Manage Analytics Solution
30-35%of exam
Ingest + Transform Data
30-35%of exam
Monitor + Optimize
30-35%of exam
Quick Facts
- Exam
- DP-700
- Credential
- Fabric Data Engineer
- Time
- 100 min
- Pass
- 700/1000
- Level
- Intermediate
- Product
- Microsoft Fabric
- Role
- Data Engineer
- Blueprint
- April 20 2026
Domain Balance
Manage, Ingest, Monitor split evenly
Manage: secureIngest: loadMonitor: tune
Pipeline vs Notebook
Pipeline
- Orchestrates activities
- Copy data
- Schedules runs
Notebook
- Transforms data
- Spark session
- Code execution
Coordinate vs compute
Security Picker
- Workspace administration→Admin(Control plane)
- Create content→Contributor(Workspace role)
- Read workspace only→Viewer(No write)
- Secure lake data→OneLake roles(Data plane)
- Filter rows→RLS(User context)
- Mask values→Dynamic masking(Sensitive data)
Workspace Settings
- Spark pool
- Workspace compute default
- Starter pools
- Faster session startup
- High concurrency
- Shared notebook session
- Resource profile
- Read/write tuning preset
- Item compute
- Per-item Spark sizing
- Domain
- Business grouping
- OneLake cache
- Shortcut read acceleration
- Dataflow scale
- Gen2 compute setting
Security Plane
Workspace acts; OneLake reads
Workspace: actionsItem: sharingOneLake: dataRLS: rows
Workspace vs OneLake Roles
Workspace role
- Control plane
- Item visibility
- Content actions
OneLake role
- Data plane
- Folder/table access
- Viewer data access
Actions vs data
Lifecycle
- Git integration
- Branch-based versioning
- Azure DevOps
- Supported Git provider
- GitHub
- Supported Git provider
- Database project
- Warehouse schema source
- Deployment pipeline
- Dev-test-prod promotion
- Deployment rules
- Stage-specific bindings
- Credentials
- Not copied
- Data refresh
- Required after deploy
Security + Governance
- Admin
- Full workspace control
- Member
- Manage content/users
- Contributor
- Create/edit items
- Viewer
- Read workspace content
- Item sharing
- Item-level access
- OneLake roles
- Data-plane access
- DefaultReader
- ReadAll data access
- Sensitivity label
- Information protection
- Endorsement
- Trusted content signal
- Audit logs
- Activity trail
Access Controls
- RLS
- Row filter
- CLS
- Column filter
- OLS
- Object hiding
- Folder security
- Path-level access
- File security
- Object-level access
- Dynamic masking
- Sensitive value masking
- Entra ID
- Identity provider
- Service principal
- App identity
Orchestration
- Pipeline
- Activity orchestration
- Dataflow Gen2
- Low-code transformation
- Notebook
- Code transformation
- Schedule
- Time-based trigger
- Event trigger
- Event-based run
- Parameter
- Runtime input
- Expression
- Dynamic value
- Notebook activity
- Pipeline-called code
Tool Stack
Flow, Pipe, Note, SQL, KQL
Flow: low-codePipe: orchestrateNote: SparkSQL/KQL: query
Lakehouse vs Warehouse
Lakehouse
- Delta files
- Spark friendly
- Flexible schema
Warehouse
- Relational SQL
- T-SQL serving
- Structured modeling
Files vs SQL
Transformation Picker
- Low-code shaping→Dataflow Gen2(Power Query)
- Custom Spark logic→Notebook(PySpark)
- Warehouse SQL→T-SQL(Relational)
- Eventhouse query→KQL(Real-time)
- Move then transform→Pipeline(Activities)
- Reusable code job→Spark job(Batch)
Stores + Patterns
- Lakehouse
- Delta lake analytics
- Warehouse
- Relational SQL analytics
- Eventhouse
- Real-time KQL store
- KQL database
- Time-series analytics
- Semantic model
- BI consumption layer
- Full load
- Reload everything
- Incremental load
- Load changes only
- Streaming load
- Continuous event ingestion
Streaming Path
Events flow, land, query, window
Eventstream: flowEventhouse: landKQL: queryWindow: time
Dataflow vs Notebook
Dataflow Gen2
- Low-code
- Power Query
- Many connectors
Notebook
- Code-first
- PySpark/Scala
- Custom logic
Low-code vs code
Storage Picker
- Delta analytics→Lakehouse(Files/tables)
- SQL serving→Warehouse(Relational)
- Telemetry analytics→Eventhouse(KQL)
- External data→Shortcut(No copy)
- Operational replica→Mirroring(Copied changes)
- Report layer→Semantic model(BI)
Batch Tools
- Copy activity
- Move source data
- Dataflow Gen2
- Power Query shaping
- Notebook
- PySpark/custom code
- T-SQL
- Warehouse transformations
- KQL
- Eventhouse transformations
- Shortcut
- Virtual data reference
- Mirroring
- Near-real-time replica
- Fast copy
- Dataflow ingestion boost
Shortcut vs Mirroring
Shortcut
- Virtual pointer
- No copy
- Target permissions
Mirroring
- Replicated data
- Source changes
- Fabric copy
Reference vs replicate
Transforms + Modeling
- Denormalize
- Flatten for analytics
- Fact table
- Measures/events
- Dimension table
- Descriptive context
- Star schema
- Facts plus dimensions
- Aggregate
- Summarize groups
- Surrogate key
- Warehouse identifier
- Late dimension
- Fact arrives first
- Inferred member
- Temporary dimension row
Eventstream vs Eventhouse
Eventstream
- Ingest events
- Route/process
- Stream pipeline
Eventhouse
- Store events
- KQL query
- Real-time analytics
Flow vs store
Data Quality
- Duplicate rows
- Deduplicate by key
- Missing data
- Impute or reject
- Late events
- Handle watermark lag
- Schema drift
- Source shape changed
- Checkpoint
- Streaming resume state
- Watermark
- Late-data boundary
- CDC
- Change data capture
- Idempotent load
- Safe rerun
Full vs Incremental
Full load
- Reload all
- Simple logic
- Longer window
Incremental
- Changed rows
- Needs watermark
- Shorter window
All vs changes
Streaming Tools
- Eventstream
- Event routing/processing
- Eventhouse
- Real-time analytics store
- Native table
- KQL-owned storage
- Shortcut table
- Referenced OneLake data
- Query acceleration
- Shortcut query boost
- Spark streaming
- Micro-batch processing
- Window function
- Time-bucket computation
- KQL
- Streaming query language
Tune Order
Prune, compact, cache, scale
Prune scansCompact filesCache hotScale compute
OPTIMIZE vs V-Order
OPTIMIZE
- Compacts files
- Delta maintenance
- Write cleanup
V-Order
- Read layout
- Parquet optimization
- Power BI speed
Compact vs layout
Troubleshooting Picker
- Pipeline failed→Run history(Activities)
- Dataflow failed→Refresh details(Mashup)
- Notebook failed→Spark logs(Driver)
- Shortcut fails→Credentials(Target access)
- Model stale→Refresh history(Semantic)
- Capacity saturated→Metrics app(Utilization)
Monitor + Alerts
- Monitoring hub
- Fabric run status
- Pipeline run
- Activity execution
- Dataflow refresh
- Gen2 execution
- Notebook run
- Spark execution
- Semantic refresh
- Model data update
- Alerts
- Condition notifications
- Capacity metrics
- Resource utilization
- Audit events
- User activity
Error Triage
- Pipeline error
- Activity failure
- Dataflow error
- Mashup/load issue
- Notebook error
- Spark/code failure
- Eventhouse error
- KQL store issue
- Eventstream error
- Event route issue
- T-SQL error
- SQL statement issue
- Shortcut error
- Target/credential issue
- Refresh error
- Semantic update failure
Performance Tuning
- OPTIMIZE
- Compact Delta files
- V-Order
- Parquet read optimization
- Partitioning
- Prune scanned data
- Z-order
- Cluster related values
- Statistics
- Query planning metadata
- Indexes
- Warehouse access paths
- Materialization
- Precomputed results
- Query acceleration
- KQL shortcut speed
Query + Spark
- Predicate pushdown
- Filter near source
- Projection pruning
- Read needed columns
- Broadcast join
- Small table join
- Shuffle
- Network data exchange
- Skew
- Uneven partition load
- Caching
- Reuse hot data
- Concurrency
- Parallel user pressure
- Explain plan
- Query execution shape
Common Traps
Workspace vs data access
Viewer sees item ≠ OneLake role reads data
Deployment copies metadata
Definitions promote ≠ Data stays behind
Shortcut vs copy
Shortcut references target ≠ Mirroring replicates changes
Pipeline vs transform
Pipeline orchestrates ≠ Notebook transforms
Low-code vs custom
Dataflow uses Power Query ≠ Notebook uses code
Streaming resume
Checkpoint preserves state ≠ Watermark bounds lateness
Telemetry store
Eventstream routes events ≠ Eventhouse stores events
Optimization scope
OPTIMIZE compacts files ≠ V-Order improves reads
Last Minute
- 1.Weights: three equal 30-35 bands
- 2.Know SQL, PySpark, KQL
- 3.Workspace roles govern actions
- 4.OneLake roles govern data
- 5.Deployment pipelines skip data
- 6.Credentials remain stage-specific
- 7.Shortcuts reference; mirroring copies
- 8.Dataflow low-code; notebooks code
- 9.Pipeline orchestrates; notebook computes
- 10.Eventstream flows; eventhouse stores
- 11.Incremental loads need watermarks
- 12.OPTIMIZE compacts Delta files
Buy on Amazon
Take courses on Udemy
Learning path on PluralsightSame family resources
Explore More Microsoft Azure Certifications
Continue into nearby exams from the same family. Each card keeps practice questions, study guides, flashcards, videos, and articles in one place.
More From This Family
Videos and articles for deeper review.
VideoAI-900 Exam Guide 2026: FREE Azure AI Fundamentals Study PlanVideoAZ-104 Study Plan 2026: Pass Azure Administrator ExamVideoAZ-900 Exam Guide 2026: Pass Azure Fundamentals Free (Complete Study Plan)VideoFREE MD-102 Exam Guide 2026: Pass Microsoft Endpoint Administrator (Intune, Autopilot, Windows 11)ArticleFREE DP-700 Exam Guide 2026: Pass Microsoft Fabric Data Engineer (Lakehouse, KQL, Pipelines)26 min readArticleAI-900 Exam Guide 2026: FREE Azure AI Fundamentals Study Plan26 min readArticleAZ-104 Study Plan 2026: Pass Azure Administrator Exam17 min readArticleAZ-900 Exam Guide 2026: Pass Azure Fundamentals Free (Complete Study Plan)28 min read