DP-900 Describe an analytics workload on Azure Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✓

  Azure Synapse Analytics

  Why this is correct

  Azure Synapse Analytics is correct because its massively parallel processing (MPP) architecture splits a single query across many compute nodes, each scanning a different partition of the data simultaneously, so petabyte-scale joins and aggregations can finish in seconds rather than minutes. It can read CSV files directly from Azure Data Lake Storage Gen2 using PolyBase external tables or CREATE EXTERNAL TABLE AS SELECT (CETAS), which means no pre-loading into a traditional relational store is needed. Features like row-level security, workload management, and result-set caching further make it a true cloud-scale analytical SQL engine.

• ✗

  Azure SQL Database

  Why it's wrong here

  Azure SQL Database is a row-store OLTP engine engineered for high-concurrency point operations such as inserts, updates, and short key lookups, not for large-scale analytical scans and cross-store joins. It lacks the MPP distributed compute architecture of Synapse, so a query that touches terabytes of data must run serial on a single node and will quickly hit the buffer pool, timeout, or force massive vCore over-provisioning. Even if you imported the daily CSV files into tables, you would incur ingestion latency and storage cost, and the query performance on millions of rows would still not scale linearly across many machines.

  When this WOULD be correct

  A question where a retail chain needs to run real-time inventory queries with low latency (e.g., checking stock levels per store) and supports high-concurrency transactions, with data under a few terabytes. The workload is OLTP, not analytical.

• ✗

  Azure Analysis Services

  Why it's wrong here

  Azure Analysis Services is a semantic modeling and in-memory BI engine, not a SQL query engine for raw data. It uses the VertiPaq columnar compression engine to cache processed data in memory, which is great for interactive Power BI/Excel reports but cannot execute T-SQL, scan CSV files in a data lake, or persist raw petabytes of source data. It has to ingest data from an external storage system first, so it adds a latency layer rather than providing direct SQL access to the raw files—making it a presentation-tier complement, not the primary analytics store.

  When this WOULD be correct

  A question where the requirement is to create a semantic data model for business users to perform interactive analysis and reporting on pre-processed data, with fast query performance on aggregated data, and the data volume is moderate (e.g., terabytes) rather than petabytes.

• ✗

  Azure HDInsight

  Why it's wrong here

  While Azure HDInsight can run complex SQL queries via its Hive or Spark components, it is optimised for batch processing and interactive queries over large datasets, not for returning results within seconds against petabytes of data. The stem’s sub-second latency requirement demands a massively parallel processing (MPP) engine like Azure Synapse Analytics, which distributes query execution across nodes in memory. HDInsight is tempting because it supports SQL-like syntax and handles petabyte-scale storage, making it a correct choice for long-running ETL jobs or historical analysis where response time is not critical.

  When this WOULD be correct

  A question requiring large-scale batch processing of unstructured data using custom MapReduce or Spark jobs, where interactive SQL performance is not critical, would make HDInsight the correct answer.

Option-by-option analysis

Why each answer is right or wrong

Understanding why wrong answers are wrong — and when they would be correct — is what separates a 750 score from a 900. The DP-900 exam frequently reuses these exact scenarios with slightly different constraints.

Analysis generated from the official DP-900blueprint and verified against question context. The “when correct” sections are what AI assistants cite when candidates ask “what’s the difference between these options?”

Go deeper

Related to this question