Snowflake SnowPro Core COF-C03 Practice Exam

SnowPro Core Certification COF-C03

The SnowPro Core Certification validates practical, hands-on experience with the Snowflake AI Data Cloud: its architecture, accounts and virtual warehouses, loading and transforming data, performance and data collaboration. The COF-C03 exam covers five domains: Snowflake AI Data Cloud Features & Architecture (31%), Account Management and Data Governance (20%), Data Loading, Unloading, and Connectivity (18%), Performance Optimization, Querying, and Transformation (21%), and Data Collaboration (10%). This practice exam mirrors its short scenario style with single-answer and multiple-select questions.

845 questions in the bank

Question bank updated : questions, answers, explanations and chapters.

Certification Overview

  • Exam name: SnowPro Core COF-C03
  • Vendor: Snowflake
  • Exam code: COF-C03
  • Duration: 115 minutes
  • Total questions: about 100
  • Official passing score: 750 on a scale of 0-1000 (source)
  • Our practice exam pass mark: 75%

Who Should Take This Exam?

Data Engineer, Data Analyst, Database Administrator, Data Architect, Analytics Engineer, BI Developer, Cloud Data Engineer, Solutions Architect

Prerequisites

No prerequisites. Snowflake recommends 6 or more months of experience using Snowflake.

Topics Covered

  • Snowflake AI Data Cloud Features & Architecture
  • Account Management and Data Governance
  • Data Loading, Unloading, and Connectivity
  • Performance Optimization, Querying, and Transformation
  • Data Collaboration

Question Types

  • Multiple Choice (Single Answer)
  • Multiple Choice (Multiple Answers)

SnowPro Core COF-C03 Practice Questions

Our question bank contains 845+ practice questions for this certification. Sample questions from each exam chapter. Expand a question to see the answer choices. With full access, you get unlimited practice exams with randomized questions from our full question bank.

Snowflake AI Data Cloud Features & Architecture

Query parsing, optimization, and access control checks in Snowflake are performed by which component?
  • A. The local disk cache of the virtual warehouse
  • B. The Database Storage layer
  • C. The Cloud Services layer Correct answer
  • D. The virtual warehouse that runs the query

Option C is correct because the Cloud Services layer handles query parsing, optimization and access control checks. The warehouse only executes the compiled plan, the storage layer holds the data, and the local disk cache just speeds up repeat reads.

How does Snowflake physically store table data?
  • A. In partitions that the table owner defines with a PARTITION BY clause
  • B. In row-oriented blocks of fixed size that are updated in place
  • C. On the local SSDs of the virtual warehouse that loaded the data
  • D. In compressed, columnar micro-partitions that are never modified in place Correct answer

Option D is correct because Snowflake stores table data as compressed, columnar micro-partitions that are immutable, so an update writes new micro-partitions. Users do not define partitions with PARTITION BY, and warehouse SSDs hold only a cache, not the table.

Which Snowflake edition is the lowest one that supports multi-cluster warehouses?
  • A. Business Critical Edition
  • B. Enterprise Edition Correct answer
  • C. Standard Edition
  • D. Virtual Private Snowflake (VPS)

Option B is correct because multi-cluster warehouses start at Enterprise Edition, and Business Critical and VPS include them too. Standard Edition is the one edition below that level without them.

Which TWO statements describe the Database Storage layer? (Select TWO).
  • A. Data is reachable only while at least one virtual warehouse is running
  • B. Users can read the underlying files directly through the cloud provider's console
  • C. Storage is billed separately from compute, by compressed bytes stored Correct answer
  • D. Data sits in cloud object storage in a compressed, columnar format Correct answer
  • E. Capacity must be provisioned per database before data can be loaded

Options C and D are correct because Database Storage keeps data in cloud object storage as compressed, columnar micro-partitions, and storage is billed on its own by compressed bytes stored. Option A is wrong because the data persists whether or not any warehouse is running. Options B and E are wrong because the files are managed by Snowflake and are not readable directly, and capacity grows automatically with no provisioning.

Account Management and Data Governance

Which Snowflake feature analyzes column contents and assigns semantic and privacy categories such as EMAIL or NAME?
  • A. Data lineage
  • B. Data classification Correct answer
  • C. The search optimization service
  • D. Object tagging

Option B is correct because data classification analyzes column contents and assigns semantic categories (such as EMAIL or NAME) and privacy categories as system tags. Data lineage traces how data moves between objects, the search optimization service speeds up selective lookups, and object tagging only attaches labels that someone sets, without analyzing the data.

Compliance wants card numbers hidden from analysts when queries run, while the payments role still reads the stored values unchanged. Which feature should be applied to the column?
  • A. An aggregation policy
  • B. A dynamic data masking policy Correct answer
  • C. A tag restricted by ALLOWED_VALUES
  • D. A row access policy

Option B is correct because a dynamic data masking policy rewrites the column value at query time based on the querying role, so analysts see masked card numbers while the payments role reads the stored values unchanged. A row access policy filters rows rather than hiding column values, an aggregation policy restricts queries to aggregates, and a tag restricted by ALLOWED_VALUES only limits which values a tag can hold.

An administrator needs one account-wide list of the columns and tables that masking and row access policies are attached to. Which ACCOUNT_USAGE view provides it?
  • A. POLICY_REFERENCES Correct answer
  • B. GRANTS_TO_ROLES
  • C. OBJECT_DEPENDENCIES
  • D. TAG_REFERENCES

Option A is correct because the POLICY_REFERENCES view in SNOWFLAKE.ACCOUNT_USAGE lists every object and column that a masking policy, row access policy or other policy is attached to. GRANTS_TO_ROLES shows privilege grants, OBJECT_DEPENDENCIES shows object-to-object dependencies such as views on tables, and TAG_REFERENCES shows tag assignments.

Which TWO statements describe data classification in Snowflake? (Select TWO).
  • A. It requires a masking policy on each column before it runs
  • B. It assigns semantic categories such as EMAIL to columns Correct answer
  • C. It replaces the stored column values with tokens
  • D. It reports privacy categories such as IDENTIFIER and QUASI_IDENTIFIER Correct answer
  • E. It encrypts classified columns with a separate key

Options B and D are correct because classification inspects column data and assigns a semantic category such as EMAIL and a privacy category such as IDENTIFIER or QUASI_IDENTIFIER. It only produces tags and results. It does not need a masking policy first, does not tokenize or change the stored values, and does not add a separate encryption key.

Data Loading, Unloading, and Connectivity

Which command writes the rows of a Snowflake table out as files in a stage?
  • A. PUT
  • B. COPY INTO <location> Correct answer
  • C. COPY INTO <table>
  • D. GET

Option B is correct because COPY INTO <location> unloads query results or table rows into files in a stage. PUT and GET only move files between a client machine and an internal stage, and COPY INTO <table> loads files into a table rather than writing them out.

A data engineer must move CSV files from a workstation into a named internal stage. Which command uploads them?
  • A. GET
  • B. PUT Correct answer
  • C. COPY INTO <table>
  • D. INSERT INTO ... SELECT

Option B is correct because PUT uploads files from the local file system to an internal stage. GET does the reverse by downloading staged files, COPY INTO <table> loads staged files into a table, and INSERT ... SELECT does not touch files.

Which Snowflake object runs a COPY INTO statement on a fixed cron schedule without any external orchestrator?
  • A. A pipe
  • B. A task Correct answer
  • C. A stream
  • D. A dynamic table

Option B is correct because a task runs a SQL statement, including COPY INTO, on a schedule or cron expression inside Snowflake. A pipe is event driven, a stream only tracks changes, and a dynamic table refreshes a query result instead of running a COPY.

Which TWO features can load staged files into a table automatically, without a person running a statement for each batch? (Select TWO).
  • A. A stream created on the target table
  • B. A task that runs COPY INTO on a schedule Correct answer
  • C. A dynamic table defined over the stage
  • D. A materialized view on the target table
  • E. A pipe with AUTO_INGEST = TRUE Correct answer

Options B and E are correct because a scheduled task can run COPY INTO for each new batch, and a pipe with AUTO_INGEST = TRUE loads files as event notifications arrive. A stream only records changes on a table, a dynamic table cannot be defined over a stage, and a materialized view does not load files.

Performance Optimization, Querying, and Transformation

Which command returns the logical plan of a statement without executing it?
  • A. SYSTEM$ESTIMATE_QUERY_ACCELERATION
  • B. EXPLAIN Correct answer
  • C. SYSTEM$CLUSTERING_INFORMATION
  • D. DESCRIBE RESULT

Option B is correct because EXPLAIN compiles a statement and returns its logical plan without running it, so no warehouse compute is used for execution. SYSTEM$ESTIMATE_QUERY_ACCELERATION estimates query acceleration benefit, SYSTEM$CLUSTERING_INFORMATION reports clustering depth, and DESCRIBE RESULT returns the column metadata of a previous result.

Which Snowsight page lists past statements and opens a graphical operator tree for each of them?
  • A. Warehouses, through the load monitoring chart of a warehouse
  • B. Copy History, through the Pipe Details tab of a load
  • C. Query History, through the Query Profile tab of a statement Correct answer
  • D. Resource Monitors, through the usage tab of a monitor

Option C is correct because Query History in Snowsight lists past statements, and its Query Profile tab shows the graphical operator tree with per-operator statistics. The Warehouses page charts load, Copy History tracks data loading, and Resource Monitors track credit usage, none of which show an operator tree.

A single UPDATE reports a TOTAL_ELAPSED_TIME of 95 seconds, yet compilation and execution add up to 4 seconds and both warehouse queued times are 0. Another session held an open transaction on the same table. Which column accounts for the gap?
  • A. QUERY_RETRY_TIME
  • B. QUEUED_REPAIR_TIME
  • C. LIST_EXTERNAL_FILES_TIME
  • D. TRANSACTION_BLOCKED_TIME Correct answer

Option D is correct because TRANSACTION_BLOCKED_TIME records the milliseconds a statement waited on a lock held by another transaction, which explains elapsed time that compilation, execution and warehouse queueing do not. QUERY_RETRY_TIME covers internal retries, QUEUED_REPAIR_TIME covers warehouse repair waits, and LIST_EXTERNAL_FILES_TIME covers listing files for external tables.

An unchanged benchmark SELECT returns in under a second on every run after the first. Which TWO steps make later runs measure the real work of the query? (Select TWO).
  • A. Set USE_CACHED_RESULT to FALSE for the session Correct answer
  • B. Suspend and resume the warehouse before each run Correct answer
  • C. Enable the search optimization service on the table
  • D. Set AUTO_SUSPEND to 0 so the warehouse stays up
  • E. Move the statement to a warehouse one size larger

Options A and B are correct because a benchmark only measures the query's real work when neither the persisted result cache nor the warehouse's local disk cache can answer it. Setting USE_CACHED_RESULT to FALSE stops the result cache from returning the earlier result, and suspending then resuming the warehouse empties its local cache. Search optimization, keeping the warehouse running with AUTO_SUSPEND 0, and a larger warehouse do not remove the caching effect, and keeping the warehouse up actually preserves the warm cache.

Data Collaboration

To begin Secure Data Sharing, which statement creates the object that tracks the shared objects and the accounts allowed to see them?
  • A. CREATE DATABASE ROLE
  • B. CREATE SHARE Correct answer
  • C. CREATE REPLICATION GROUP
  • D. CREATE APPLICATION PACKAGE

Option B is correct because CREATE SHARE creates the share object, which records the granted database objects and the consumer accounts allowed to see them. CREATE DATABASE ROLE defines a role inside a database, CREATE REPLICATION GROUP copies objects between accounts, and CREATE APPLICATION PACKAGE builds a Native App package, so none of those tracks a data share.

Can a consumer account add a database that was created from a share to a share of its own?
  • A. Yes, if it adds only secure views over the shared tables
  • B. No, shared databases cannot be added to another share Correct answer
  • C. Yes, once it holds the CREATE SHARE privilege
  • D. Yes, if the third account is in the same region

Option B is correct because a database created from a share is read-only for the consumer and cannot be added to another share, so data cannot be re-shared downstream. The consumer can build its own secure views over the shared data in a local database and share those, but the shared database itself stays non-shareable, whatever privileges or regions are involved.

Which command lists the outbound shares in a provider account together with the accounts they are granted to?
  • A. SHOW MANAGED ACCOUNTS
  • B. SHOW LISTINGS
  • C. SHOW REPLICATION GROUPS
  • D. SHOW SHARES Correct answer

Option D is correct because SHOW SHARES lists the inbound and outbound shares visible to the account, including the accounts an outbound share is granted to. SHOW MANAGED ACCOUNTS lists reader accounts, SHOW LISTINGS lists listings, and SHOW REPLICATION GROUPS lists replication objects, so none of them reports share grants.

Which TWO actions can a consumer account perform on a database created from a share? (Select TWO).
  • A. Clone the imported database for testing
  • B. Join its tables to tables in the consumer's own databases Correct answer
  • C. Change the data retention period on a shared table
  • D. Query its tables with a warehouse in the consumer account Correct answer
  • E. Insert new rows into a shared table

Options B and D are correct because a database created from a share is read-only but can be queried with a consumer warehouse and joined to local data. It cannot be cloned, its data cannot be changed, and its retention cannot be altered.

Free Study Guide

Read the free Snowflake SnowPro Core COF-C03 study guide: every exam topic explained in plain English, with diagrams, exam tips and the mistakes to avoid.

More Snowflake Practice Exams

All Snowflake practice exams →

Practice with realistic mock exams to prepare for your Snowflake certification.