AZ-305 Question 107
Single answerA large retail chain wants to analyze both historical point-of-sale (POS) data and near-real-time promotional campaigns. They need a scalable, cost-effective, and flexible Azure solution that can handle data ingestion, orchestration, data transformation, and advanced analytics. Which of the following approaches best meets these requirements?
- A
Implement Azure HDInsight with Hadoop clusters to store all data and run MapReduce jobs for daily analytics.
- B
Combine Azure Data Factory for orchestration, Azure Databricks for data transformation, and Azure Synapse Analytics for scalable data warehousing and advanced analytics.
- C
Use only Azure Stream Analytics to process both near-real-time data and historical batch data in a single pipeline.
- D
Store all data in Azure SQL Database and rely on vertical scaling to handle both OLTP and advanced analytics workloads.
Show answer and explanation
Correct answer: B
Explanation
The recommended approach for a scenario requiring combined batch and real-time analytics at scale is to orchestrate data flows with Azure Data Factory, transform data with Azure Databricks, and manage large-scale analytical queries in Azure Synapse Analytics. This pattern follows Microsoft best practices for modern data warehousing and analytics (see Microsoft documentation: https://learn.microsoft.com/azure/architecture/data-guide/relational-data/modern-data-warehouse). The combination of these services supports both day-to-day operational reporting and near-real-time streams for advanced analytics, making it a comprehensive and scalable solution.
- A. Incorrect.
Option 1 (Azure HDInsight with Hadoop) is a viable big data platform, but it can be more complex to maintain than other services and does not natively provide the same level of integration with advanced analytics and orchestration tools as newer Azure services like Synapse and Databricks. While it might handle large volumes, it’s not always the most efficient or cost-effective choice for both real-time and batch analytical workloads.
- B. Correct.
Option 2 (Azure Data Factory + Azure Databricks + Azure Synapse) is correct. Azure Data Factory can orchestrate both batch and streaming data pipelines, Azure Databricks can process and transform data at scale (including ML workloads if needed), and Azure Synapse offers a powerful data warehouse and analytical engine. This combination provides a comprehensive solution for ingesting, storing, transforming, and analyzing data from multiple sources in both real-time and batch scenarios.
- C. Incorrect.
Option 3 (Using only Azure Stream Analytics) focuses primarily on near-real-time event processing. Stream Analytics can handle streaming data well, but it lacks native capabilities for large-scale historical data transformation and comprehensive data warehousing. You would still need additional services like Azure Data Lake Storage and an analytics engine for in-depth, large-scale batch analytics.
- D. Incorrect.
Option 4 (Storing all data in Azure SQL Database) is typically best for OLTP or smaller analytics scenarios. Although you can scale up Azure SQL Database, it can become cost-prohibitive at higher data volumes and lacks the specialized features of an enterprise-scale warehouse or a distributed processing engine. It is not designed to handle massive batch data transformations and advanced analytics as efficiently as Azure Synapse or Databricks.