All 33 lessons
Start here: cloud and data basics
- 1. Start here: the exam, the plan and what you already know
- 2. Cloud in plain words: SaaS, PaaS, IaaS and where Fabric fits
- 3. Data basics: files, tables, formats and the lakehouse idea
- 4. A tour of Microsoft Fabric: OneLake, workspaces and items
- 5. The three languages: SQL, PySpark and KQL side by side
- 6. Getting data in: the map of Fabric tools
Store, ingest and transform batch data (exam area 2, 30-35%)
- 7. Choosing a data store and a transformation tool
- 8. Shortcuts and mirroring: use data without copying it
- 9. Ingest data with pipelines and Copy job
- 10. Transform with PySpark and SQL: join, denormalize, aggregate
- 11. Messy data: duplicates, missing values and late arrivals
- 12. Full and incremental loads, and loading a star schema
Orchestrate and stream (exam areas 1 and 2)
Set up and secure a Fabric solution (exam area 1, 30-35%)
- 18. Workspace settings: Spark, domains, OneLake and data workflows
- 19. Version control with Git and database projects
- 20. Deployment pipelines: development, test and production
- 21. Who can see what: workspace roles and item permissions
- 22. Fine-grained security: rows, columns, objects, files and masking
- 23. Governance: sensitivity labels and endorsement
Monitor, fix and optimize (exam area 3, 30-35%)
- 24. Monitoring hub, alerts and Activator
- 25. Fixing errors in pipelines, dataflows and notebooks
- 26. Fixing errors in Eventhouse, Eventstream and T-SQL
- 27. Optimize lakehouse tables: V-Order, OPTIMIZE, VACUUM and partitions
- 28. Optimize Spark, pipelines and warehouses
- 29. Optimize eventstreams, eventhouses and queries
Exam practice and final review
15 min
Getting data in: the map of Fabric tools
Know every way data gets into Fabric and the one question that picks each tool.
By the end of this lesson you can
- Tell the tools that copy data from the ones that leave it in place
- Sort ingestion tools into low-code and code-first
- Pick the tool for streaming events
- Use a three-question map to choose an ingestion tool
From earlier lessons · 1 of 2
In a Fabric notebook, which line saves the DataFrame df as a Delta table, replacing any old data?
Pick one answer.
From earlier lessons · 2 of 2
Which KQL step groups rows and counts them in one go?
Pick one answer.
Northwind Rides has four kinds of data to bring into Fabric. Scooters send trip events every few seconds. Customers and payments sit in a SQL Server database. A partner keeps station maintenance CSVs in its own Amazon S3 bucket. Fabric offers more than six ways in, and your manager wants a plan by Friday: which source gets which tool?
Question 1: copy the data, or leave it where it is?
Exam depth: know it cold.
Most tools copy data. They read it from a source and write a new copy into OneLake. Two features skip the copy job you'd otherwise build.
- A shortcut is a pointer in OneLake to data stored elsewhere, such as Azure Data Lake Storage Gen2 (Azure's file storage for analytics), Amazon S3 or another workspace. Nothing is copied. Queries read the data where it lives.
- Mirroring keeps a copy of an outside database in OneLake as Delta tables, in near real time. You build nothing. Fabric watches the source for changes and applies them.
So mirroring usually does move data; it just does it for you, all the time. A few sources, such as Azure Databricks, are mirrored as metadata only, through shortcuts. Lesson 8 covers that.
| Tool | Copies data? | Who runs it? |
|---|---|---|
| Pipeline (Copy activity) | Yes. | You build and schedule it. |
| Copy job | Yes. | You set it up in a wizard and schedule it. |
| Dataflow Gen2 | Yes. | You build and schedule it. |
| Notebook | Yes. | You write and schedule it. |
| Mirroring | Database mirroring: yes, a read-only replica. Metadata mirroring: no. | Fabric, continuously. |
| Shortcut | No. | Nobody. It's a pointer. |
Database mirroring is also cheap. Its sync compute doesn't use your capacity units (CUs), the compute you pay for. Mirrored storage is free up to a limit tied to your capacity size; an F64 includes 64 TB.
Decision rule: if data must stay in the source, use a shortcut. If you need an always-current copy of a database with no pipelines, use mirroring.
You are a data engineer at Northwind Rides. A partner keeps station maintenance files in its own Amazon S3 bucket. The contract says the files must not be copied out of that bucket. Analysts need to join them with Lakehouse tables. What should you use?
Pick one answer.
Question 2: low-code or code-first?
Exam depth: know it cold.
If data must be copied, ask who builds it. Low-code tools use a visual editor:
- Dataflow Gen2 cleans and reshapes data in Power Query, the point-and-click editor also found in Excel and Power BI. It has 300+ transformations.
- A Pipeline (older pages: Data pipeline) is a canvas of steps. Its Copy activity moves data. Other activities run dataflows, notebooks and loops in order, on a schedule or when an event happens.
- A Copy job copies data with a wizard, no pipeline needed. It does bulk copy, incremental copy (only new or changed rows, found through a watermark column such as LastModified) and change data capture (CDC) replication. CDC is a database feature that records every insert, update and delete.
Code-first tools need code:
- A Notebook runs PySpark or Spark SQL, for any volume and any logic.
- T-SQL in a Warehouse loads and transforms with
COPY INTO,INSERTorCREATE TABLE AS SELECT.
Match
Match each Northwind task to the tool built for it
Pick an item on the left, then its match on the right.
Question 3: batch or streaming?
Exam depth: know it cold.
Batch data arrives in chunks: a file each night, a table once an hour. Every tool above handles batch. Streaming data arrives as a constant flow of small events, such as each scooter's position every few seconds. For that, Fabric has Eventstream:
- It captures events from sources such as Azure IoT Hub, Kafka or a database's CDC feed.
- It transforms them with no code: filter, aggregate over time windows, join streams.
- It routes them to destinations: an Eventhouse, a Lakehouse, Activator (Fabric's alerting item) or a custom endpoint.
An Eventstream doesn't store events. It passes them through memory to its destinations. An Eventhouse can also pull data in by itself with its Get data wizard. For streaming in code, Notebooks offer Spark Structured Streaming, covered in a later lesson.
You are a data engineer at Northwind Rides. Scooter battery events must reach an Eventhouse within seconds, and low-battery events must raise an alert. The team prefers no code. What should you use?
Pick one answer.
The decision map
Exam depth: know it cold.
Ask the three questions in order: streaming first, then copy or not, then who builds it. The first answer that settles the question wins. You'll use this map for the rest of the course.
Each tool has words that give it away in a scenario. ETL means extract, transform, load: the classic copy-and-clean job.
| The scenario says | Pick |
|---|---|
| "streaming", "events", "within seconds". | Eventstream |
| "no data movement", "data must stay in the source". | Shortcut |
| "near real-time copy of a database", "no ETL to maintain". | Mirroring |
| "Power Query", "low-code transformation". | Dataflow Gen2 |
| "copy only", "incremental", "CDC", "no pipeline". | Copy job |
| "orchestrate", "run in order", "loop over items". | Pipeline |
| "PySpark", "complex logic", "custom library". | Notebook |
| "SQL skills", "stored procedure", "Warehouse". | T-SQL |
Exam-style practice
You are a data engineer at Northwind Rides. Customer and payment data lives in Azure SQL Database. Analysts need it in OneLake within minutes of each change. The team wants no pipelines to maintain and no capacity cost for syncing. What should you use?
Pick one answer.
You are a data engineer at Northwind Rides. An on-premises SQL Server has 40 tables, each with a LastModified column. You must load them into a Lakehouse: a full load first, then only new or changed rows every hour. The team wants a wizard, no pipeline, and no state to track by hand. What should you use?
Pick one answer.
You are a data engineer at Northwind Rides. A business analyst knows Power Query from Excel. Each week she must clean the partner's CSVs, fix date formats and merge them with the station list, then load a Lakehouse table. She can't write code. What should she use?
Pick one answer.
Worth keeping
- Shortcuts copy nothing; mirroring keeps a read-only copy in sync for you.
- Low-code: Dataflow Gen2, Pipeline, Copy job. Code-first: Notebook, T-SQL.
- Streaming events go through an Eventstream to an Eventhouse or Lakehouse.
- Ask in order: streaming? copy or not? who builds it?
- Find the constraint word in the scenario; it names the tool.
On the job
Here's the plan you hand your manager on Friday.
Scooter trip events: an Eventstream reads them from Azure IoT Hub, the Azure service that collects device messages. It drops test rides and sends the rest to an Eventhouse for live queries. A second route sends battery alerts to Activator.
Customers and payments: SQL Server is mirrored into Fabric. Analysts get read-only Delta tables that stay current, and nobody maintains a load job.
Partner maintenance CSVs: the contract keeps them in the partner's S3 bucket, so a shortcut makes them visible in the Lakehouse without a copy.
The nightly clean-up: a Pipeline runs a Notebook that joins trips, stations and maintenance into clean tables. Then it refreshes the reports.
Four sources, four different tools, and each choice came from one constraint.
You know how data gets in. But in to what: a Lakehouse, a Warehouse or an Eventhouse? And once it's there, who reshapes it: Dataflow Gen2, a Notebook or T-SQL? Next lesson builds the decision tables for both.
Leaderboard
Loading…
50 XP for each lesson you finish, 10 for each right answer. Names show only for learners who turn on Show me on leaderboards.
Sources for this lesson (9)
Each fact in the lesson is marked with the number of the page it comes from.
- [4]Microsoft Learn: Study Guide for Exam DP-700
- [33]Microsoft Learn: Lakehouse and Delta Lake tables
- [64]Microsoft Learn: Microsoft Fabric decision guide: Choose between Warehouse and Lakehouse
- [67]Microsoft Learn: Get data into Microsoft Fabric
- [68]Microsoft Learn: Microsoft Fabric decision guide: copy activity, Copy job, dataflow, Eventstream, or Spark
- [69]Microsoft Learn: Microsoft Fabric decision guide: Choose a data movement strategy
- [70]Microsoft Learn: Microsoft Fabric decision guide: Choose a data integration strategy
- [71]Microsoft Learn: What is Dataflow Gen2?
- [72]Microsoft Learn: Unify data with OneLake shortcuts and mirroring