DP-900 Practice Test

Sample DP-900 Practice Questions

10 real questions from the free DP-900bank. Pick an answer in your head, then open “Show answer” for the explanation.

1. 1. A multinational company stores employee records and needs guaranteed data consistency, enforced cross-record relationships, and complex join query support. Which data model is most appropriate?

   • A. Relational database with normalized tables and defined primary/foreign key relationships
   • B. Document database with JSON records for each employee
   • C. Key-value store with employee ID as the key
   • D. Graph database with employee nodes and relationship edges
   Show answer

   Correct answer: A

   A relational database is the best choice when you need referential integrity (enforced via foreign keys), guaranteed consistency (ACID transactions), and complex JOIN queries. These are defining features of the relational model.

2. 2. JSON and XML are both examples of which type of data?

   • A. Structured data, because they conform to formal syntax rules with a rigid predefined schema applied consistently across the data store.
   • B. Semi-structured data, because they have self-describing organizational elements (JSON keys, XML tags) but no rigid schema enforcement.
   • C. Unstructured data, because they use plain text encoding without inherent key-value or hierarchical organizational structure.
   • D. Binary data, because they are encoded in a machine-readable format that requires a specific parser to decode the content.
   Show answer

   Correct answer: B

   JSON and XML are semi-structured. They have organizational elements (tags in XML, key-value pairs in JSON) but do not enforce a rigid schema where every record must have the same fields. This flexibility distinguishes them from fully structured relational data.

3. 3. A company needs to analyze customer sentiment from social media posts, emails, and recorded phone calls. Which data processing challenge is unique to this unstructured data scenario?

   • A. Standard SQL queries on the raw data files require pre-processing to remove stop words and normalize character encoding before any analysis can begin.
   • B. Standard SQL queries are insufficient for extracting sentiment directly, since the unstructured text must first be flattened into a normalized columnar format.
   • C. Extracting meaningful insights requires Natural Language Processing for text and speech-to-text for audio, rather than standard SQL queries on raw content.
   • D. Sentiment signals are available from text-based sources, but images and audio require conversion to a structured format before any signals can be extracted.
   Show answer

   Correct answer: C

   Unstructured data like text and audio requires specialized processing techniques to extract insights. NLP is needed for text sentiment analysis; speech-to-text is needed for audio. Unlike structured data, you cannot simply write a SQL query to extract sentiment.

4. 4. Which of the following statements correctly describes the ORC (Optimized Row Columnar) file format?

   • A. ORC is a columnar format with efficient compression, optimized for Hive and Hadoop analytics
   • B. ORC stores data in a row-based format optimized for transactional write performance and OLTP workloads
   • C. ORC is a text-based columnar format similar to CSV but with embedded column type definitions and schema
   • D. ORC is a real-time streaming format for IoT data that supports exactly-once message delivery guarantees
   Show answer

   Correct answer: A

   ORC (Optimized Row Columnar) is a columnar storage format that provides highly efficient compression and encoding schemes. It was designed for Apache Hive workloads in the Hadoop ecosystem, offering better performance than plain text formats for large-scale analytics.

5. 5. A streaming service stores millions of video files, subtitle files, and thumbnail images. How should this content data be classified?

   • A. Semi-structured because the files have associated metadata records
   • B. Structured because the files are organized in folder hierarchies
   • C. Transactional because the files represent business assets
   • D. Unstructured data because the primary content (video bytes and image pixels) has no predefined organizational schema.
   Show answer

   Correct answer: D

   Video files, subtitle files, and images are unstructured data. While associated metadata (filename, upload date, duration) may be structured, the actual file content has no queryable schema - this makes the files themselves unstructured. Option A is incorrect because having metadata records does not make the content semi-structured; the classification applies to the data type being stored, not its metadata.