Describe an analytics workload on Azure Study Guide — DP-900

Describe an Analytics Workload on Azure – DP-900 Study Guide

This domain accounts for 25–30% of the exam — tied for the largest — and covers how organizations turn raw data into insight: the analytics pipeline, data-warehousing concepts, batch vs real-time processing, Microsoft Fabric, and Power BI fundamentals including choosing the right visualization. Questions blend conceptual (what is a fact table?) with practical (which chart shows a trend?). Learn the pipeline shape and the visualization rules and this domain scores well.

Elements of a Modern Analytics Solution

Data flows through predictable stages — knowing the order is testable:

• Ingest — capture data from sources (databases, files, streams, APIs).
• Store — land it in a data lake (cheap, flexible storage for any data).
• Process/transform — clean, join, and reshape it.
• Model — organize it for analysis in a data warehouse or lakehouse.
• Visualize — deliver insight through reports and dashboards.

A key architectural idea: separate storage from compute, so each scales independently and multiple engines can query the same lake without duplicating it (the foundation of Fabric's OneLake).

Data Warehousing Concepts

Analytical data is usually modeled as a star schema:

• Fact table — the center; holds measurable, numeric events (sales quantity, revenue) plus foreign keys to dimensions.
• Dimension tables — surround the fact table; hold descriptive attributes you filter and group by (product, customer, date, geography).

This denormalized shape is optimized for reads and aggregation — the opposite priority of a normalized OLTP schema, which is why operational data is transformed into star schemas when it lands in the warehouse. When a report needs to join eight normalized tables, that's the signal analytics wants a star schema instead.