ExamFinch.com My Study

Cheat sheet

SnowPro Core Cheat Sheet

Practice this exam

AI Data Cloud Architecture

  • 31% of exam
  • Architecture Layers Objects + Tables Warehouses Compute Picker

Account Management + Governance

  • 20% of exam
  • Governance + Access Policies + Security Monitoring + Cost Governance Picker

Loading + Connectivity

  • 18% of exam
  • Loading + Connectivity Semi-Structured + Unload Load Picker Snowpipe vs COPY Automated Ingestion Pipeline Picker

Performance + Transformation

  • 21% of exam
  • Performance Transformation Transform Picker Query Profile

Data Collaboration

  • 10% of exam
  • Collaboration Sharing Picker Direct Share vs Listing Provider vs Consumer Recovery + Cloning Time Travel vs Fail-safe

Quick Facts

  • Exam COF-C03
  • Credential SnowPro Core
  • Questions 100
  • Time 115 min
  • Pass 750 scaled
  • Cost $175 USD
  • Format Multiple choice/select
  • Launch Feb 16 2026

Three Layers AR

  • Storage -> Compute -> Services
  • Storage: data Compute: warehouses Services: metadata Optimizer: plans

Warehouse vs Database AR

Warehouse

  • Runs compute
  • Consumes credits
  • Can suspend

Database

  • Stores objects
  • Uses privileges
  • No compute
  • Compute vs container

Compute Picker AR

  • Heavy single query → Resize up ( More resources )
  • Many queued users → Multi-cluster ( Scale out )
  • Idle credit burn → Auto-suspend ( Cost guard )
  • Start on demand → Auto-resume ( Convenience )
  • Warehouse credit cap → Resource monitor ( Suspend option )
  • Serverless spend → Budget ( Monitor cost )

Architecture Layers AR

  • Storage Managed columnar data
  • Compute Virtual warehouses
  • Cloud Services Metadata + optimization
  • Metadata Object and file stats
  • Optimizer Query planning
  • Result cache Reused query output
  • Services cache Warehouse-independent
  • Separation Storage apart compute

Warehouse Scaling AR

  • Up = size | Out = clusters
  • Up: heavy query Out: concurrency Suspend: save credits

Size vs Clusters AR

Size

  • Single-query resources
  • Scale up
  • Bigger cluster

Clusters

  • Concurrency
  • Scale out
  • More clusters
  • Power vs concurrency

Objects + Tables AR

  • Database Schema container
  • Schema Object namespace
  • Table Managed rows
  • View Stored query
  • Secure view Protected definition
  • Dynamic table Declarative pipeline
  • Iceberg table Open table format
  • Stage File location

Transient vs Temporary AR

Transient

  • Persists until dropped
  • No Fail-safe
  • Lower recovery

Temporary

  • Session scoped
  • Hidden after session
  • No Fail-safe
  • Persistent vs session

Warehouses AR

  • Warehouse Compute cluster
  • Size Resources per cluster
  • Auto-suspend Stop idle compute
  • Auto-resume Start on demand
  • Multi-cluster Concurrency scale-out
  • Scaling up Bigger cluster
  • Scaling out More clusters
  • Resource monitor Warehouse credit guard

Policy Scope GO

  • Mask columns; row policies filter
  • Masking: values Row: visibility Tags: metadata

Masking vs Row Policy GO

Masking

  • Protect columns
  • Return altered values
  • Tag-based option

Row policy

  • Filter rows
  • Predicate logic
  • Table/view level
  • Columns vs rows

Governance Picker GO

  • Object access → Grants ( Roles )
  • Centralized grants → Managed access ( Schema owner )
  • Hide columns → Masking policy ( Column )
  • Filter rows → Row policy ( Predicate )
  • Classify data → Classification ( Sensitive data )
  • Label assets → Tags ( Metadata )

Governance + Access GO

  • User Login principal
  • Role Privilege holder
  • Privilege Allowed action
  • RBAC Privileges via roles
  • DAC Owner grants access
  • UBAC Direct user grants
  • Secondary role Extra active roles
  • Managed access Centralized grants

Policies + Security GO

  • Masking policy Column protection
  • Row policy Row filtering
  • Tag Metadata label
  • Tag masking Policy via tag
  • Network policy IP allow/deny
  • MFA Extra authentication
  • Tri-Secret Customer key control
  • Access History Usage audit trail

Monitoring + Cost GO

  • Account Usage Shared usage views
  • Information Schema Object metadata
  • Snowsight Web UI
  • Query History Executed statements
  • Warehouse load Queue pressure
  • Budget Spend monitoring
  • Resource monitor Warehouse suspension
  • Credits Compute billing unit
  • Serverless Snowflake-managed compute

Load Flow LO

  • Stage -> Format -> COPY -> Table
  • Stage: files Format: parsing COPY: load Table: rows

Snowpipe vs COPY LO

Snowpipe

  • Continuous files
  • Serverless compute
  • Pipe object

COPY

  • Bulk files
  • Warehouse compute
  • Manual/scheduled
  • Continuous vs batch

Load Picker LO

  • Local files → PUT + COPY ( Internal stage )
  • Cloud files → External stage ( S3/GCS/Azure )
  • Batch load → COPY INTO ( Warehouse compute )
  • Continuous files → Snowpipe ( Serverless )
  • App event stream → Snowpipe Streaming ( No files )
  • Export files → COPY INTO location ( Unload )

Loading + Connectivity LO

  • Internal stage Snowflake file store
  • External stage Cloud storage path
  • File format Parse rules
  • COPY INTO Bulk load/unload
  • Snowpipe Continuous micro-batch
  • Pipe COPY definition
  • PUT Upload local files
  • ON_ERROR Load error action
  • VALIDATION_MODE Check load errors
  • Connector Client integration

CDC Flow LO

  • Stream records; Task consumes
  • Stream: changes Task: runs SQL MERGE: apply

Stream vs Task LO

Stream

  • Tracks changes
  • CDC offset
  • Consumed by DML

Task

  • Runs SQL
  • Schedule/trigger
  • Can form DAGs
  • Detect vs execute

Pipeline Picker LO

  • Need CDC → Stream ( Change offset )
  • Scheduled SQL → Task ( Timed run )
  • Complex DAG → Task graph ( Dependencies )
  • Freshness target → Dynamic table ( Declarative )

Semi-Structured + Unload LO

  • VARIANT Semi-structured value
  • OBJECT Key-value structure
  • ARRAY Ordered values
  • FLATTEN Explode nested data
  • LATERAL Correlated inline view
  • Parquet Columnar file
  • UNLOAD Export with COPY
  • MATCH_BY_COLUMN Name-based mapping

Dynamic Table vs Task LO

Dynamic table

  • Declarative query
  • Target lag
  • Automated refresh

Task

  • Imperative SQL
  • Schedule control
  • Procedure calls
  • Desired state vs procedure

Automated Ingestion LO

  • Stream CDC offset
  • Task Scheduled SQL
  • Task graph DAG orchestration
  • Dynamic table Target-lag refresh

Transform Picker PE

  • Deduplicate rows → QUALIFY ( Window filter )
  • Nested JSON rows → FLATTEN ( Explode )
  • Apply source changes → MERGE ( Upsert )
  • Rebuild from query → CTAS ( New table )

Performance PE

  • Micro-partition Immutable storage unit
  • Pruning Skip partitions
  • Clustering key Improve pruning
  • Query Profile Execution diagnostics
  • Local cache Warehouse cache
  • Result cache Query result reuse
  • Search optimization Point lookup speed
  • QAS Query acceleration

Transformation PE

  • MERGE Upsert changes
  • QUALIFY Filter window results
  • Window function Partitioned calculation
  • CTAS Create from query

Sharing Roles CO

  • Provider publishes; consumer queries
  • Provider: data Consumer: access Listing: product

Time Travel vs Fail-safe CO

Time Travel

  • User queryable
  • Clone/restore
  • Up to 90 days

Fail-safe

  • Snowflake only
  • 7 days
  • After retention
  • User recovery vs support

Sharing Picker CO

  • Same-region account → Direct share ( No copy )
  • Cross-region consumers → Listing ( Auto-fulfill )
  • Public data product → Marketplace ( Public listing )
  • No Snowflake account → Reader account ( Provider managed )
  • Group governance → Data Exchange ( Curated members )
  • Privacy collaboration → Clean room ( Controlled joins )

Recovery + Cloning CO

  • Time Travel Historical query/restore
  • Fail-safe Snowflake recovery only
  • Zero-copy clone Metadata copy
  • Permanent Time Travel + Fail-safe
  • Transient No Fail-safe
  • Temporary Session scoped
  • UNDROP Restore dropped object
  • Retention Historical data window

Direct Share vs Listing CO

Direct share

  • Same region
  • Specific accounts
  • No marketplace

Listing

  • Any region
  • Rich metadata
  • Marketplace capable
  • Relationship vs product

Collaboration CO

  • Provider Shares data
  • Consumer Uses shared data
  • Share Secure data access
  • Direct share Same-region sharing
  • Listing Marketplace-style sharing
  • Marketplace Public listings
  • Reader account Non-customer access
  • Data Exchange Managed sharing group
  • Clean room Privacy-safe collaboration

Common Traps

Compute vs storage

  • Warehouses run queries ≠ Databases store objects

Power vs concurrency

  • Resize for heavy work ≠ Clusters for queued users

Batch vs continuous

  • COPY is bulk ≠ Snowpipe is continuous

CDC vs scheduler

  • Stream records changes ≠ Task executes SQL

Column vs row security

  • Masking changes values ≠ Row policy filters rows

Recovery window confusion

  • Time Travel is queryable ≠ Fail-safe needs Snowflake

Share vs copy

  • Sharing avoids copies ≠ Unload creates files

Direct vs marketplace

  • Direct share is narrow ≠ Listing is productized

Monitor vs budget

  • Resource monitor suspends ≠ Budget monitors spend

Last Minute

  • 1. Weights: 31/20/18/21/10
  • 2. Pass: 750 scaled
  • 3. Time: 115 minutes
  • 4. Warehouse = compute, not storage
  • 5. Database = objects, not compute
  • 6. COPY uses user warehouse
  • 7. Snowpipe uses serverless compute
  • 8. Streams track table changes
  • 9. Tasks execute scheduled SQL
  • 10. Fail-safe is Snowflake-only recovery
  • 11. Listings cross regions/clouds
  • 12. Direct shares same region
  • 13. Resize for heavy queries
  • 14. Clusters for concurrency queues