Question 39
Domain 10: Data ModellingFor a new large Unity Catalog managed Delta table with evolving filter patterns, which layout strategy is generally recommended over rigid partitioning?
Correct answer: B
Explanation
Liquid clustering is recommended because it organizes data for changing query patterns without requiring fixed partition columns. Unlike rigid partitioning, it lets Unity Catalog managed Delta tables adapt as filters evolve, improving data skipping and maintenance flexibility.
Why each option is right or wrong
A. High-cardinality partitioning on customer_id
High-cardinality columns usually make poor partition keys and can create too many small partitions.
B. Liquid clustering
The exam guide explicitly lists “Simplify data layout decisions and optimize query performance using Liquid Clustering” and “Identify the benefits of using liquid Clustering over Partitioning and ZOrder” under Data Modelling, so for a new Unity Catalog managed Delta table with changing filter predicates, the expected choice is liquid clustering rather than fixed partitions. Partitioning is rigid because the partition columns are chosen up front and are costly to change later, whereas liquid clustering is designed for evolving access patterns on managed tables and preserves optimization flexibility without requiring a repartition/rewrite cycle.
C. Writing every day to a new table
Creating a new table daily fragments data and does not solve layout optimization for evolving filters.
D. Turning off data skipping
Data skipping improves query performance; disabling it removes an optimization Databricks uses for large datasets.