a designated channel.
E) create a custom Snowflake Alert that triggers when the function for the 'SALES_DATA' table indicates significant delay in data loading. Use a Snowflake Notification Integration to send alerts via email or Slack.

4. You are responsible for ensuring data consistency across multiple Snowflake tables involved in a financial reporting system. You've noticed discrepancies in aggregate calculations between a 'TRANSACTIONS" table and a summary table 'MONTHLY REPORTS'. The 'TRANSACTIONS' table is frequently updated via streams and tasks. Which combination of the following strategies would be MOST effective in identifying and resolving these inconsistencies in near real-time?

A) Utilize Snowflake's Time Travel feature to compare the ' TRANSACTIONS' table and 'MONTHLY _ REPORTS' table at a specific point in time and identify the changes that led to the discrepancies.
B) Implement a Snowflake task that periodically recalculates the 'MONTHLY_REPORTS' table from the 'TRANSACTIONS table and compares the results with the existing data, logging any discrepancies. Use a smaller warehouse size to minimize cost.
C) Implement data validation checks within the data pipeline (streams and tasks) that update the 'TRANSACTIONS' table to reject transactions that violate predefined business rules.
D) Create a Snowflake alert that triggers when the difference in the total 'SALE_AMOUNT between the 'TRANSACTIONS' table and 'MONTHLY REPORTS' exceeds a predefined threshold within a specified time window.
E) Use Snowflake's row access policies to restrict access to the 'TRANSACTIONS' table, forcing users to only access the 'MONTHLY REPORTS table.

5. A data engineering team is building a real-time fraud detection system. They have a large 'TRANSACTIONS table that grows rapidly. They need to calculate the average transaction amount per merchant daily. The following query is used:

This query is run every hour and is performance-critical. Which of the following materialized view definitions would provide the BEST performance improvement, considering the need for near real-time data and minimal latency?

A) Option E
B) Option D
C) Option C
D) Option A
E) Option B

Solutions: