# Free DP-900 practice questions

> 10 free DP-900 practice questions with the correct answer and an explanation.

Page: https://devopscodelab.com/exams/microsoft-dp-900/free-practice-questions/
Practice exams: https://devopscodelab.com/exams/microsoft-dp-900/

### Question 1 of 10

A globally replicated database lets a write in one region become visible in other regions after a short delay, and readers may briefly see older values. Which consistency model does this describe?

- A. Eventual consistency
- B. Immediate consistency
- C. Strong consistency
- D. Two-phase commit consistency

**Answer:** A

**Explanation:** With eventual consistency, replicas converge over time, so a read in another region may return a stale value until the write propagates. Strong or immediate consistency would guarantee that every read sees the latest committed write, and two-phase commit is a protocol that coordinates a transaction across participants so they commit together, which avoids rather than allows stale reads.

### Question 2 of 10

A shopping cart service only ever looks up a cart by its cart ID and replaces the whole cart value on every change. Which type of non-relational data store is the simplest fit?

- A. A relational database with foreign keys
- B. A column-family store
- C. A graph database
- D. A key-value store

**Answer:** D

**Explanation:** A key-value store retrieves and writes an opaque value by its unique key, which matches lookups and replacements by cart ID with very low latency. Graph databases model relationships between entities, column-family stores organize wide rows into column groups for large sparse datasets, and a relational database adds schema and join overhead that this access pattern does not need.

### Question 3 of 10

A reporting query sums one column across 500 million rows of a 60-column table. Why would a columnar store usually complete this query faster than a row store?

- A. It reads only the needed column's data, which is stored together and compresses well
- B. It stores each row as a JSON document, so the engine can read individual properties
- C. It enforces foreign keys between tables, which lets the optimizer skip unrelated rows
- D. It keeps every row in memory, so no data has to be read from disk at query time

**Answer:** A

**Explanation:** Columnar storage groups each column's values together, so an aggregate over one column reads a small fraction of the data, and similar values compress efficiently. Columnar stores do not require holding all rows in memory, foreign keys do not reduce the data scanned for a single-table aggregate, and storing rows as JSON documents describes a document store, not a columnar one.

### Question 4 of 10

A company is moving an on-premises SQL Server application to Azure. The application uses SQL Server Agent jobs and cross-database queries, and the team wants Microsoft to handle patching and backups. Which service should they choose?

- A. Azure SQL Managed Instance
- B. Azure SQL Database
- C. Azure Database for PostgreSQL flexible server
- D. SQL Server on Azure Virtual Machines

**Answer:** A

**Explanation:** Azure SQL Managed Instance is a platform service with near-complete SQL Server engine compatibility, including SQL Server Agent and cross-database queries, while Microsoft manages patching and backups. Azure SQL Database does not support SQL Server Agent or cross-database queries in the same way. A virtual machine supports everything but leaves patching and backup management to the customer, and PostgreSQL is a different engine that would require rewriting the application.

### Question 5 of 10

A startup has built its application on open-source PostgreSQL and wants a managed Azure service that handles backups, patching, and high availability without changing the database engine. Which service should it use?

- A. Azure Cosmos DB for NoSQL using a relational container
- B. Azure SQL Database with the PostgreSQL compatibility level
- C. Azure Database for PostgreSQL flexible server
- D. SQL Server on Azure Virtual Machines with a PostgreSQL driver

**Answer:** C

**Explanation:** Azure Database for PostgreSQL flexible server is a fully managed PostgreSQL service with automated backups, patching, and optional zone-redundant high availability. Azure SQL Database runs the SQL Server engine and has no PostgreSQL compatibility level, Cosmos DB for NoSQL is a document database, and a SQL Server virtual machine would require migrating to a different engine.

### Question 6 of 10 (choose 2)

A team plans to migrate an existing MongoDB application to Azure Cosmos DB with minimal code changes, and a second team needs to store a highly connected social network for graph traversals. Which two Azure Cosmos DB APIs should they use? (Choose two.)

- A. API for Apache Gremlin
- B. API for NoSQL with a relational schema
- C. API for MongoDB
- D. API for Apache Cassandra
- E. API for Table

**Answer:** A, C

**Explanation:** The API for MongoDB lets existing MongoDB applications use Cosmos DB through the MongoDB wire protocol with minimal changes, and the API for Apache Gremlin stores and traverses graph data such as social networks. The Table API targets key-value workloads migrated from Azure Table storage, the Cassandra API serves wide-column Cassandra workloads, and the NoSQL API is a document API that does not use a relational schema.

### Question 7 of 10

Branch offices keep using local Windows file servers for fast access, but the company wants the files centralized in Azure with cloud tiering of rarely used files. Which solution meets this requirement?

- A. A Data Lake Storage account with one container per branch
- B. Blob Storage lifecycle management moving files to the archive tier
- C. An Azure Data Factory pipeline that copies the files to Blob nightly
- D. Azure File Sync with an Azure file share as the cloud endpoint

**Answer:** D

**Explanation:** Azure File Sync synchronizes Windows file servers with an Azure file share and can tier infrequently accessed files to the cloud while keeping hot files cached locally. Lifecycle management only tiers blobs already in Azure, a nightly copy pipeline provides no local cache or tiering, and a Data Lake Storage account with per-branch containers stores data in the cloud but does not sync with or cache for file servers.

### Question 8 of 10 (choose 2)

A company is choosing between ETL and ELT for loading data into its analytical store. Which two statements about ELT are correct? (Choose two.)

- A. Raw data is loaded into the target store first and transformed there
- B. ELT relies on the compute power of the target store or lake to run transformations
- C. ELT prevents analysts from keeping a copy of the original raw data
- D. ELT requires that all transformations finish before data leaves the source system
- E. ELT can only be used when the target is an on-premises relational database

**Answer:** A, B

**Explanation:** In ELT, data is extracted and loaded in raw form, then transformed inside the target using that platform's scalable compute, such as a lakehouse or warehouse engine. Transforming before data reaches the target describes ETL, ELT is common with cloud data lakes rather than being limited to on-premises databases, and loading raw data first actually makes it easier to retain the original data.

### Question 9 of 10

Which statement correctly describes how Power BI Desktop and the Power BI service are typically used together?

- A. Power BI Desktop stores data for the organization, and the service is used only to design visuals
- B. Reports are authored in Power BI Desktop and then published to the Power BI service for sharing
- C. Power BI Desktop is used only on mobile devices, while the service runs only on Windows computers
- D. Reports are authored in the Power BI service and then downloaded to Power BI Desktop to be viewed

**Answer:** B

**Explanation:** Power BI Desktop is the authoring tool for connecting to data, modeling, and building reports, which are then published to the cloud-based Power BI service where users view, share, and collaborate. The service is the sharing and consumption platform rather than a download source, Desktop is not a storage service, and mobile access is provided by the Power BI mobile apps.

### Question 10 of 10

A Stream Analytics job must report the count of transactions for each fixed, non-overlapping five-minute interval, with every event counted exactly once. Which window type should the query use?

- A. A sliding window
- B. A tumbling window
- C. A session window
- D. A hopping window

**Answer:** B

**Explanation:** Tumbling windows are fixed-size, contiguous, non-overlapping intervals, so each event belongs to exactly one window. Hopping windows overlap when the hop is shorter than the window size, sliding windows produce output when events enter or leave and can overlap, and session windows group events separated by gaps of inactivity rather than fixed intervals.

Ready for a timed run? Take the free 15-question DP-900 mini exam: https://devopscodelab.com/app/exams/microsoft-dp-900/mini/

DevOpsCode Lab is not affiliated with Microsoft. DP-900 and related names are trademarks of their owners. We write every question ourselves; these are not real exam questions.
