DP-900 Practice Questions (2026): 10 Free Azure Data Fundamentals Questions with Answers

JSON is semi-structured: it uses tags/keys to describe structure but does not force every record into the same fixed schema, so products can have varying fields. Structured data (A) fits a rigid table of rows and columns. Unstructured data (C) has no organising schema at all — images, video, free-form text. Relational data (D) is a specific form of structured data stored in tables.

OLTP handles high volumes of small, fast read/write transactions with strong (ACID) consistency — exactly what order capture needs. OLAP (A) and data warehousing (B) are read-heavy systems that aggregate historical data for analysis, not for recording live transactions. Batch processing (D) collects data and processes it in scheduled bulk groups, not one transaction at a time.

The data engineer designs, builds, and operates the pipelines that ingest, clean, transform, and load data. The database administrator (A) owns database availability, security, backups, and performance. The data analyst (C) explores and visualises data to surface insights and build reports. A business user (D) consumes the reports rather than building the plumbing.

Azure SQL Managed Instance is a PaaS offering with near-complete SQL Server engine compatibility, including instance-scoped features such as SQL Server Agent and cross-database queries — ideal for lift-and-shift migrations. Azure SQL Database (A) is also PaaS but is scoped to a single database and lacks some instance-level features. SQL Server on a VM (C) is IaaS: full control, but you patch and manage the OS yourself. PostgreSQL (D) is a different database engine, not SQL Server compatible.

An index is a sorted structure that lets the engine find rows by a key (here CustomerID) without scanning the whole table, which speeds up filtered lookups transparently. A view (A) is a saved query definition — it does not by itself make the underlying search faster. A stored procedure (B) is reusable SQL logic. A trigger (D) runs automatically in response to data changes; it is not a performance object.

Azure Blob storage is built for massive amounts of unstructured object data — images, video, backups — accessed over HTTP/HTTPS, with hot, cool, cold, and archive access tiers to balance cost against access frequency. Table storage (A) is a NoSQL key-value store for structured non-relational data. Azure Files (B) provides managed SMB/NFS file shares. Queue storage (D) is a messaging service, not a file store.

The API for MongoDB implements the MongoDB wire protocol, so existing MongoDB drivers, SDKs, and tools keep working with little or no code change. The API for NoSQL (A) is the native Cosmos DB API but uses its own SDKs, so it would mean rewriting data access. The API for Cassandra (C) targets Cassandra/CQL workloads, and the API for Table (D) targets apps written for Azure Table storage.

A data lake stores large volumes of raw structured, semi-structured, and unstructured data cheaply in its native format, applying schema on read so many engines can consume it later. An OLTP database (A) and a single Azure SQL Database (C) enforce a fixed relational schema up front and are not cost-optimised for raw big data. Azure Cache for Redis (D) is an in-memory cache for fast key lookups, not bulk storage.

Power BI is Microsoft's business-intelligence platform for modelling data and building interactive reports and dashboards with filtering and drill-down. Azure Data Factory (A) orchestrates data movement and transformation, not visualisation. Azure Synapse Analytics (C) is a large-scale analytics and data-warehousing service that Power BI can sit on top of. Table storage (D) is a NoSQL data store.

Stream (real-time) processing handles data continuously as each event arrives, with very low latency — required to update a live map within seconds. Batch processing (A, C) collects data into groups and processes it on a schedule, which is high latency by design. OLTP (D) records individual transactions but is not the analytics model used for continuous telemetry streams.