AZ-305 Question 76
Select 2A healthcare organization needs to store and analyze large volumes of medical images (unstructured data), logs from hospital devices (semi-structured data), and real-time IoT sensor data. They plan to use Azure Synapse Analytics for advanced analytics and machine learning. Which two design recommendations should you implement to handle both unstructured and semi-structured data in a cost-effective, scalable, and near real-time fashion?
- A
Store all data in a single Azure Cosmos DB container regardless of data type
- B
Store unstructured data (e.g., images and logs) in Azure Data Lake Storage Gen2 and ingest real-time sensor data into Azure Cosmos DB with autoscale enabled
- C
Directly stream all IoT sensor data into a single Azure SQL Database using built-in streaming
- D
Use Azure Files to handle all types of data and attach it to Synapse for analytics
- E
Enable Azure Synapse Link on Azure Cosmos DB to support near real-time analytics on sensor data
Show answer and explanation
Correct answers: B, E
Explanation
When storing unstructured data such as images and semi-structured data like device logs, Azure Data Lake Storage Gen2 provides scalable, cost-optimized storage. For real-time or near real-time data ingestion, Azure Cosmos DB is highly suited, especially with autoscale to handle variable throughput. Enabling Azure Synapse Link on Cosmos DB further allows near real-time analytics without a complex ETL pipeline, following Microsoft best practices for modern data warehousing and analytics.
- A. Incorrect.
Option 1 is incorrect because storing all data types, especially large unstructured files, into a single Cosmos DB container can become very costly and complex. Azure Cosmos DB is excellent for high throughput, low-latency scenarios, but large files like images are generally better suited for a data lake.
- B. Correct.
Option 2 is correct. Azure Data Lake Storage Gen2 provides cost-effective, scalable storage for unstructured and semi-structured data such as logs and images. Azure Cosmos DB with autoscale is well-suited for high-ingestion scenarios like IoT sensor data, allowing elastic scaling of throughput.
- C. Incorrect.
Option 3 is incorrect. Directly streaming large volumes of IoT data to a single Azure SQL Database can be cost-inefficient and difficult to scale for near-real-time ingestion. SQL database is typically not the best choice for time-series sensor data at massive scale.
- D. Incorrect.
Option 4 is incorrect. While Azure Files can store unstructured data, it does not seamlessly integrate with analytics tools for large-scale, near real-time processing the way Data Lake Storage Gen2 does. It also lacks the advanced features needed for analytics workloads in Azure Synapse.
- E. Correct.
Option 5 is correct. By enabling Azure Synapse Link on Azure Cosmos DB, you can run near real-time analytics on the sensor data without impacting the performance of the operational store.