5.6 Power BI Basics for Data Fundamentals

Power BI Basics for Data Fundamentals

Power BI is Microsoft's flagship business intelligence platform and the most-tested visualization tool on DP-900. You will not be asked to write DAX, but you must know the components, the report-building workflow, the difference between datasets, reports, and dashboards, and the available storage modes.

Quick Answer: Power BI is delivered as Desktop (free Windows authoring tool), Service (the SaaS site at app.powerbi.com), Mobile apps, and Report Server (on-prem). You build content by connecting to data, transforming it with Power Query, modeling it with DAX and relationships, visualizing it as reports, and sharing through workspaces with optional row-level security.

Components

The Build Workflow

1. Connect — Pick from 200+ connectors (Excel, SQL Server, Synapse, Fabric, Salesforce, Web). Desktop is the usual starting point.
2. Transform — Use Power Query to clean and reshape data. Steps are recorded as M-language code and run on every refresh.
3. Model — Build the semantic model: tables, relationships (one-to-many is the default), calculated columns, and DAX measures.
4. Visualize — Drag fields onto a canvas to create charts, tables, slicers, maps, and custom visuals.
5. Publish — Push the .pbix file to a workspace in the Power BI Service.
6. Share — Grant access through workspace roles or build an app that bundles related reports for distribution.

Datasets, Reports, and Dashboards

These three terms describe distinct items in the service. The exam loves to test that you can tell them apart.

• Dataset (also called a semantic model in modern terminology) — The underlying data plus relationships, measures, and security. One dataset can power many reports.
• Report — A multi-page collection of visuals built against one dataset. Interactive: filtering one visual highlights others.
• Dashboard — A single page of pinned tiles. Tiles can come from many reports and datasets, but a dashboard exists only in the Power BI Service (not in Desktop) and is read-only.

Dataset (semantic model)
  -> Report (multi-page, interactive, .pbix file)
       -> Dashboard tile (single-page, service-only)

Storage Modes

A table in a Power BI semantic model has one of four storage modes. Picking the right mode trades freshness against performance and scale.

Most production datasets default to Import. DirectQuery is chosen when data must be live or when the dataset exceeds Import limits. Direct Lake is the strategic 2026 choice for Fabric workloads.

Workspaces and Sharing

Power BI content lives in workspaces. The default My Workspace is for personal scratch work. Shared workspaces are the collaboration unit and have four roles:

Beyond workspaces, you can publish an App that packages selected reports for a wider audience, or use Sharing links for one-off distribution. Power BI Premium capacities and Microsoft Fabric capacities unlock features such as paginated reports, larger datasets, and Direct Lake.

Row-Level Security (RLS)

Row-level security (RLS) lets one report show different data to different users without building separate reports. You define roles in Power BI Desktop and assign DAX filter expressions, for example [Region] = "EMEA". In the service, you map Microsoft Entra users or groups to those roles. When a user opens the report, the filter is silently applied. Object-level security (OLS) extends the idea to entire tables or columns.

Licensing in Brief

For DP-900 you do not need to memorize prices, but recognize the tiers:

• Power BI Free — Personal use only; cannot share.
• Power BI Pro — Per-user license required to share content.
• Power BI Premium Per User (PPU) — Adds Premium-only features like paginated reports and larger models, still per-user.
• Power BI Premium / Fabric capacity — Dedicated capacity; free Pro for viewers when consuming content from F64+ Fabric capacity.

Choosing the Right Visualization

The DP-900 Power BI objective explicitly includes "identify appropriate visualizations for data," so know which chart answers which question.

The exam tell: "over time" → line; "compare categories" → bar/column; "contribution to total" → pie/treemap; "single headline number" → card or KPI; "by region/country" → map.

Power Query vs DAX

A frequent point of confusion: Power Query (the M language) runs at data-load/refresh time to clean, reshape, merge, and filter data before it enters the model. DAX (Data Analysis Expressions) runs at query time to compute measures and calculated columns after the model is built, responding live as users slice and filter. Rule of thumb: shape and clean data in Power Query; calculate business metrics in DAX. Recognizing that transformation happens in Power Query while measures are written in DAX is a classic distractor pair.

Refresh and Gateways

Import-mode datasets need a scheduled refresh to stay current; the Power BI Service can refresh up to 8 times per day on Pro and more on Premium/Fabric capacity. When the data source is on-premises or in a private network, refresh flows through an on-premises data gateway installed inside that network — the same bridging idea as ADF's self-hosted integration runtime. DirectQuery and Direct Lake reduce or remove the refresh burden because they read closer to live data, which is part of why Direct Lake is the strategic 2026 choice for Fabric-backed reports.

Semantic Models, Dataflows, and Reuse

Modern Power BI promotes reuse: a single certified semantic model (dataset) can power many reports across an organization, and dataflows (Power Query in the service) let teams share standardized, pre-cleaned tables. This avoids every analyst re-importing and re-cleaning the same source. For DP-900, the takeaway is that the dataset/semantic model is the reusable data-and-logic layer, while reports and dashboards are presentation layers built on top of it.