Power BI Architecture

Power BI is Microsoft’s business intelligence and data visualization platform (Now a tool in Microsoft Fabric), widely used for analytics and reporting. Many organizations are using this tool to generate reports and dashboards to make effective business decisions.

Before using Power BI services and features, it is essential to understand its architecture. It helps you know how the Power BI services and components are used to transform data and create reports and dashboards.

In this blog, we provide detailed information about the Power BI Architecture and its components. Let’s get started!

Table of Contents

•  Power BI Architecture Overview
• Components
• Power BI Service Overview
• Power BI Storage Modes Comparison
• Semantic Model Design in Power BI
• Copilot in Power BI
• FabricIQ
• Pricing and Plans
• Power BI Deployment Modes Comparison
• FAQs
• Conclusion

Microsoft Power BI Architecture - An Overview

Power BI Architecture contains several steps. Let us discuss these steps in detail:

1. Data Sources

Data is extracted from various sources, such as different servers or databases. Data from various sources can come in different types and formats. You can choose storage modes such as import, DirectQuery, and direct lake based on the size of the semantic model. You can also create composite semantic models that combine multiple storage modes.

Power BI can connect to various data sources, as shown in the image below:

2. Data Connectivity

Power BI retrieves data using multiple connectivity modes:

3. Data Transformation (Power Query)

The integrated data is not ready to visualize because it needs to be transformed. To transform the data, it should be cleaned or pre-processed.

For example, duplicates are removed, and missing values are handled according to business requirements. After the data is pre-processed or cleaned, business rules are applied to transform it. After processing the data, it can be loaded into a semantic model, a fabric Lakehouse, or a warehouse. Data is cleaned and prepared with Power Query.

Power Query allows users to:

4. Data Modeling

After data transformation, data is organized into a semantic model by:

Composite semantic models in Power BI let you combine data from various sources and use different storage modes within one semantic model.

For example, a composite semantic model can include Import, DirectQuery, and Direct Lake tables together. It helps you balance performance, data freshness, and scalability.

Composite semantic models are especially helpful when some data changes often and needs to stay live, whereas other data is mostly stable and can be imported for quicker performance.

5. Report and Publish

After sourcing and cleaning the data, you can create reports. Reports are visualizations of data, such as slicers, graphs, and charts. Power BI offers many custom visualizations for creating reports. You can create interactive reports using charts, tables, maps, KPIs, gauges, and cards.

After creating reports, you can publish them to Power BI Service or publish them to an on-premises Power BI Report Server.

6. Creating Dashboards

You can create dashboards after publishing reports to Power BI Service by pinning individual visuals from reports. The visual retains the filters when the report is saved with the individual elements. Pinning the live report page allows the dashboard users to interact with the visual by selecting slicers and filters.

These are the basic steps in the Power BI Architecture. Now we are going to discuss the components of Power BI and how they work together in the Power BI Architecture.

Components of Power BI Architecture

Let’s learn the components of Power BI Architecture in detail. These components play an important role in delivering the Power BI capabilities.            

1. Data Sources

Power BI can retrieve data from various online sources and file types. It imports information into Power BI or configures a Live Connection to receive it.

2. Power BI Desktop

It is free software that enables you to connect, transform, and visualize the data on your desktop. You can connect to various data sources using Power BI Desktop and combine them into a data model. It uses Power Query, DAX, semantic models, direct lake, and composite models.

This data model enables you to create interactive reports and visualizations to share information across the organization. Most users working on Business Intelligence projects use Power BI Desktop to create and share their reports.

3. Power BI Report Server

Power BI Report Server is an on-premises server platform that supports only on-premises deployment and a subset of cloud capabilities.

Using Power BI Report Server, organizations can secure their data. It enables users to create reports and dashboards and share them with other users or organizations, with proper security protocols.

4. Power BI Gateway

Power BI Gateway keeps information up to date by connecting to your on-premises data sources without transferring data. It provides secure data transfer between Microsoft cloud services and on-premises services.

Microsoft cloud services include PowerApps, Power BI, Azure Analysis Services, Power Automate, and Azure Logic Apps. By using a gateway, organisations can securely maintain databases and other data sources within cloud services.

5. Power BI Mobile Apps

Using Power BI Mobile Apps, you can stay connected with on-premises data from anywhere. Power BI apps are available for iOS and Android platforms.

6. Power BI Embedded

Power BI Embedded is an Azure cloud service for embedding analytics into custom applications. It offers APIs for embedding the reports and dashboards into custom applications. So far, we have discussed the major components of Power BI, and now we will cover the remaining components as well.

Here is the list of the remaining Power BI Components.

7. Power Query

Power Query is the data connectivity feature that enables business users to access data from multiple sources and transform it to meet their business requirements. Power Query offers a custom connectors SDK so that third-party users can create their own data connectors.

8. Power Q&A

Power Q&A is a Power BI feature that lets you explore your data in your own words. In other words, you can use natural language and ask a question to get an answer from your data.

Apart from the Power BI architecture components discussed above, let’s look at some of the key features of Power BI.

9. Power BI Report Format (PBIR)

PBIR is Power BI’s developer-friendly report file format, which replaces the earlier binary .pbix file format. It saves report definitions as Git-friendly, human-readable text files (JSON) within the Power BI project (PBIP).

By using PBIR, you can improve source control, versioning, code reviews, and more. The important thing is that you can manage report definitions like application source code.

10. Tabular Model Definition Language (TMDL)

Tabular Model Definition Language (TMDL) is a text-based language for defining and managing tabular semantic models in Power BI, Microsoft Fabric, and SQL Server Analysis Services (SSAS) Tabular.

TMDL represents semantic model metadata in human-readable text files. This language simplifies the development, versioning, review, and deployment of semantic models.

Power BI Service - An Overview

Power BI Service is a cloud-based SaaS service used to share and publish reports created in Power BI Desktop. It enables collaboration among users and creates dashboards.

Power BI Service enables users to create and access reports and dashboards on client platforms, including mobile devices and websites. Users interact with the Power BI Service whenever they want to access data created in Power BI. So, now, we will learn how the Power BI Service works.

How Does the Power BI Service Work?

• Power BI stores data in two primary repositories: Azure SQL Database and Azure Blob Storage. Azure Blob Storage enables users to store semantic models, while all system-related data and metadata are stored in Azure SQL Database.
• It authenticates the users’ requests and sends them to the Gateway Role. It processes requests and assigns them to the appropriate components, such as the Background Job Processing Role, Data Movement Role, Presentation Role, and Data Role.
• The Presentation Role manages all visualization queries, such as reports and dashboards.
• The Presentation Role sends requests to the Gateway Role, which then routes them to the Data Movement Role or the Data Role for all relevant semantic models.
• Azure Service Bus is used to connect to and retrieve data from on-premises data sources in the cloud. It sends a request to execute queries on the on-premises data source and retrieve the data from the cloud service.

Power BI Storage Modes Comparison

Let’s look at the differences between Power BI storage modes, including Import, DirectQuery, and Direct Lake, in the table below:

Features	Import	DirectQuery	Direct lake
Data location	Data is copied into the Power BI semantic model	Data is stored in the source systems	Data is located in Fabric OneLake
Need for Refreshes	Yes	No	No
Query latency	Very low	Depends on the performance of source systems	Very low
Best for	High performance dashboards and historical analysis	Operational reporting and real-time dashboards	Large-scale analytics on Fabric
Requirement for Fabric	No	No	Yes

Semantic Model Design in Power BI

It is the process of organizing business data into a logical, optimized, and reusable structure for analytics. A well-designed semantic model defines tables, relationships, measures, hierarchies, security, and business logic.

The semantic model serves as the single source of truth for reports, dashboards, scorecards, and AI features like Copilot. The core components of the semantic model design are listed below:

Most importantly, the star schema is the preferred design pattern for Power BI semantic models. It improves performance, simplifies DAX, and makes models easier to understand.

Copilot in Power BI

Copilot in Power BI is an AI assistant that helps you create, analyze, and improve Power BI content using natural language. As part of Microsoft Fabric, Copilot helps build semantic models, generate DAX formulas, summarize data, generate reports, and derive insights.

Copilot works efficiently when Fabric-ready semantic models are AI-ready. In other words, the models must have defined relationships, business-friendly field naming, and AI instructions.

Let’s explore more about Copilot:

• Both business users and BI professionals can use Copilot. 
• It automates repetitive tasks while keeping human oversight. 
• It uses large language models (LLMs) and your organization's data, metadata, and Microsoft Fabric services to understand user requests and generate relevant analytics.
• With the help of Copilot, you can develop analytics solutions in Power BI much faster and with less manual work. 
• Copilot is available across all Microsoft Fabric SKUs starting with F2.

Use cases of Copilot:

• Generate reports using natural language prompts
• Generate DAX measures for common business calculations
• Summarize reports highlighting key trends, insights, and anomalies
• Optimize semantic models.

Learn Power BI Interview Questions and Answers that help you grab high-paying jobs.

FabricIQ - An Overview

Fabric IQ is Microsoft's AI-powered knowledge and reasoning layer for Microsoft Fabric. It supports AI agents and Copilot in gaining a deep understanding of your organization's data, metadata, business context, and relationships. As a result, they can generate more accurate, context-aware responses and automate analytics tasks.

Fabric IQ combines large language models (LLMs) with metadata from Microsoft Fabric, including semantic models, Lakehouses, Warehouses, notebooks, and governance information.

Power BI Pricing and Plans

As Power BI is a workload within Microsoft Fabric, Power BI licensing is closely tied to Microsoft Fabric features.

The table below shows the Power BI and Fabric plans and pricing details as of August 2026.

Plan	Pricing (As of August 2026)
Power BI Free	Free
Power BI Pro	₹1165 or $12 per user/month and yearly payment
Power BI Premium Per User (PPU)	₹1995 or $21 per user/month and yearly payment
Power BI Embedded	Variable
Microsoft Fabric Free	Free
Microsoft Fabric Reservation	Variable
Microsoft Fabric Pay-as-you-go	Variable

If you want further information, you can refer to the official Microsoft pricing webpage.

Power BI Deployment Modes Comparison

Next, we’ll compare various Power BI deployment modes based on their advantages and limitations.

Deployment mode	Best For	Limitations
Power BI Service	Enterprise-level BI Dashboard creation Self-service analytics	It needs cloud connectivity
Power BI Report Server	On-premises reporting with compliance	It doesn't support Fabric workloads
Power BI Embedded (A-SKU & F-SKU)	It enables embedded analytics into applications	It needs capacity planning
Microsoft Fabric capacity (F-SKU)	It is a unified analytics platform that includes Power BI, OneLake, lakehouses, and more.	It needs Fabric capacity licensing

Frequently Asked Questions

1. What is the architecture of Power BI?

Power BI architecture is the complete framework that connects data from various sources, transforms and models it, creates interactive reports and dashboards, and securely shares insights with users and devices.

2. What is Direct Lake mode in Power BI?

Direct Lake is a fast data connection option in Power BI, which allows reports and semantic models to access data directly from Microsoft Fabric OneLake. You don’t need to import data into Power BI or send live SQL queries to the original data source while using this mode.

3. What is Microsoft Fabric and how does Power BI fit in?

Power BI is the Business Intelligence (BI) and data visualization tool within Microsoft Fabric. It allows users to connect to Fabric data, build semantic models, create interactive reports and dashboards, and share insights throughout the organization.

4. What is a Power BI semantic model?

A Power BI semantic model serves as the business layer that organizes, secures, and defines data for analysis and reporting. It transforms raw data into a structured model that you can easily understand and query.

A semantic model includes tables, relationships, calculations, and security rules that allow multiple reports and dashboards to use the same business data.

5. Which roles need Power BI architecture skills?

Power BI Developers, Data Analysts, BI Architects, and Data Engineers must clearly understand Power BI architecture and how it works.

6. Compare Power BI: Import vs DirectQuery vs Direct Lake.

Import mode excels in loading data into the Power BI semantic model. It is best for dashboard creation and historical reporting. DirectQuery stores data in the source system and retrieves it in real time. It is best for operational reporting.

Direct Lake is made for Microsoft Fabric. It allows Power BI to access data directly from OneLake with the same performance as Import mode. It eliminates the need for scheduled refreshes and reduces data duplication. It is best for modern Fabric-based analytics solutions.

Conclusion

In this blog, we shared detailed information about the Power BI Architecture, how it works, and its components. We have also explained the Power BI Service and how it works.

If you want to learn more about Power BI, you can register for a Power BI course at MindMajix, which will help you delve deeper into the Power BI features and provide you with the expertise to develop powerful data visualizations.