TL;DR:DP-700 doesn't test definitions in isolation — it tests which of several similar Fabric tools fits a described scenario. This sheet skips prose and goes straight to a glossary and five decision tables built around the exact choices Microsoft's objectives call out by name.
MSCertQuiz sells a 500-question DP-700 practice bank; this cheat sheet is written by the same team using Microsoft's DP-700 study guide (skills measured as of July 21, 2026, checked September 7, 2026) as the source for every table below.
DP-700 at a Glance
Passing score
700 / 1000
Level
Associate
Renewal
Free, annual
Prerequisites
None formal
Implement & Manage
30–35%
Ingest & Transform
30–35%
Monitor & Optimize
30–35%
Microsoft does not publish a DP-700-specific price, question count, or exam duration on the credential page — those show at registration and aren't repeated here as if they were fixed figures. Microsoft Q&A threads about DP-700 registration report $165 USD for US test-takers.
Fabric Data Engineering Glossary
| Term | What it is |
|---|---|
| OneLake | The single, tenant-wide data lake underlying every Fabric workspace — one copy of data, referenced by multiple engines. |
| Lakehouse | A Fabric item combining file storage (Delta/Parquet) with SQL and Spark access over the same data. |
| Warehouse | A fully T-SQL, transactional-style relational engine in Fabric, built on the same OneLake storage. |
| Shortcut | A reference to data that already exists elsewhere (another Fabric item, ADLS Gen2, S3) — no data is copied. |
| Mirroring | Continuous, near real-time replication of an external database (e.g. Azure SQL DB, Cosmos DB) into OneLake. |
| Dataflow Gen2 | A low-code, Power Query-style item for building transforms without writing code. |
| Pipeline | An orchestration item that runs and sequences activities — copies, notebooks, Dataflows — on a schedule or trigger. |
| Notebook | A code-first item (PySpark, SQL, KQL) for custom data transformation logic. |
| Eventstream | A no-code item for routing and lightly transforming streaming data before it lands. |
| Eventhouse | A KQL-based database purpose-built for high-volume, low-latency event/telemetry data (Real-Time Intelligence). |
| Deployment pipeline | Stages and promotes Fabric items through Dev/Test/Prod workspaces. |
| Semantic model | The data model layer (relationships, measures) that Power BI reports query. |
| V-Order | A write-time optimization on Delta/Parquet files in a Lakehouse that speeds up downstream reads. |
| Dynamic data masking | Hides sensitive column values from unauthorized users at query time without changing stored data. |
| Sensitivity label | A classification tag (e.g. Confidential) applied to an item for governance/compliance. |
Decision Table: Dataflow Gen2 vs Pipeline vs Notebook
Microsoft calls this choice out explicitly as a testable skill — it recurs across all three DP-700 domains.
| If the question says… | Think… |
|---|---|
| Business users, no code, self-service | Dataflow Gen2 |
| Schedule, sequence multiple steps, trigger on event | Pipeline |
| Custom code, conditional logic, PySpark/SQL/KQL by hand | Notebook |
| Chain a notebook and a Dataflow together with dependencies | Pipeline (orchestrates the other two) |
Decision Table: OneLake Shortcuts vs Mirroring vs Copy Activity
All three get data into a Lakehouse or Warehouse without you writing custom ingestion code — the exam tests knowing which one avoids duplication vs which one replicates.
| If the question says… | Think… |
|---|---|
| Reference data in place, avoid duplicating it | OneLake shortcut |
| Continuously synced copy of an operational database, no pipeline to maintain | Mirroring |
| One-time or scheduled batch move between locations | Pipeline Copy activity |
Decision Table: T-SQL vs PySpark vs KQL
DP-700's audience profile explicitly names all three languages — expect the exam to test picking the right one, not writing syntax.
| If the question says… | Think… |
|---|---|
| Set-based logic against a Warehouse, relational joins | T-SQL |
| Large-scale files in a Lakehouse, custom/branching logic | PySpark (in a notebook) |
| Querying an Eventhouse, telemetry/time-series data | KQL |
Drill These Decision Tables Against Real Scenarios
40 free DP-700 practice questions to test whether these choices stick under exam-style scenarios.
Start Free DP-700 Practice →Decision Table: Real-Time Intelligence Storage Options
A narrower, exam-specific ambiguity inside Domain 2's streaming objectives.
| Option | Use when… |
|---|---|
| Native Eventhouse table | Fastest queries needed; data is written directly and optimized for KQL at ingestion. |
| OneLake shortcut with query acceleration | Data must stay accessible to other engines without duplication, but query speed still matters — adds a cache layer. |
| Standard OneLake shortcut | Zero duplication is the priority and query latency is not critical. |
Security & Governance Quick Reference
| Control | Scopes access to… |
|---|---|
| Workspace-level access | Whether a user can enter the workspace at all. |
| Item-level access | A specific report, Lakehouse, or pipeline within the workspace. |
| Row-level security | Which rows a user sees within a table they can otherwise query. |
| Column-level security | Which columns a user sees within a visible row. |
| OneLake security | The underlying files/folders, independent of which Fabric item reads them (including shortcuts). |
| Dynamic data masking | Column values shown at query time, without altering stored data. |
Last-Minute DP-700 Questions
Is this cheat sheet enough on its own?
No — it's a compression of concepts you should already understand, for final review. DP-700 tests applying these decisions to a described scenario, so pair this with the practice questions before test day.
What's the single most-repeated decision on DP-700?
Dataflow Gen2 vs pipeline vs notebook. It appears in Microsoft's objectives under orchestration but shows up in ingestion and transformation scenarios too.
What's the difference between a shortcut and mirroring again?
A shortcut points at data that already exists elsewhere — nothing is copied. Mirroring continuously copies (replicates) a source database into OneLake. If the scenario says "avoid duplicating," think shortcut; if it says "keep a synced copy without a pipeline," think mirroring.
Do I need to memorize KQL syntax for DP-700?
You need to recognize when KQL is the right choice (Eventhouse/telemetry queries) more than write exact syntax from memory — the exam tests tool selection more heavily than hand-writing code.
What's the passing score and how is it scored?
700 out of 1000 on Microsoft's standard scaled scoring system, stated directly on the DP-700 study guide's own scoring link.
Is V-Order something I need to actively run, or automatic?
It's a write-time optimization applied to Delta/Parquet files; the exam tests recognizing it as the fix for slow reads caused by table layout, distinct from a compute-side fix like resizing a Spark pool.
More DP-700 Resources
DP-700 Study Guide 2026
Full domain-by-domain breakdown, including a real DP-700 vs DP-203 vs DP-600 comparison.
Free DP-700 Practice Questions
14 scenario questions across 3 case studies, one per domain.
DP-700 Readiness Quiz
Free timed mock exam to gauge whether you're ready to book.
DP-700 Certification Page
Full 500-question practice bank details and pricing.