BenchMark Quizzes by MDFT Pro

Resolve Direct Lake Performance Issues

You work as a Performance Optimization Specialist for Mark at MDFT Pro, a well-known training agency. MDFT Pro operates a Fabric F32 capacity with a workspace containing a warehouse named DW1 that stores comprehensive training analytics data. The warehouse is modeled using MD5 hash surrogate keys for dimension tables. Over the past year, the main fact table in DW1 has grown significantly from 200 million rows to 500 million rows as MDFT Pro expanded its course offerings and student base. The agency has deployed Microsoft Power BI reports based on Direct Lake that show year-over-year comparisons of student enrollment, course completion rates, and revenue metrics. Recently, users have reported that the performance of some reports has degraded noticeably, and some visuals are displaying errors. Mark has asked you to resolve these performance issues while ensuring the solution provides the best query performance and minimizes operational costs.

What should you do?

Choose the correct answer from the options below.

Explanations for each answer:

Change the MD5 hash to SHA256 is incorrect. Changing the hash algorithm from MD5 to SHA256 would not improve query performance; in fact, SHA256 is more computationally expensive than MD5 and would likely degrade performance further.
Increase the capacity is incorrect. While increasing capacity might temporarily improve performance, it would significantly increase operational costs and doesn't address the root cause of inefficient data storage for Direct Lake queries.
Enable V-Order is correct. V-Order is a write-time optimization that organizes Parquet files for faster reads in Direct Lake mode. It provides the best query performance for Direct Lake reports while minimizing operational costs since it's a one-time write optimization that doesn't require ongoing capacity increases.
Modify the surrogate keys to use a different data type is incorrect. Changing the data type of surrogate keys might reduce storage slightly but won't address the fundamental performance issues with reading large fact tables in Direct Lake mode.
Create views is incorrect. Creating views would add an additional layer of abstraction but wouldn't improve the underlying data read performance in Direct Lake mode, which depends on how efficiently the Parquet files are organized.

Learn more about V-Order optimization in Fabric:

V-Order Optimization

Practice Exam

Resolve Direct Lake Performance Issues

Discuss this question on social media: