DP-900 Data Formats and Data Stores Practice Questions
The DP-900 starts with how data is represented. You need to recognize structured, semi-structured and unstructured data from a short scenario, and know common file formats: delimited text such as CSV, JSON and XML, and optimized formats such as Avro, ORC and Parquet.
Expect questions on when a columnar format like Parquet is the better choice, and on the main types of databases: relational, and non-relational stores such as key-value, document, column-family and graph databases.
Below are 10 free questions on this topic. Try each one before you open the answer.
Question 1
A data engineering team needs to store large datasets for analytical queries that only access a subset of columns. Which file format is most appropriate?
A.CSV
B.JSON
C.Parquet
D.TXT
chevron_rightShow answer and explanation
Correct answer: C. Parquet
Parquet is a columnar storage format optimized for analytical queries that read only specific columns, making it highly efficient for large datasets. CSV and TXT are row-based flat formats with no compression optimization. JSON is semi-structured and not optimized for column-based analytical access.
Question 2
A company wants to store raw, unprocessed log files, sensor readings, and images for future analysis. Which data storage concept is most appropriate?
A.Data lake
B.Data mart
C.Data warehouse
D.Relational database
chevron_rightShow answer and explanation
Correct answer: A. Data lake
A data lake stores raw data in its native format — structured, semi-structured, and unstructured — at scale, making it ideal for future processing or exploratory analytics. A data warehouse stores processed, structured data optimized for analytical queries. A data mart is a focused subset of a data warehouse for a specific business area. A relational database is designed for transactional workloads, not raw file storage.
Question 3
Which columnar file format is heavily optimized for read performance in Apache Hive and Hadoop-based ecosystems?
A.Avro
B.CSV
C.JSON
D.ORC
chevron_rightShow answer and explanation
Correct answer: D. ORC
ORC (Optimized Row Columnar) is a columnar format designed for high read performance in the Hadoop ecosystem, particularly with Apache Hive. Avro is a row-based binary format used for serialization and streaming. CSV and JSON are text-based formats that lack the compression and columnar optimizations of ORC.
Question 4
A team loads raw CSV and JSON files from various sources directly into Azure Data Lake Storage Gen2, then uses Azure Synapse Analytics to apply transformations and build aggregated tables. Which data integration pattern does this describe?
A.ETL
B.ELT
C.Batch processing only
D.Streaming ingestion
chevron_rightShow answer and explanation
Correct answer: B. ELT
This scenario describes ELT (Extract, Load, Transform): raw data is first loaded into the storage layer (Azure Data Lake Storage Gen2), and transformation logic is applied afterward using the compute power of Azure Synapse Analytics. ELT is a common pattern in modern cloud data architectures where scalable compute makes in-place transformation practical and efficient.
Question 5
Which statement best defines metadata?
A.The actual business data stored in rows and columns of a database
B.Data that describes other data — such as column names, data types, creation dates, and data lineage
C.Binary data encoded for secure transmission over a network
D.Aggregated summary statistics generated from analytical queries
chevron_rightShow answer and explanation
Correct answer: B. Data that describes other data — such as column names, data types, creation dates, and data lineage
Metadata is "data about data." It provides descriptive, structural, or administrative information about data assets — for example, a table's column names and data types (structural metadata), when a file was created or modified (operational metadata), or what a field means in business terms (business metadata). Metadata does not contain the actual business values stored in the data.
Question 6
What does data lineage describe?
A.The physical location where data is stored across different geographic regions
B.The origin of data and the transformations it has undergone as it flows through systems over time
C.The compression algorithm used to reduce the size of stored data files
D.The schedule on which a data pipeline is triggered to run
chevron_rightShow answer and explanation
Correct answer: B. The origin of data and the transformations it has undergone as it flows through systems over time
Data lineage tracks where data comes from (its source), how it has been transformed, and where it has moved to — creating an end-to-end trace of data flow. Lineage is essential for troubleshooting data quality issues, understanding the impact of upstream changes, and meeting compliance requirements. Tools like Microsoft Purview automatically capture lineage across Azure services such as Data Factory and Synapse.
Question 7
Which three of the following are examples of unstructured data? Each correct answer presents a complete solution. *(meerdere antwoorden)*
A.Audio recordings of customer service calls
B.A relational table with customer orders
C.Scanned PDF documents of invoices
D.Social media posts written in natural language
chevron_rightShow answer and explanation
Correct answer: A. Audio recordings of customer service calls
Unstructured data has no predefined schema or organizational structure that makes it directly queryable. Audio files, scanned PDFs (which contain text as images), and free-text social media posts are all unstructured. A relational table with customer orders is structured data — it has a fixed schema with defined columns and data types. Organizations often use AI and machine learning to extract insights from unstructured data.
Question 8
A team is choosing a file format for storing large datasets in Azure Data Lake Storage Gen2 that will be queried frequently by Apache Spark jobs. They want efficient columnar reads, built-in compression, and schema enforcement. Which file format should they choose?
A.CSV
B.JSON
C.Parquet
D.XML
chevron_rightShow answer and explanation
Correct answer: C. Parquet
Parquet is a columnar file format designed for analytical workloads. It stores data column-by-column rather than row-by-row, which means Spark queries that read only specific columns can skip unneeded data entirely — dramatically reducing I/O. Parquet includes built-in compression (Snappy, GZIP) and encodes schema information in the file footer, enabling schema enforcement at read time. CSV is row-based, has no schema, no compression, and requires full file scans. JSON is also row-oriented and verbose — poor for large-scale analytics. XML is even more verbose and has no native support in most big data processing engines.
Question 9
Determine whether each statement about non-relational databases is correct.
1.Graph databases store entities as nodes with links to define relationships between them.
2.Column family databases store data as simple key-value pairs where each key maps to a single string value.
3.Document databases are a specific form of key-value database in which the value is a JSON document.
chevron_rightShow answer and explanation
Statement 1: Yes
Statement 2: No
Statement 3: Yes
Statement 1 is correct: graph databases use nodes and edges (links) to represent entities and their relationships. Statement 2 is incorrect: column family databases store tabular data organized into column families — groups of logically related columns. Key-value databases store simple key-value pairs, not column family databases. Statement 3 is correct: document databases extend the key-value model where the value is specifically a JSON document optimized for parsing and querying.
Question 10
The ___ file format is row-based, stores a JSON header to describe the binary data structure, and is particularly well suited for compressing data to minimize storage and network bandwidth.
A.Parquet
B.ORC
C.Avro
D.CSV
chevron_rightShow answer and explanation
Correct answer: C. Avro
Avro is a row-based format created by Apache that stores a JSON header followed by binary data, and is designed for efficient compression. Parquet and ORC are both columnar formats. CSV is a plain text delimited format with no compression optimizations.
These are 10 of 45 Data Formats and Data Stores questions in the full bank
Practice all 400 DP-900 questions in exam format, including drag-and-drop and multi-select questions, with a mock exam and progress tracking.