DP-750

DP-750 Cheat Sheet: Task-to-Command Reference

Print-friendly12 min read

Quick answer

This is a reference, not a tutorial: exam facts, then every task the DP-750 skills outline names, mapped to the exact Databricks UI path, SQL statement, or CLI command that does it, plus a pattern table for exam phrasing that maps to a specific concept. Pair it with the full study guide for the reasoning behind each choice.

Exam Snapshot

Exam codeDP-750
CertificationMicrosoft Certified: Azure Databricks Data Engineer Associate
Domains4 — two at 15–20%, two at 30–35%
Passing score700 / 1000 (scaled)
Exam / seat timeNot published per-exam by Microsoft as of this writing; general Associate/Expert (no-lab) allocation is 100 min exam / 120 min seat
PrerequisitesNone enforced; SQL, Python, Git/SDLC, Microsoft Entra, Azure Data Factory, and Azure Monitor familiarity recommended
LanguagesSQL and Python only — Scala is not named on the current audience profile
RenewalAnnual, free online assessment on Microsoft Learn
Practice AssessmentAvailable via AI Skills Navigator (sign-in required)

Source: Microsoft Learn's DP-750 study guide, checked September 7, 2026. Price and per-exam duration/question count are not published on that page — we left them out rather than borrow a figure from a different exam.

Drill this into memory

Practice the Real Task-Selection Logic

40 free DP-750 questions, scored instantly. No credit card required.

Start Free Practice →

Environment and Compute Tasks

Domain 1 (15–20%) is largely about choosing and configuring the right compute and Unity Catalog objects up front.

TaskWhere / How
Choose compute for interactive BI/SQLServerless SQL warehouse
Choose compute for a scheduled batch jobJob compute
Configure autoscaling / node type / poolingCompute → Edit → Performance settings
Enable faster query executionCompute → Photon acceleration toggle
Create a catalog / schema / volumeCREATE CATALOG / CREATE SCHEMA / CREATE VOLUME
Connect to an external data source as a catalogCatalog Explorer → Foreign catalog → Connection
Let business users query data conversationallyCatalog Explorer → AI/BI Genie instructions

Unity Catalog Governance Tasks

Domain 2 (15–20%) — the same weight as Domain 1, and easy to underestimate.

TaskWhere / How
Grant a user/group access to a tableGRANT SELECT ON TABLE … TO `group`
Hide specific column values by roleColumn mask function on the table
Restrict which rows a user can seeRow filter function on the table
Read a secret from Key Vault in a notebookdbutils.secrets.get(scope, key)
Authenticate a pipeline (not a user) to dataService principal or managed identity
Tag-driven access rules across many objectsAttribute-based access control (ABAC) policy
See who queried a table and whenCatalog Explorer → Lineage / Audit logs
Share curated tables with another organizationDelta Sharing share/recipient

Data Ingestion and Processing Tasks

Domain 3 (30–35%) — the largest domain, together with Domain 4.

TaskWhere / How
One-time batch load from filesCREATE TABLE … AS / COPY INTO
Continuously ingest new arriving filesAuto Loader (inside a Declarative Pipeline)
Ingest from a managed SaaS/DB source with minimal setupLakeflow Connect
Continuous low-latency processingSpark Structured Streaming
Ingest a live event streamAzure Event Hubs connector
Apply upserts from a source system\'s change feedCDC feed + MERGE
Preserve full change history on a dimensionSCD Type 2 (versioned rows)
Enforce column types / block malformed rowsSchema enforcement + data type checks
Declare validation rules with a fail thresholdPipeline expectations (Lakeflow Declarative Pipelines)

Pipeline and Job Tasks

Domain 4 (30–35%) — shipping and keeping pipelines alive.

TaskWhere / How
Build a declarative, expectation-driven pipelineLakeflow Spark Declarative Pipelines
Run tasks on a fixed scheduleLakeflow Job → Cron trigger
Start a run the moment a file landsLakeflow Job → File-arrival trigger
Recover automatically after a transient failureJob → Automatic restarts / retry policy
Package job/pipeline config as version-controlled codeDatabricks Asset Bundles
Deploy a bundle from a CI/CD pipelinedatabricks bundle deploy (CLI) or REST API
Review/approve a teammate\'s pipeline changeGit pull request

Monitoring and Optimization Tasks

TaskWhere / How
Diagnose one slow task in an otherwise-fast stageSpark UI + query profile → check for skew
Visualize a job\'s execution stagesDAG (Directed Acyclic Graph) view
Reclaim space from old Delta file versionsVACUUM table_name
Compact small files for faster readsOPTIMIZE table_name
Improve reads filtered on a changing, high-cardinality columnLiquid clustering
Send job logs to a central monitoring workspaceLog Analytics via Azure Monitor
Get notified when a job fails or runs longAzure Monitor alert rule

If the Exam Says This, Think That

DP-750 scenarios describe a situation rather than naming the feature directly. This table maps common phrasing to the concept it is actually testing.

If the exam says…Think
"…must preserve every historical version of a row"SCD Type 2 / temporal table
"…one task is taking far longer than the rest of the stage"Data skew
"…needs to start the moment a file arrives, not on a schedule"File-arrival trigger
"…must hide a column\'s value but keep the table queryable"Column mask
"…share curated data with another org, no copy, no full access"Delta Sharing
"…the pipeline should call itself to a service, not a person"Service principal / managed identity
"…stop the run automatically if too many rows fail a rule"Pipeline expectations
"…same job config deployed identically via a Git-reviewed workflow"Databricks Asset Bundles
"…reads filtered on a column whose values keep changing"Liquid clustering (over static Z-ordering)
"…recover automatically without manual intervention"Automatic restarts / retry policy

Exam-day reminder:DP-750's audience profile names only SQL and Python. If a scenario shows Scala syntax, treat it as legacy/inherited code being described, not a language you need to write yourself on this exam.

Common Cheat-Sheet Questions

Can I print this for exam-day review?

Yes — every table on this page uses plain borders with no interactive elements, so it prints cleanly from any browser's print dialog.

Is memorizing this task list enough to pass DP-750?

No. DP-750 is scenario-based, and roughly two-thirds of the exam sits in the two largest domains — this sheet is for fast recall of terms and task mappings you've already studied, not a substitute for the full study guide or working through practice scenarios.

Why isn't Scala anywhere on this sheet?

Because Microsoft's official DP-750 audience profile only names SQL and Python, unlike the retired DP-203 exam, which explicitly required Scala. This cheat sheet reflects what the current exam actually tests.

Does this cover Databricks CLI syntax in depth?

No — this sheet maps tasks to the tool or command category that does them (CLI, REST API, SQL, or UI path), since the exam tests knowing which mechanism fits a scenario, not memorizing every CLI flag.

What's the real difference between liquid clustering and Z-ordering?

Z-ordering is a static optimization you re-run manually; liquid clustering adapts more gracefully as data and query patterns change over time — the exam tests picking the right one for a described access pattern, not defining either term.

Is this cheat sheet useful for DP-700 (Fabric) too?

No — DP-700 tests Microsoft Fabric-native tools (Lakehouses, Fabric pipelines), which use different terminology and portal paths than Azure Databricks. Use this sheet only for DP-750 prep.

About This Reference

MSCertQuiz sells a 500-question DP-750 practice bank alongside this cheat sheet, built by the same team. Facts above trace to Microsoft's DP-750 study guide and Azure Databricks documentation, checked September 7, 2026.