Free Microsoft Certified: Azure Data Fundamentals (DP-900) Sample Questions

Correct answer: C — Unstructured

• A. Incorrect. Structured data conforms to a strict, predefined schema, typically organized in a tabular format with rows and columns, like in a relational database. Medical imaging files and PDFs do not fit this classification as they lack a fixed schema for their content.
• B. Incorrect. Semi-structured data does not have a rigid schema but contains organizational properties like tags or keys that make it self-describing, such as JSON or XML files. While image files and PDFs may contain metadata, their core content (image pixels, text, and layout) does not have this inherent self-describing structure.
• C. Correct. Unstructured data lacks any predefined data model or schema. This category includes binary files such as images (X-rays, MRI scans), videos, audio files, and documents like PDFs. This type of data is often stored in data lakes or object/blob storage.
• D. Incorrect. Relational data is a specific type of structured data that is organized into tables with defined schemas and relationships between them. This model is not suitable for storing and classifying raw binary files like images or PDFs, even though a relational database might store a pointer to such a file or the file itself as a BLOB (Binary Large Object).

Correct answer: B — Relational data

• A. Incorrect. Non-relational data, often associated with NoSQL databases, does not follow the rigid tabular structure of rows and columns. It encompasses various data models like document, graph, wide-column, and key-value stores, which offer more flexibility than the strict schema described.
• B. Correct. This is the definition of relational data. It is structured into tables (relations) consisting of rows (tuples) and columns (attributes). Each column enforces a specific data type, and relationships between tables are explicitly defined and maintained using primary and foreign keys to ensure data integrity.
• C. Incorrect. Unstructured data lacks a predefined data model or schema. It does not fit into the organized structure of rows and columns found in relational databases. Examples include text files, images, videos, and audio files.
• D. Incorrect. Key-value data is a type of non-relational data where information is stored as a collection of key-value pairs. While it is structured, it does not use the tabular format with typed columns and defined relationships between tables that characterizes relational data.

Correct answer: A — Relational database

• A. Correct. Relational databases are the classic example of a 'schema-on-write' model. They enforce a strict, predefined schema consisting of tables, columns with specific data types, and constraints (like NOT NULL or CHECK). Data is validated against this schema before it can be written (inserted or updated), ensuring high data integrity and consistency.
• B. Incorrect. Key-value stores are a type of NoSQL database that is typically schema-less. They store a simple key paired with a value, which is often treated as an opaque blob of data. The structure and interpretation of the value are left to the application, which is a 'schema-on-read' approach.
• C. Incorrect. Document stores are NoSQL databases that are schema-flexible, storing data in formats like JSON or BSON. While some modern document databases allow for optional schema validation rules, their fundamental design allows documents within the same collection to have different structures. This flexibility is characteristic of a 'schema-on-read' approach.
• D. Incorrect. Column-family stores, also known as wide-column stores, are a type of NoSQL database designed for high scalability and schema flexibility. They do not enforce a uniform schema for every row, allowing different rows to have different columns. This is the opposite of the strict 'schema-on-write' model.

Correct answers: A, B — Azure SQL Database; Azure Synapse Analytics (Dedicated SQL Pool)

• A. Correct. Azure SQL Database is a fully managed relational database service (PaaS) that stores structured data in tables with predefined schemas. It supports T-SQL and ACID transactions, making it a primary choice for storing relational data for various applications, including analytics.
• B. Correct. Azure Synapse Analytics (specifically its Dedicated SQL Pool feature) is a distributed, Massively Parallel Processing (MPP) relational data warehouse. It is purpose-built to store and process extremely large volumes of structured, relational data for high-performance analytics and business intelligence.
• C. Incorrect. Azure Blob Storage is an object storage service designed for storing large amounts of unstructured or semi-structured data, such as files, images, and backups. It does not enforce a relational schema or provide native relational query capabilities.
• D. Incorrect. The Azure Cosmos DB Gremlin API provides a graph database. This is a type of NoSQL database that stores data as nodes and edges, which is a non-relational model designed for traversing relationships, not for storing tabular, relational data.
• E. Incorrect. Azure Data Lake Storage Gen2 is a hierarchical file system built on Azure Blob Storage, optimized for big data analytics. While it can store files containing structured data (like CSV or Parquet), it is fundamentally a file storage system, not a relational database engine that manages tables and enforces schemas.

Correct answer: B — Transactional

• A. Incorrect. Analytical workloads, also known as Online Analytical Processing (OLAP), are read-heavy and focus on complex queries and aggregations over large datasets to generate business insights. They are not optimized for handling a high volume of small, frequent write operations like inserts and updates.
• B. Correct. Transactional workloads, or Online Transaction Processing (OLTP), are characterized by a high volume of short, atomic transactions involving frequent reads, inserts, and updates. This perfectly matches the scenario of an e-commerce platform processing customer orders, which requires high throughput and data integrity (ACID properties).
• C. Incorrect. Batch processing involves collecting and processing large volumes of data at scheduled intervals or when certain conditions are met. It is not suitable for handling the immediate, small, and frequent transactions typical of an online ordering system.
• D. Incorrect. Streaming workloads involve processing a continuous flow of data in near real-time. While an e-commerce platform might use streaming for real-time analytics, the core task of reliably recording individual customer orders is a transactional workload, not a stream processing one.

Correct answer: B — OLTP

• A. Incorrect. OLAP (Online Analytical Processing) systems are optimized for complex queries and analysis over large historical datasets. They are used for business intelligence and reporting, not for high-frequency, real-time transactional updates like inventory management.
• B. Correct. OLTP (Online Transaction Processing) systems are specifically designed to handle a large volume of short, concurrent transactions with low latency and immediate consistency. This is ideal for applications like real-time inventory management where immediate updates are crucial to maintain data integrity and prevent issues like overselling.
• C. Incorrect. A Data Warehouse is a centralized repository for storing large amounts of historical data for analysis and reporting. It is typically populated through batch processes (ETL/ELT) and is not designed for real-time transactional updates.
• D. Incorrect. Batch Analytics involves processing large volumes of data in scheduled batches. This introduces significant latency between when data is generated and when it is processed, making it unsuitable for the immediate, real-time updates required by an inventory management system.

Correct answer: B — False

• A. The statement is false. Transactional workloads, also known as Online Transaction Processing (OLTP), are characterized by a high volume of short, simple transactions like INSERTs, UPDATEs, and DELETEs. These operations typically affect a small number of records and are optimized for low latency and high concurrency.
• B. The statement is false. The workload described—being read-heavy and involving complex queries across millions of rows—is characteristic of an analytical workload (Online Analytical Processing or OLAP), not a transactional one. OLAP systems are designed for business intelligence and data analysis, whereas OLTP systems handle day-to-day operational data processing.

Correct answer: C — Avro

• A. Incorrect. JSON (JavaScript Object Notation) is a text-based, human-readable format, not a binary format. While common in data pipelines, it does not have a built-in mechanism for formal schema evolution.
• B. Incorrect. Parquet is a binary format that supports schema evolution, but it is a columnar storage format. The question specifically asks for a row-based format. Columnar formats are optimized for analytical queries rather than row-based serialization for ingestion.
• C. Correct. Avro is a row-based binary serialization format. It stores its schema (in JSON format) within the data file, which provides robust support for schema evolution. These characteristics make it a popular choice for serializing data in data ingestion pipelines and streaming systems.
• D. Incorrect. CSV (Comma-Separated Values) is a row-oriented, plain-text format. It is not a binary format and has no built-in support for schemas or schema evolution, making it less robust when data structures change.

Correct answer: B — Document database

• A. Incorrect. Relational databases enforce a predefined, fixed schema with tables and columns. Storing items with vastly different attributes would require a very wide table with many nullable columns or a complex Entity-Attribute-Value (EAV) model, neither of which is efficient or flexible for this use case.
• B. Correct. Document databases are ideal for this scenario. They store data in flexible, JSON-like documents where each document can have its own unique structure. This allows different products with varying attributes to be stored as self-contained units in the same collection, perfectly matching the requirement for a flexible schema.
• C. Incorrect. While column-family databases support flexible schemas with wide, sparse rows and are excellent for high-throughput, large-scale workloads, they are not as naturally suited as document databases for storing and querying nested, self-contained objects like individual product specifications.
• D. Incorrect. Graph databases are specialized for modeling and querying complex relationships between entities (nodes and edges). While you can store properties on nodes, their primary strength is not storing heterogeneous product records as self-contained units but rather analyzing the connections between them.

Correct answers: B, D — Nodes (Vertices); Edges (Relationships)

• A. Incorrect. Rows are fundamental components of a relational or tabular data model, not a graph data model. Graph databases work with nodes (vertices) and edges, not rows and columns.
• B. Correct. Nodes, also known as vertices, are the fundamental entities in a graph data model. They represent objects, people, or concepts. In Azure Cosmos DB for Apache Gremlin, developers query these vertices to retrieve entities and their properties.
• C. Incorrect. Tables are data structures used in relational databases or key-value stores with a tabular structure (like Azure Cosmos DB for Table). They are not a component of a graph data model.
• D. Correct. Edges, also known as relationships, represent the connections between nodes (vertices) in a graph. Queries in a graph database traverse these edges to understand and analyze the relationships between entities. Edges can also have their own properties.
• E. Incorrect. Columns are part of a relational or tabular data structure. In a graph model, data is stored as properties on nodes and edges, not in fixed columns within a table.

Correct answer: C — Database Administrator

• A. Incorrect. A Data Engineer's primary focus is on designing, building, and maintaining data pipelines and ETL/ELT processes. They are responsible for data integration and transformation, not for the operational management of database recovery and SLA compliance.
• B. Incorrect. A Data Analyst is a consumer of data who focuses on analyzing data to extract insights, create reports, and support business decision-making. They do not manage database infrastructure, backups, or disaster recovery settings.
• C. Correct. A Database Administrator (DBA) is responsible for the overall management, maintenance, and performance of databases. This includes configuring backup strategies, high availability, and disaster recovery options, which directly determine the Recovery Point Objective (RPO). Ensuring these configurations meet the business's Service Level Agreement (SLA) is a core responsibility of the DBA role.
• D. Incorrect. An AI Engineer focuses on developing, training, and deploying artificial intelligence and machine learning models. Their responsibilities do not include managing database infrastructure or ensuring disaster recovery objectives are met.

Correct answers: B, C — Designing an Azure Data Factory pipeline.; Cleaning and transforming raw data into a usable format.

• A. Incorrect. Creating reports and visualizations, such as a paginated report in Power BI, is typically the responsibility of a Data Analyst or a Business Intelligence (BI) Developer who consumes the data prepared by data engineers.
• B. Correct. Designing and building data pipelines using tools like Azure Data Factory is a core responsibility of a Data Engineer. This involves creating and orchestrating ETL (Extract, Transform, Load) or ELT (Extract, Load, Transform) workflows for data movement and transformation.
• C. Correct. A primary responsibility of a Data Engineer is to take raw data from various sources and apply processes for cleaning, transforming, normalizing, and enriching it. This makes the data reliable and usable for analysis and other applications.
• D. Incorrect. Managing user access, permissions, and login credentials for a database is a classic task for a Database Administrator (DBA) or a security administrator, not a Data Engineer.
• E. Incorrect. Performing operational tasks like backing up databases and transaction logs is a fundamental duty of a Database Administrator (DBA) to ensure data availability, integrity, and disaster recovery.

Correct answer: C — Data Engineer prepares; Data Analyst finds insights

• A. Incorrect. This option reverses the typical roles. Data Engineers are responsible for creating data pipelines to process raw, messy data, while Data Analysts consume the prepared data to find insights.
• B. Incorrect. A Database Administrator's primary role is managing and maintaining database systems, not preparing unstructured data for analysis. While the Data Analyst is correctly identified as the one finding insights, the preparation role is wrong.
• C. Correct. This accurately reflects the standard division of responsibilities. A Data Engineer builds the infrastructure and pipelines to ingest, clean, and transform messy, unstructured data into a usable format. A Data Analyst then uses this prepared data to perform analysis, create visualizations, and extract business insights.
• D. Incorrect. This option misattributes both roles. Data Analysts are not typically responsible for preparing large-scale unstructured data, and Database Administrators are not primarily tasked with finding business insights through data analysis.

Correct answer: A — True

• A. The statement is true because creating an index on a frequently queried column provides a fast lookup mechanism for the database engine. Instead of performing a full table scan to find data, the engine can use the index to locate the required rows much more efficiently, which significantly improves the performance of read queries (e.g., those with WHERE or ORDER BY clauses on the indexed column).
• B. The statement is false because the primary purpose of creating an index in this scenario is precisely to improve read query performance. While indexes consume storage and can add overhead to write operations (INSERT, UPDATE, DELETE), the performance gains on read-heavy workloads almost always justify their use.

Correct answer: A — True

• A. The statement is true because stored procedures in relational databases are designed to be reusable code blocks that encapsulate logic. They can accept input parameters to allow for dynamic execution and can return data to the calling application through various means, including output parameters, return codes indicating execution status, and by returning one or more result sets.
• B. The statement is false because it incorrectly describes the functionality of stored procedures. A key feature and common use case for stored procedures is their ability to be parameterized with both input and output parameters, which allows for flexible and powerful data manipulation and retrieval.

Correct answer: B — False

• A. The statement is false because a primary key constraint's core purpose is to uniquely identify each row in a table. This is enforced through two implicit rules: the values in the key column must be unique, and they cannot be NULL. Allowing NULL values would violate the principle of unique identification.
• B. The statement is false because any column defined with a Primary Key constraint automatically has a NOT NULL constraint applied to it. This ensures data integrity by guaranteeing that every record has a unique, non-missing identifier, thus prohibiting any NULL entries.

Correct answer: B — Azure SQL Managed Instance

• A. Incorrect. Azure SQL Database is a Platform-as-a-Service (PaaS) that focuses on the database itself, abstracting away instance-level features. It does not natively support SQL Server Agent jobs or cross-database queries in the same way as an on-premises instance, failing the near 100% compatibility requirement for these specific features.
• B. Correct. Azure SQL Managed Instance is a fully managed PaaS offering designed for this exact scenario. It provides near 100% compatibility with the on-premises SQL Server engine, including support for instance-scoped features like Cross-Database queries and SQL Server Agent jobs, while minimizing the effort of managing the underlying infrastructure.
• C. Incorrect. While SQL Server on Azure Virtual Machines offers 100% compatibility, it is an Infrastructure-as-a-Service (IaaS) solution. This means the company would still be responsible for managing the operating system, patching, and database software, which contradicts the key requirement to minimize infrastructure management effort.
• D. Incorrect. Azure Database for MySQL is a service for the MySQL database engine, which is a completely different platform from Microsoft SQL Server. It is not compatible with SQL Server features like SQL Agent or T-SQL syntax for cross-database queries.

Correct answer: B — Microsoft

• A. Incorrect. In a Platform-as-a-Service (PaaS) model like Azure SQL Database, the cloud provider handles the underlying infrastructure. The database administrator's (DBA) role shifts from managing hardware and OS patching to focusing on database schema, performance tuning, security configuration, and data management.
• B. Correct. Azure SQL Database is a fully managed Platform-as-a-Service (PaaS) offering. According to the shared responsibility model for PaaS, Microsoft is responsible for managing the underlying infrastructure, which includes applying all software patches and updates to both the operating system and the SQL Server database engine.
• C. Incorrect. The application developer is responsible for writing, deploying, and maintaining the application code that interacts with the database, but they are not responsible for managing or patching the underlying database platform or operating system.
• D. Incorrect. A cloud architect is responsible for designing the overall cloud solution, making decisions on which services to use, and defining the architecture. They do not perform the operational, day-to-day tasks like applying patches to a managed service.

Correct answer: B — False

• A. The statement is false. While Azure SQL Database is a fully managed PaaS service that Microsoft keeps updated, its codebase and release cycle are distinct from the on-premises SQL Server box product. It does not necessarily run the same version as the latest stable on-premises release; in fact, new features often appear in Azure first.
• B. The statement is false. Azure SQL Database is based on the SQL Server engine but operates on a continuous update model with its own versioning, which does not directly correspond to on-premises versions like SQL Server 2022. It is an 'evergreen' service that is constantly evolving, and database compatibility levels are used to manage feature behavior for applications.

Correct answer: C — Azure Files

• A. Incorrect. Azure Blob Storage is an object store for large amounts of unstructured data, such as text or binary files. It is accessed via REST APIs or SDKs and does not natively support the SMB protocol. Using Blob Storage would require significant code changes to the legacy application.
• B. Incorrect. Azure Queue Storage is a messaging service designed for asynchronous communication between application components. It is used to store and retrieve messages and does not provide file share capabilities or support for the SMB protocol.
• C. Correct. Azure Files provides fully managed file shares in the cloud that support the Server Message Block (SMB) protocol. This allows you to mount the file share just like a traditional on-premises network share, making it the ideal choice for a 'lift-and-shift' migration of applications that depend on SMB, with minimal or no code changes required.
• D. Incorrect. Azure Table Storage is a NoSQL key-value store for structured, non-relational data. It is not a file system and does not expose a file share interface or support the SMB protocol, making it unsuitable for this scenario.

Correct answers: B, D — PartitionKey; RowKey

• A. Incorrect. Timestamp is a system-managed property automatically added by Azure Table Storage to record when an entity was last modified. It is useful for tracking changes but is not part of the unique identifier (primary key).
• B. Correct. The PartitionKey is the first part of the compound primary key. It is used to group related entities and partition the data across storage nodes, which is crucial for scalability and efficient query performance.
• C. Incorrect. The ETag (Entity Tag) is a system property used for optimistic concurrency control. It helps detect and prevent conflicting updates to an entity but is not part of the entity's unique identifier.
• D. Correct. The RowKey is the second part of the compound primary key. It provides a unique identifier for an entity within a specific partition. The combination of PartitionKey and RowKey uniquely identifies every entity in a table.
• E. Incorrect. TenantId is not a standard system property used by Azure Table Storage to form the unique identifier. While a developer could choose to include a TenantId as a custom property within their data model, it is not an inherent part of the table's keying scheme.

Correct answers: A, B, C — Locally-redundant storage (LRS); Zone-redundant storage (ZRS); Geo-redundant storage (GRS)

• A. Correct. Locally-redundant storage (LRS) is a valid redundancy option where data is replicated synchronously three times within a single physical location (a single data center) in the primary region. It protects data from server rack and drive failures.
• B. Correct. Zone-redundant storage (ZRS) is a valid redundancy option that replicates data synchronously across three Azure Availability Zones in the primary region. This provides high availability by protecting against data center-level failures within a single region.
• C. Correct. Geo-redundant storage (GRS) is a valid redundancy option that copies data synchronously three times in the primary region and then asynchronously replicates the data to a secondary region hundreds of miles away. It protects against regional outages or disasters.
• D. Incorrect. 'Container-redundant storage (CRS)' is not a valid Azure Storage redundancy option. Redundancy is configured at the storage account level, and this specific SKU does not exist.
• E. Incorrect. 'File-redundant storage (FRS)' is not a recognized Azure Storage redundancy SKU. Azure Files uses the standard redundancy options available for the storage account it is created in, such as LRS, ZRS, or GRS.

Correct answer: A — True

• A. The statement is true. The Azure Blob Storage Archive tier is designed for long-term data retention at the lowest storage cost. To achieve this, data is stored in an offline state. To access a blob in the Archive tier, it must first be rehydrated to an online tier (Hot or Cool). This rehydration process is not instantaneous and can take several hours, depending on the selected rehydration priority.
• B. The statement is false. The description provided in the question accurately reflects the behavior of the Azure Blob Storage Archive tier. Data in this tier is indeed offline and requires a rehydration process that can take hours before it becomes accessible.

Correct answer: A — Azure Cosmos DB for Table

• A. Correct. Azure Cosmos DB for Table is specifically designed as a premium offering for applications that outgrow Azure Table Storage. It provides a compatible API, allowing for migration with minimal code changes, while adding the core benefits of Azure Cosmos DB, including global distribution and guaranteed single-digit millisecond latency, which directly address the stated requirements.
• B. Incorrect. The Azure Cosmos DB for MongoDB API is designed for applications written for MongoDB and uses a document data model. Migrating from a key-value store like Azure Table Storage would require significant changes to both the application code and the data model, making it unsuitable.
• C. Incorrect. The Azure Cosmos DB for Apache Gremlin API is for graph databases and is not compatible with the key-value model used by Azure Table Storage. This would require a complete application and data model redesign.
• D. Incorrect. The Azure Cosmos DB for NoSQL (also known as the Core or SQL API) is a document-oriented API. While it offers global distribution and low latency, it is not API-compatible with Azure Table Storage. A migration would require rewriting the data access code and is not the most direct path.

Correct answers: A, C, E — Core (SQL); MongoDB; Gremlin

• A. Correct. The Core (SQL) API is the native and default API for Azure Cosmos DB. It is used to work with JSON document data using a familiar SQL-like query language.
• B. Incorrect. Oracle is a relational database management system and is not an API provided by Azure Cosmos DB. Azure Cosmos DB does not expose an Oracle-compatible wire protocol or API.
• C. Correct. Azure Cosmos DB provides an API for MongoDB that implements the MongoDB wire protocol. This allows existing MongoDB applications, drivers, and tools to interact with Azure Cosmos DB as if it were a MongoDB database.
• D. Incorrect. Neo4j is a separate and distinct graph database product. For graph database workloads, Azure Cosmos DB provides the Gremlin API, not a Neo4j-compatible API.
• E. Correct. The Gremlin API, based on Apache TinkerPop, is supported by Azure Cosmos DB for building graph databases. It allows users to perform graph traversals using the Gremlin query language.
• F. Incorrect. Redis is an in-memory key-value store. While Azure offers a managed Redis service called Azure Cache for Redis, it is a separate service and not an API for Azure Cosmos DB.

Correct answer: A — True

• A. The statement is true because Eventual consistency is the most relaxed of the five consistency levels in Azure Cosmos DB. It allows reads to be served from any available replica without needing to coordinate with other replicas to guarantee the absolute latest data. This lack of coordination overhead significantly reduces read latency and maximizes read throughput, making it the fastest option.
• B. The statement is false because Strong consistency provides the tightest guarantees, ensuring that every read operation returns the most recently committed version of an item. To achieve this, it requires consensus across a quorum of replicas, which inherently introduces network latency and coordination overhead. This results in higher latency and lower throughput compared to more relaxed levels like Eventual consistency.

Correct answer: B — Azure Databricks

• A. Incorrect. Azure Stream Analytics is a real-time analytics and complex event-processing engine, but it is not based on Apache Spark. It uses a proprietary SQL-like query language and is not designed for the type of collaborative, Spark-based development the company requires.
• B. Correct. Azure Databricks is a fully managed analytics platform built around and optimized for Apache Spark. It supports real-time processing via Spark Structured Streaming and provides an integrated, collaborative workspace with features like interactive notebooks, enabling data engineers, data scientists, and business analysts to work together.
• C. Incorrect. Azure SQL Database is a fully managed platform-as-a-service (PaaS) relational database. It is designed for transactional workloads and is not an Apache Spark platform for real-time data processing.
• D. Incorrect. Azure Analysis Services is a fully managed platform-as-a-service (PaaS) that provides enterprise-grade semantic data models for Business Intelligence (BI) and OLAP-style queries. It does not use Apache Spark and is not intended for real-time data processing workloads.

Correct answers: A, C — Azure Stream Analytics; Spark Structured Streaming

• A. Correct. Azure Stream Analytics is a fully managed, low-latency stream processing service designed specifically for real-time analytics and complex event processing. It can ingest data from streaming sources like Azure Event Hubs, apply transformations and windows, and output results with minimal latency, making it a perfect fit for the Speed Layer in a Lambda architecture.
• B. Incorrect. The Archive Tier of Azure Data Lake Storage is optimized for long-term, low-cost storage of data that is rarely accessed. It has very high retrieval latency (potentially hours), which is the opposite of the low-latency processing required for the Speed Layer. This tier is suitable for cold data storage, not real-time processing.
• C. Correct. Spark Structured Streaming, available in services like Azure Databricks and Azure Synapse Analytics, is a scalable and fault-tolerant stream processing engine. It handles real-time data processing using either a micro-batch or continuous processing model, making it another excellent and powerful choice for implementing the Speed Layer.
• D. Incorrect. The Azure Data Factory Copy Activity is designed for orchestrating batch-oriented data movement and ETL/ELT workflows. It operates on a scheduled or triggered basis, moving chunks of data, and is not designed for the continuous, low-latency stream processing characteristic of the Speed Layer. It is typically used in the Batch Layer of a Lambda architecture.

Correct answers: A, B, C — Tumbling Window; Hopping Window; Sliding Window

• A. Correct. Tumbling windows segment a data stream into a series of fixed-size, non-overlapping, and contiguous time intervals. An event can only belong to one tumbling window. This function is useful for generating periodic reports, such as a count of events every 5 minutes.
• B. Correct. Hopping windows are used to model overlapping time windows. They are defined by a window size (duration) and a hop size (the period by which the window moves forward). Since windows can overlap, an event can belong to more than one hopping window. This is useful for creating moving averages or finding patterns over overlapping time periods.
• C. Correct. Sliding windows, unlike Tumbling or Hopping windows, produce an output only when an event occurs. The window 'slides' forward with each new event. Every window will have at least one event, and it groups events that fall within a specified time duration. This is useful for scenarios that require rolling aggregations where a calculation is needed whenever a new event arrives.
• D. Incorrect. 'Broken Window' is not a valid windowing function in Azure Stream Analytics. The standard windowing functions are Tumbling, Hopping, Sliding, and Session.
• E. Incorrect. 'Curtain Window' is not a recognized term or function within the Azure Stream Analytics query language. It is a fabricated option.

Correct answer: C — Paginated Report

• A. Incorrect. Power BI dashboards are single-page, tile-based canvases intended for on-screen monitoring and getting quick insights. They are not designed for multi-page, pixel-perfect layouts required for printing.
• B. Incorrect. Standard, interactive Power BI reports (.pbix files) are optimized for on-screen data exploration and interaction. Their layouts are responsive, which means they are not suitable for the strict, fixed-layout control needed for multi-page printed documents.
• C. Correct. Paginated reports, which use the RDL (Report Definition Language) file format, are specifically designed for creating pixel-perfect, fixed-layout reports that can span multiple pages. They are ideal for documents that will be printed or exported to PDF, such as invoices, statements, and catalogs.
• D. Incorrect. Scorecards in Power BI are used to track and monitor key performance indicators (KPIs), goals, and other metrics. They are not a tool for creating formatted, multi-page reports intended for physical printing.

Correct answers: A, B — Pin a visual from a report; Pin a live report page

• A. Correct. This is a primary method for building dashboards. In the Power BI service, you can select the pin icon on any visual within a report to add it as a tile to a dashboard. This tile serves as a snapshot of the data and links back to the source report.
• B. Correct. Pinning a live report page allows you to add an entire, interactive report page to a dashboard. Unlike pinning individual visuals, this method preserves the interactivity between visuals on the page (like cross-filtering and slicers) directly on the dashboard.
• C. Incorrect. You cannot directly import a visual from the Excel desktop application into a Power BI dashboard. While Power BI can connect to Excel data, and you can pin elements from Excel workbooks hosted in OneDrive for Business or SharePoint Online, the action of 'importing a visual' is not a standard feature.
• D. Incorrect. Power BI dashboards do not support pasting or copying visuals directly from external applications like Microsoft Word. Dashboard content must be generated from within the Power BI ecosystem, primarily from reports.
• E. Incorrect. This is not a feature in Power BI. While you can add web content, including videos from services like YouTube, to a dashboard using a dedicated web content tile, you cannot embed a video file directly into the header of another visual.

Correct answer: B — False

• A. The statement is false because Power BI dashboards are a feature exclusive to the Power BI service, which is the online platform. Power BI Desktop is the authoring tool used to create reports and data models. To create a dashboard, you must first publish a report from Power BI Desktop to the Power BI service and then pin visuals from the published report(s) onto a dashboard within the service.
• B. The statement is false because Power BI dashboards are a feature exclusive to the Power BI service, which is the online platform. Power BI Desktop is the authoring tool used to create reports and data models. To create a dashboard, you must first publish a report from Power BI Desktop to the Power BI service and then pin visuals from the published report(s) onto a dashboard within the service.

Correct answer: B — Azure Databricks

• A. Incorrect. Azure Functions is a serverless compute service for running event-driven code. It is not designed for managing or running large-scale, interactive Apache Spark clusters for data processing.
• B. Correct. Azure Databricks is an Apache Spark-based analytics platform optimized for the Microsoft Azure cloud services platform. It provides a collaborative, interactive workspace with notebooks and integrates seamlessly with other Azure services like Azure Data Lake Storage, making it the ideal choice for large-scale Spark workloads.
• C. Incorrect. Azure SQL Managed Instance is a fully managed Platform as a Service (PaaS) offering for relational database workloads (SQL Server). It is not designed for running Apache Spark clusters.
• D. Incorrect. Power BI is a business analytics and data visualization service. While it can connect to data processed by Spark, it is used for reporting and creating dashboards, not for performing the large-scale data processing itself.

Correct answer: A — Data Ingestion

• A. Data Ingestion is the correct phase. It is the process of moving data from numerous disparate sources, such as on-premises databases, SaaS applications, or IoT devices, into a centralized storage location like a data lake or cloud blob storage. This is the first critical step in an analytics pipeline.
• B. Data Visualization is the process of presenting data and insights in a graphical format, like charts, graphs, and dashboards. This phase occurs after the data has already been ingested, processed, and analyzed, making it easier for users to understand the findings.
• C. Data Analysis involves examining, querying, and applying algorithms to the data to extract meaningful insights and identify trends. This phase presumes the data has already been collected and stored in the analytics environment.
• D. Data Modeling is a design activity that defines the logical structure for storing data, including schemas, relationships, and constraints. It focuses on how data is organized for efficient querying, not the physical act of moving the data from its source.

Correct answers: A, B, C — Azure Data Factory (or Synapse Pipelines); Azure Data Lake Storage Gen2; Azure Synapse Analytics

• A. Correct. Azure Data Factory and its equivalent, Synapse Pipelines, are core components for data integration and orchestration in a modern data warehouse. They are used to create, schedule, and manage ETL/ELT (Extract, Transform, Load / Extract, Load, Transform) data-driven workflows to move and transform data from various sources to the analytics platform.
• B. Correct. Azure Data Lake Storage Gen2 serves as the scalable, secure, and cost-effective storage layer for a modern data warehouse. It is designed to store vast amounts of structured and unstructured data and features a hierarchical namespace, making it an ideal central repository (data lake) for raw and processed data.
• C. Correct. Azure Synapse Analytics is a key component, acting as the central analytics and serving layer. It's an integrated service that combines big data analytics (using Apache Spark) and data warehousing (using a massively parallel processing SQL engine) into a single unified experience, accelerating time to insight.
• D. Incorrect. Azure DNS is a hosting service for Domain Name System (DNS) domains. It is a fundamental networking and infrastructure service, not a component directly involved in the data processing, storage, or analytics layers of a modern data warehouse architecture.
• E. Incorrect. Azure Active Directory (Azure AD) is used for identity and access management in Azure. However, Azure AD Domain Services provides managed domain services like LDAP and Kerberos, typically for legacy applications. It is not a core component of a modern data warehouse architecture, which would use Azure AD for authentication and authorization.