Microsoft AZ-104 - Microsoft Azure Administrator - Azure Data Concepts - Review Questions

Azure Data Concepts, Tools, and Processes — 15 Review Questions

1. An online store must record orders, update inventory, and process payments through many short transactions. Which workload type best describes this requirement?

A. OLAP

B. OLTP

C. Data cataloging

D. Historical aggregation

2. A company needs to calculate sales trends across five years of records from multiple business systems. Which workload type is most appropriate?

A. OLTP

B. File synchronization

C. OLAP

D. Resource provisioning

3. A pipeline extracts records, loads them into a data lake, and then transforms them. Which processing pattern does this represent?

A. ELT

B. ETL

C. OLTP

D. ACID

4. You need to schedule data ingestion, run a Databricks notebook after a successful copy, and follow an error-handling path if processing fails. Which service is designed to coordinate these activities?

A. Azure Storage Explorer

B. Power BI

C. Microsoft Purview Data Map

D. Azure Data Factory

5. Device messages arrive continuously through IoT Hub. You need to calculate one-minute temperature averages and write results to a data lake. Which service is the best fit?

A. SQL Server Management Studio

B. Azure Stream Analytics

C. Azure Resource Manager

D. Azure Container Registry

6. Why might an organization retain raw data after creating a cleaned, curated dataset?

A. Raw data automatically prevents duplicate pipeline runs.

B. Curated datasets cannot be stored in a data lake.

C. Raw data can be reprocessed for new business rules or future use cases.

D. Retaining raw data removes the need for access controls.

7. You need complex joins and transformations across large datasets using Python and distributed Spark processing. Which option is most appropriate?

A. Azure Databricks

B. Azure Storage Explorer

C. Azure Resource Manager

D. Microsoft Purview Unified Catalog

8. In a pipeline where curated lake files are loaded into Azure Synapse Analytics warehouse tables, which component is the Load destination?

A. Synapse Studio

B. Azure Data Factory trigger

C. Synapse serverless SQL pool

D. Synapse dedicated SQL pool

9. Which statement correctly distinguishes Synapse serverless SQL from a dedicated SQL pool?

A. Serverless SQL requires loading every file into warehouse tables first.

B. Serverless SQL can query supported lake files directly; a dedicated SQL pool provides provisioned warehouse compute and tables.

C. A dedicated SQL pool cannot load data from ADLS Gen2.

D. Serverless SQL is a replacement for Blob Storage.

10. An application requires access to the guest operating system and additional software installed alongside SQL Server. Which Azure deployment option best meets this requirement?

A. Azure SQL Database single database

B. Azure SQL Database elastic pool

C. SQL Server on Azure Virtual Machines

D. Azure SQL Managed Instance

11. An application stores product photographs and flexible JSON product records. Which design is most appropriate?

A. Store photographs in Blob Storage and product records in Cosmos DB.

B. Store photographs in Azure Resource Manager and product records in Activity Log.

C. Store both exclusively in a Purview catalog.

D. Store product records in function keys and photographs in SQL role assignments.

12. A DBA needs detailed Microsoft SQL database administration, while another user needs to browse and upload ADLS Gen2 files. Which tool pairing is most appropriate?

A. Power BI for administration; SSMS for file uploads

B. Storage Explorer for administration; Purview for file uploads

C. AzCopy for administration; Power BI for file uploads

D. SSMS for administration; Storage Explorer for file access

13. A sales report shows that revenue declined. An analyst filters results by region and product to investigate contributing factors. Which type of analysis is this?

A. Descriptive

B. Diagnostic

C. Predictive

D. Prescriptive

14. A file upload must start custom validation and transformation code. Which architecture best fits this requirement?

A. A Purview classification directly modifies the file.

B. A Power BI visual executes the transformation on upload.

C. A BlobCreated event is routed through Event Grid to an Azure Function.

D. An Azure role assignment automatically starts a Databricks notebook.

15. You need to discover data assets, identify sensitive data categories, and trace supported pipeline relationships without moving the source datasets into a central warehouse. Which service should you use?

A. Microsoft Purview

B. Azure Resource Manager

C. Azure Storage Explorer

D. Azure Functions

Answer Key and Explanations

1. B — OLTP

Online Transaction Processing supports operational applications that perform many short reads and writes. Examples include recording orders, updating inventory, and transferring funds. These workloads prioritize reliable transactions, concurrency, and fast responses. Relational transactional systems commonly provide ACID guarantees.

2. C — OLAP

Online Analytical Processing supports complex queries and aggregations over large datasets, often including historical information from several sources. It helps answer questions such as how sales changed over time. OLAP is read-heavy, but not necessarily read-only: analytical data still needs to be loaded and maintained.

3. A — ELT

Extract → Load → Transform means data is loaded into the destination platform before transformation. In ETL, transformation occurs before loading into the target. A larger architecture can combine both patterns—for example, landing raw files in a lake, transforming them, and then loading prepared results into a warehouse.

4. D — Azure Data Factory

Data Factory provides pipeline orchestration. It can schedule runs, copy data, invoke processing services, pass parameters, and control dependencies and failure paths. It coordinates the workflow while the selected processing engine—such as Databricks or a mapping data flow—performs the transformation.

5. B — Azure Stream Analytics

Stream Analytics processes arriving events using queries that can filter, transform, join, and aggregate data over time windows. It can read IoT Hub messages and write results to supported destinations such as ADLS Gen2. Window duration and event-ordering policies influence when results become available.

6. C — Reprocess data for new requirements

Retained raw data provides a starting point when business rules change or a new use case requires details absent from an earlier curated dataset. It can also support correction and historical reprocessing. Retention must still follow privacy, security, cost, and lifecycle requirements; retaining data does not automatically prevent duplicate processing.

7. A — Azure Databricks

Databricks supports distributed data processing using Apache Spark and development tools such as Python/PySpark and SQL. It is suitable for complex joins, cleaning, enrichment, and large-scale transformations. Storage Explorer manages files, while Purview catalogs and governs metadata rather than executing these transformations.

8. D — Synapse dedicated SQL pool

A dedicated SQL pool can receive prepared data into analytical warehouse tables. Loading methods include the T-SQL COPY statement and PolyBase-based approaches. Synapse Studio is the interface, while Data Factory can orchestrate the load. Synapse’s broader capabilities can also participate in other pipeline stages.

9. B — Direct lake querying versus provisioned warehousing

Serverless SQL can query supported files in the data lake without first copying them into dedicated warehouse tables. A dedicated SQL pool provides provisioned analytical compute and stores warehouse tables. Choose according to query patterns, performance requirements, data organization, and cost.

10. C — SQL Server on Azure Virtual Machines

SQL Server on Azure VMs provides guest OS and database-engine control, including the ability to install additional software. The customer consequently has greater responsibility for patching, configuration, backup, and availability. PaaS options reduce infrastructure administration but do not provide equivalent OS access.

11. A — Blob Storage plus Cosmos DB

Blob Storage is appropriate for files and objects such as photographs. Cosmos DB can store queryable JSON product records with varying attributes. The records can contain references to the photographs. This separates object storage from application data access and querying.

12. D — SSMS and Storage Explorer

SQL Server Management Studio supports detailed administration and querying of Microsoft SQL database engines. Storage Explorer provides graphical access to storage containers, files, and directories. Browsing lake files is different from analyzing their contents through SQL or Spark. Both tools still require permissions and network connectivity.

13. B — Diagnostic

Diagnostic analysis investigates why something happened by exploring contributing factors. Descriptive analysis summarizes the decline; predictive analysis estimates future results; prescriptive analysis compares possible actions. A relationship observed in a report does not, by itself, prove causation.

14. C — Event Grid and Azure Functions

A storage event can notify an Event Grid-triggered function that a file was created. The function then reads the actual file using authorized access, validates or transforms it, and writes results. The notification is not the entire file payload. Design processing to handle retries and duplicate delivery safely.

15. A — Microsoft Purview

Purview Data Map discovers metadata and classifications across supported sources, while catalog capabilities help users find and understand assets. Supported integrations can capture lineage from data pipelines. Coverage varies by connector and file type. Cataloging data does not automatically grant source access, encrypt it, or replace the source’s security controls.