ExamFinch.com My Study

Cheat sheet

Databricks Data Engineer Cheat Sheet

Practice this exam

Databricks Intelligence Platform

  • 0% of exam
  • Platform Core Compute Picker Delta + UC Control Plane

Data Ingestion + Loading

  • 0% of exam
  • Ingestion Methods Auto Loader COPY INTO Lakeflow Connect

Data Transformation + Modeling

  • 0% of exam
  • Medallion Layers Transform Ops Gold Objects Data Quality

Lakeflow Jobs

  • 0% of exam
  • Job Orchestration Task Types Triggers Run Repair

CI/CD

  • 0% of exam
  • Git Workflow DAB Bundle Targets CLI Deploy

Troubleshooting + Optimization

  • 0% of exam
  • Spark UI Bottlenecks Liquid Clustering Predictive Optimization

Governance + Security

  • 0% of exam
  • Unity Catalog Privileges Masking Delta Sharing

Quick Facts

  • Exam Data Engineer Associate
  • Questions 45 scored
  • Time 90 min
  • Fee USD 200
  • Type Multiple choice
  • Delivery Online/test center
  • Aides None
  • Validity 2 years
  • Weights Not published
  • Code SQL when possible

Platform Core PL

  • Lakehouse Warehouse + lake
  • Delta Lake ACID table layer
  • Unity Catalog Governance plane
  • Workspace Collaborative UI
  • Control plane Databricks services
  • Data plane Customer compute
  • Photon Vectorized engine
  • SQL warehouse SQL compute

Compute Picker PL

  • Serverless Hands-off compute
  • SQL warehouse BI queries
  • All-purpose Interactive notebooks
  • Job compute Per-run compute
  • Pool Fast startup
  • Policy Compute guardrails
  • Databricks Connect Local IDE
  • Single node Small tests

Ingest CAL IN

  • COPY, Auto Loader, Lakeflow Connect
  • COPY: batch Auto: files Lakeflow: connectors

COPY INTO vs Auto Loader IN

COPY INTO

  • Batch files
  • SQL command
  • Simple replay

Auto Loader

  • Continuous files
  • cloudFiles source
  • Schema drift
  • Batch vs stream

Ingestion Picker IN

  • One-time files → COPY INTO ( Batch )
  • Continuous files → Auto Loader ( cloudFiles )
  • Enterprise SaaS → Lakeflow Connect ( Managed )
  • Relational source → JDBC/ODBC ( Client )
  • API source → REST ( Notebook )
  • Nested JSON → Auto Loader ( Rescue )
  • UC file path → Volume ( Governed )

Ingestion Methods IN

  • COPY INTO Idempotent batch
  • Auto Loader Cloud file stream
  • Lakeflow Connect Managed connectors
  • JDBC/ODBC Database clients
  • REST API ingestion
  • Partner External connector
  • Local file Manual upload
  • Volume UC file storage

Lakeflow Connect vs JDBC IN

Lakeflow Connect

  • Managed connector
  • Enterprise sources
  • UC landing

JDBC/ODBC

  • Client access
  • Custom logic
  • Notebook code
  • Managed vs custom

Auto Loader IN

  • cloudFiles Streaming source
  • schemaLocation Schema tracking
  • Checkpoint Processed file state
  • File events Notification mode
  • Directory listing Listing mode
  • addNewColumns Fail then evolve
  • rescue Capture drift
  • _rescued_data Unexpected fields

Medallion MO

  • Bronze -> Silver -> Gold
  • Bronze: raw Silver: clean Gold: BI Quality rises

View vs Materialized View MO

View

  • Query definition
  • Computed on read
  • No stored result

Materialized view

  • Stored result
  • Refreshed data
  • Faster reads
  • Virtual vs stored

Modeling Picker MO

  • Raw landing → Bronze ( Preserve )
  • Clean entities → Silver ( Validate )
  • BI metrics → Gold ( Aggregate )
  • Array column → explode ( Flatten )
  • Duplicate records → dropDuplicates ( Dedup )
  • Small dimension → Broadcast join ( Avoid shuffle )

Medallion Layers MO

  • Bronze Raw landing
  • Silver Cleaned entities
  • Gold Business outputs
  • Raw columns Bronze preserve
  • Dedup Silver cleanup
  • Metrics Gold aggregate
  • Lineage Layer trace
  • Quality Silver/gold checks

Transform Ops MO

  • Inner join Matched rows
  • Left join Keep left
  • Broadcast join Small side shipped
  • Cross join Cartesian product
  • Union Distinct rows
  • Union all Keep duplicates
  • explode Array to rows
  • dropDuplicates Remove duplicates
  • approx_count_distinct Fast cardinality
  • summary Column statistics

Gold Objects MO

  • Table Stored Delta data
  • View Stored query
  • Materialized view Precomputed query
  • Streaming table Incremental table
  • BI table Consumption model
  • Metric view Governed metrics
  • Dashboard Visual consumer
  • Quality rule Reliability check

Jobs DAG JO

  • Tasks depend, triggers start, repair reruns
  • DAG: order Trigger: start Repair: failed tasks

Scheduled vs Data Trigger JO

Scheduled

  • Clock based
  • Fixed cadence
  • May run empty

Data trigger

  • Data arrival
  • Table/file change
  • Event aligned
  • Time vs readiness

Production Picker JO

  • Task dependencies → Lakeflow Jobs ( DAG )
  • Clock cadence → Scheduled ( Time )
  • New files → File arrival ( Trigger )
  • Source tables change → Table update ( Trigger )
  • Failed task → Repair run ( Rerun )
  • Dev to prod → DAB ( Deploy )

Job Orchestration JO

  • Lakeflow Job Pipeline orchestration
  • Task Work unit
  • DAG Dependency graph
  • Retry Failure rerun
  • Branching Conditional path
  • Looping Repeated task
  • Repair run Failed-task rerun
  • Run history Execution trends

Triggers JO

  • Scheduled Time based
  • File arrival New files
  • Table update Table changed
  • Continuous Always running
  • Manual Run now
  • External Outside orchestrator
  • Dependency Upstream task
  • Failure rate Health signal

Git Folder vs DAB CI

Git folder

  • Branch code
  • Commit changes
  • Pull request

DAB

  • Package assets
  • Target config
  • Deploy environments
  • Version vs deploy

DAB Bundle CI

  • DAB Declarative deployment
  • databricks.yml Bundle config
  • Target Environment config
  • Variables Reusable settings
  • Override Target-specific value
  • Validate Check bundle
  • Deploy Promote assets
  • Git folder Workspace repo

Spark UI OP

  • Skew, shuffle, spill explain slowness
  • Skew: uneven Shuffle: exchange Spill: disk

OPTIMIZE vs VACUUM OP

OPTIMIZE

  • Compact files
  • Improve reads
  • Layout rewrite

VACUUM

  • Delete old files
  • Reduce storage
  • Limits time travel
  • Speed vs cleanup

Spark Tuning OP

  • Skew Uneven partitions
  • Shuffle Data exchange
  • Spill Disk overflow
  • Partitions Parallelism units
  • autoBroadcast Small join threshold
  • Executor memory Worker heap
  • Driver memory Coordinator heap
  • Spark UI Stage diagnostics

UC Access GO

  • Grant base, restrict finer
  • GRANT: allow DENY: block Mask: values Filter: rows

Managed vs External Table GO

Managed

  • UC storage
  • UC lifecycle
  • Drop deletes data

External

  • External location
  • Shared files
  • Drop keeps data
  • Owned vs referenced

Governance Picker GO

  • Central governance → Unity Catalog ( UC )
  • Read access → SELECT ( Privilege )
  • Hide rows → Row filter ( RLS )
  • Hide values → Column mask ( CLS )
  • Tag-driven rules → ABAC ( Central )
  • External sharing → Delta Sharing ( Read-only )

Governance Core GO

  • Catalog Top namespace
  • Schema Database namespace
  • Managed table UC lifecycle
  • External table External location
  • GRANT Add permission
  • REVOKE Remove permission
  • DENY Explicit block
  • Lineage Data flow
  • Delta Sharing Read-only share
  • Federation External querying

Row Filter vs Column Mask GO

Row filter

  • Restrict rows
  • Tenant isolation
  • WHERE-like logic

Column mask

  • Transform values
  • PII redaction
  • Column function
  • Rows vs values

Security Controls GO

  • Principal User/group/SP
  • USE CATALOG Catalog access
  • USE SCHEMA Schema access
  • SELECT Read rows
  • MODIFY Write rows
  • Row filter Hide rows
  • Column mask Transform values
  • ABAC Tag-based control
  • Audit logs Access evidence
  • System tables Operational telemetry

ABAC vs Table Controls GO

ABAC

  • Tag matched
  • Central policy
  • Owner cannot remove

Table controls

  • Per-table setup
  • ALTER TABLE
  • Owner managed
  • Central vs local

Common Traps

Current weights

  • May guide omits weights ≠ Avoid legacy percentages

COPY vs Auto Loader

  • COPY is batch ≠ Auto Loader streams

Rescue vs evolve

  • Rescue captures drift ≠ addNewColumns evolves

View vs MV

  • View stores logic ≠ MV stores results

External drop

  • Table metadata drops ≠ Files remain

Privilege stack

  • Need USE parents ≠ SELECT reads table

ABAC access

  • ABAC restricts only ≠ GRANT still required

VACUUM risk

  • Cleans old files ≠ Breaks old time travel

Triggers

  • Schedule uses clock ≠ Data trigger waits

Last Minute

  • 1. 45 scored; 90 minutes
  • 2. May guide: no weights
  • 3. SQL when possible
  • 4. Auto Loader = cloudFiles
  • 5. COPY INTO = batch
  • 6. Bronze raw; Silver clean
  • 7. Gold = BI outputs
  • 8. Jobs are DAG tasks
  • 9. DAB deploys assets
  • 10. Spark UI finds skew
  • 11. UC governs data
  • 12. Masks values; filters rows
  • 13. ABAC needs base grants
  • 14. Delta Sharing is read-only

Databricks Certified Data Engineer Associate

  • Practice Questions 194 questions Study Guide 40 topics Cheat Sheet 110 facts Flashcards 50 cards Shorts 1 short Videos 1 video Blogs 1 blog Free CE Podcast