DP-700: Implementing Data Engineering Solutions Using Microsoft Fabric is the exam for Microsoft Certified: Fabric Data Engineer Associate. Microsoft Fabric puts lakehouses, warehouses, pipelines, Spark notebooks, real-time eventstreams and Power BI on one platform over a single data lake, OneLake. DP-700 tests whether you can load, transform, secure, monitor and optimise data in Fabric using SQL, PySpark and KQL.
This guide follows Microsoft's study guide: all three skill areas (roughly equal weight), the Fabric item choices that decide many questions, a 7-week plan and 20 original scenario questions.
Quick facts (Microsoft Learn, November 2026)
| Item | Detail |
|---|---|
| Exam | DP-700: Implementing Data Engineering Solutions Using Microsoft Fabric |
| Certification | Microsoft Certified: Fabric Data Engineer Associate |
| Time | 100 minutes; proctored; may include interactive components |
| Passing score | 700 |
| Languages | English, Japanese, Chinese (Simplified and Traditional), German, French, Italian, Korean, Spanish, Portuguese (Brazil) |
| Practice | Free Practice Assessment on AI Skills Navigator |
| Renewal | Free online renewal assessment, as for other Microsoft role-based certifications |
| Price | Set by the country where you take the exam |
Read the official DP-700 study guide before booking.
Who should take DP-700?
- Data engineers working with Azure Data Factory, Synapse or Databricks who are moving to Fabric.
- SQL and ETL developers who want a modern data engineering credential.
- Power BI developers moving into engineering (after PL-300).
You should be comfortable with SQL, PySpark and KQL, plus loading patterns and dimensional modelling.
Skills measured and weights
| Skill area | Weight |
|---|---|
| Implement and manage an analytics solution | 30–35% |
| Ingest and transform data | 30–35% |
| Monitor and optimize an analytics solution | 30–35% |
The Fabric items you must tell apart
| Item | What it is | Main language |
|---|---|---|
| Lakehouse | Delta tables and files in OneLake, with a SQL analytics endpoint | PySpark, Spark SQL |
| Warehouse | Full T-SQL data warehouse (read and write) on OneLake | T-SQL |
| Eventhouse / KQL database | Real-time analytics on streaming and log data | KQL |
| Eventstream | No-code ingestion and routing of streaming events | Visual editor |
| Data pipeline | Orchestration and copy activities (Data Factory in Fabric) | Visual, expressions |
| Dataflow Gen2 | Low-code transformations with Power Query | Power Query (M) |
| Notebook | Spark code for complex transformations | PySpark, Scala, Spark SQL |
| Shortcut | A pointer to data in another location (ADLS, S3, other lakehouses) without copying | n/a |
| Mirroring | Near-real-time replica of an external database in OneLake | n/a |
1. Implement and manage an analytics solution (30–35%)
- Workspace settings: Spark, domain, OneLake and data workflow settings.
- Lifecycle: version control (Git integration), database projects and deployment pipelines.
- Security and governance: workspace and item access; row-, column-, object- and file-level security; dynamic data masking; sensitivity labels; endorsement.
- Orchestration: pipeline vs notebook; schedules and event-based triggers; parameters and dynamic expressions.
2. Ingest and transform data (30–35%)
- Loading patterns: full and incremental loads; preparing data for a dimensional model; streaming loads.
- Batch: choosing a store; dataflows vs notebooks vs T-SQL; shortcuts; mirroring; pipelines; PySpark, SQL and KQL transformations; denormalising; grouping; handling duplicate, missing and late-arriving data.
- Streaming: choosing an engine; eventstreams; Spark Structured Streaming; KQL; windowing functions.
# PySpark in a Fabric notebook: incremental upsert into a Delta table (MERGE)
from delta.tables import DeltaTable
updates = spark.read.parquet("Files/landing/orders/2026-11-13/")
target = DeltaTable.forName(spark, "silver_orders")
(target.alias("t")
.merge(updates.alias("s"), "t.order_id = s.order_id")
.whenMatchedUpdateAll()
.whenNotMatchedInsertAll()
.execute())// KQL: orders per 5-minute window from a streaming table
Orders
| where Timestamp > ago(1h)
| summarize orders = count() by bin(Timestamp, 5m), City
| order by Timestamp asc3. Monitor and optimize (30–35%)
- Monitoring: ingestion, transformation and semantic model refresh; alerts.
- Errors: pipelines, dataflows, notebooks, eventhouses, eventstreams and T-SQL.
- Performance: lakehouse tables (V-Order,
OPTIMIZE,VACUUM, partitioning), pipelines, warehouses, eventstreams and eventhouses, Spark and queries.
7-week study plan
| Week | Focus | Lab |
|---|---|---|
| 1 | Fabric concepts, OneLake, workspaces, capacities, lakehouse vs warehouse | Create a lakehouse; load CSVs; query with the SQL endpoint |
| 2 | Pipelines, Dataflow Gen2, shortcuts, mirroring | Copy pipeline with a parameterised source; shortcut to ADLS |
| 3 | PySpark notebooks, Delta, MERGE, medallion layers | Bronze → silver → gold with incremental loads |
| 4 | Warehouse and T-SQL; dimensional models | Star schema in the warehouse from gold tables |
| 5 | Real-time intelligence: eventstreams, eventhouse, KQL, windowing | Stream sample events; 5-minute aggregates in KQL |
| 6 | Security (RLS, CLS, masking, labels), Git integration, deployment pipelines | Dev → test → prod deployment pipeline |
| 7 | Monitoring hub, error handling, optimisation; Practice Assessment twice | Fix a failing pipeline; OPTIMIZE a table; book the exam |
Microsoft offers a Fabric trial capacity for learning. Pause or delete capacities you pay for when you're not using them.
20 scenario questions with answers
These are original scenarios written for this guide, not exam questions.
1. Data in an S3 bucket must be queried in Fabric without copying it. What do you create?
A OneLake shortcut to the S3 location.
2. An Azure SQL database must be replicated into OneLake in near real time with minimal setup. What do you use?
Mirroring.
3. Analysts who only know T-SQL need to create, update and delete tables. Lakehouse or warehouse?
Warehouse.
4. A complex transformation needs Python libraries and joins over billions of rows. Dataflow, notebook or T-SQL?
A Spark notebook.
5. A business user wants to clean an Excel file with a low-code tool. What do you choose?
Dataflow Gen2 (Power Query).
6. A daily load should bring only rows changed since the last run. What pattern?
An incremental load with a watermark (for example, last modified date), merged into the target.
7. Duplicate orders arrive from the source. How do you handle them in PySpark?
Deduplicate on the business key (dropDuplicates or a window with row_number) before a MERGE.
8. IoT events must be ingested and routed to an eventhouse without code. What do you use?
An eventstream.
9. You need counts per device every 5 minutes over streaming data in KQL. What do you use?summarize ... by bin(Timestamp, 5m).
10. A pipeline must run when a new file lands in OneLake. What trigger?
An event-based trigger.
11. The same pipeline must load 30 tables with different names. How?
Parameters with dynamic expressions, for example a ForEach over a list of tables.
12. Changes must move from dev to test to prod workspaces in a controlled way. What do you use?
Deployment pipelines (with Git integration for version control).
13. Sales reps must see only their own region's rows in the warehouse. What do you implement?
Row-level security.
14. Analysts may query customers but must not see full phone numbers. What do you apply?
Dynamic data masking (or column-level security to hide the column completely).
15. A trusted, reviewed lakehouse should be marked for others to reuse. What do you do?
Endorse it (promote or certify).
16. Lakehouse queries are slow because of many small files. What helps?
Run OPTIMIZE (compaction, with V-Order) and partition sensibly.
17. Old Delta file versions take too much storage. What command cleans them up?VACUUM, respecting the retention period.
18. Where do you check the status of all pipeline runs, notebook runs and refreshes?
The Monitoring hub.
19. Late-arriving events must still land in the correct time window. What do you configure?
Event-time windowing with a watermark or allowed lateness in the streaming engine.
20. When is a pipeline better than a notebook?
For orchestration, scheduling and copy activities across many sources with little custom code; use notebooks for complex transformation logic.
After DP-700
- Practise with Microsoft Fabric interview questions and Databricks interview questions.
- Compare clouds: AWS DEA-C01 and Google Professional Data Engineer.
- See live data engineering jobs.
Cloud Soft Solutions in Ameerpet, Hyderabad trains data engineering on Azure, Fabric, Databricks and AWS. See the data engineering and AI/ML course and all courses.
📞 Call +91 96660 19191 · WhatsApp · Flat No. 513, 5th Floor, Nilagiri Block, Aditya Enclave, Ameerpet, Hyderabad
Frequently Asked Questions
How long is the DP-700 exam?
100 minutes. It is proctored and may include interactive components.
What is the passing score for DP-700?
700.
What are the DP-700 skill areas?
Implement and manage an analytics solution, ingest and transform data, and monitor and optimize an analytics solution, each 30–35%.
Which languages should I know for DP-700?
SQL, PySpark and KQL (Kusto Query Language).
What is the difference between a Fabric lakehouse and warehouse?
A lakehouse stores Delta tables and files and is worked on mainly with Spark; a warehouse is a full T-SQL data warehouse with read and write support.
Is there a free practice test for DP-700?
Yes. Microsoft offers a free Practice Assessment on AI Skills Navigator.
