LFC Medallion Architecture Layers 3 — Questions and Answers
Question 1: A pipeline ingests CDC (Change Data Capture) events from an OLTP database. Where should these raw CDC records land first?
- Silver layer after deduplication
- Gold layer for real-time analytics
- Bronze layer without transformation (Correct answer)
- A separate quarantine zone outside the medallion
Correct answer: Bronze layer without transformation
CDC events are raw source data and should be written to the Bronze layer exactly as received before any processing.
Question 2: Which scenario would most likely cause data to be routed to a 'quarantine' table rather than Silver?
- Records with a valid but unexpected schema version
- Records that fail Silver-layer data quality checks (Correct answer)
- Records ingested more than 24 hours late
- Records originating from a non-primary source system
Correct answer: Records that fail Silver-layer data quality checks
Records failing quality checks (invalid formats, business rule violations) are quarantined so clean data flows through while bad data is investigated.
Question 3: Why might a Gold layer table in Medallion Architecture be pre-aggregated rather than computed at query time?
- To avoid granting users access to raw data
- To reduce query latency and compute cost for frequent BI workloads (Correct answer)
- Because Delta Lake does not support aggregation functions
- To comply with data retention policies
Correct answer: To reduce query latency and compute cost for frequent BI workloads
Pre-aggregating in Gold shifts heavy compute to the pipeline, so dashboards and reports execute quickly without re-scanning large datasets.
Question 4: Which statement about schema enforcement is TRUE for the Bronze layer in a typical Medallion implementation?
- Bronze enforces a strict predefined schema to prevent ingestion errors
- Bronze uses schema-on-read, accepting data in its original form (Correct answer)
- Bronze automatically converts all sources to a canonical schema
- Bronze rejects records that do not match the Silver schema
Correct answer: Bronze uses schema-on-read, accepting data in its original form
Bronze typically uses schema-on-read or flexible schema to absorb diverse source formats without transformation.
Question 5: A data engineer needs to join customer data from two Silver tables and expose the result for a Tableau dashboard. What should they create?
- A new Bronze table containing the joined raw events
- A Gold layer table with the joined and aggregated business view (Correct answer)
- A separate data mart outside the lakehouse
- A Silver view that mirrors the join logic in real time
Correct answer: A Gold layer table with the joined and aggregated business view
Joined, business-ready outputs for BI consumption belong in the Gold layer of the Medallion Architecture.
Question 6: In Medallion Architecture, what is the relationship between Silver and Bronze layer data?
- Silver replaces Bronze once processing is complete
- Silver is a refined subset derived from Bronze; Bronze is retained (Correct answer)
- Silver data is written back to Bronze for archival
- Silver and Bronze contain identical data in different formats
Correct answer: Silver is a refined subset derived from Bronze; Bronze is retained
Silver is derived from Bronze through transformation; Bronze is kept intact as the immutable source of truth for reprocessing.
Question 7: Which of the following is an example of a Silver-layer transformation?
- Storing raw JSON API responses
- Computing monthly revenue by product category
- Standardizing phone number formats and removing duplicates (Correct answer)
- Training a machine learning model on customer features
Correct answer: Standardizing phone number formats and removing duplicates
Standardizing formats and deduplicating are data cleansing operations that define Silver-layer work.
LFC Medallion Architecture Layers 3
A pipeline ingests CDC (Change Data Capture) events from an OLTP database.
Where should these raw CDC records land first?