Snowflake • COF-C03
SnowPro Core Certification (COF-C03) Practice Test
Validates hands-on expertise with the Snowflake AI Data Cloud, covering architecture, data loading, performance optimization, governance, and data collaboration. Designed for data engineers, DBAs, and cloud professionals working with Snowflake.
Exam Details
Questions
592
Duration
115 minutes
Passing Score
750/1000
Difficulty
AssociateLast Updated
May 2026
Topics Covered
Snowflake AI Data Cloud Features and ArchitectureAccount Management and Data GovernanceData Loading, Unloading, and ConnectivityPerformance Optimization, Querying, and TransformationData Collaboration
Exam Domain Breakdown
Snowflake AI Data Cloud Features and Architecture31%
Account Management and Data Governance20%
Data Loading, Unloading, and Connectivity18%
Performance Optimization, Querying, and Transformation21%
Data Collaboration10%
Exam Overview
The SnowPro Core Certification (COF-C03) is Snowflake's foundational technical credential, validating practical, hands-on expertise with the Snowflake AI Data Cloud. It covers the full breadth of Snowflake's platform: its unique multi-cluster, shared-data architecture that separates storage, compute, and cloud services; account and virtual warehouse management; structured, semi-structured, and unstructured data handling; performance monitoring and optimization; Role-Based Access Control (RBAC) and data governance; and secure data collaboration and sharing.
The COF-C03 version launched on February 16, 2026, replacing the retired COF-C02 exam. A key update in this version is an expanded emphasis on Snowflake's AI and ML capabilities, including Cortex AI, Snowpark, Iceberg tables, Notebooks, and Git integration — reflecting Snowflake's strategic evolution toward AI-driven data workflows. The certification is valid for two calendar years from the date of passing.
Who Should Take This Exam
The SnowPro Core Certification is designed for data engineers, database administrators, cloud architects, data analysts, and BI professionals who work with Snowflake in a hands-on capacity. It is particularly well-suited for professionals who build and maintain data pipelines, manage Snowflake accounts, optimize query performance, or implement data governance policies on the platform.
Snowflake recommends candidates have at least six months of practical experience with Snowflake before attempting the exam. This means the certification is appropriate for working professionals with meaningful platform exposure rather than those entirely new to cloud data warehousing. It also serves as a prerequisite stepping stone toward Snowflake's advanced certifications, including SnowPro Advanced: Data Engineer, Administrator, and Architect.
Prerequisites
There are no formal prerequisites required to register for the COF-C03 exam. However, Snowflake strongly recommends that candidates have a minimum of six months of hands-on experience working with the Snowflake platform before sitting for the exam, as questions are scenario-based and require applied knowledge rather than theoretical recall.
Candidates should be comfortable with SQL fundamentals, core cloud computing concepts (particularly as they apply to one or more of AWS, Azure, or GCP), and key Snowflake constructs such as virtual warehouses, stages, file formats, micro-partitions, and the COPY INTO command. Familiarity with data loading patterns (including Snowpipe for continuous ingestion), RBAC security models, and Snowflake's data sharing and collaboration features is also expected.
Exam Format
The COF-C03 exam consists of approximately 100 questions in multiple-choice and multiple-select formats. Candidates have 115 minutes to complete the exam. The exam is delivered in a proctored environment through Pearson VUE, available as either an online proctored session (via OnVUE) or at a physical Pearson VUE testing center.
Scoring uses a scaled score system with a maximum of 1,000 points; the minimum passing score is 750 out of 1,000 (75%). The exam costs $175 USD per attempt. The certification remains valid for two calendar years, after which candidates must recertify to maintain their status.
Skills Measured
- 1.Domain 1 — Snowflake AI Data Cloud Features and Architecture (31%): Covers Snowflake's multi-cluster, shared-data architecture including the separation of storage, compute, and cloud services layers. Topics include virtual warehouses, micro-partitions, data caching (result cache, local disk cache, remote disk cache), clustering keys, Snowflake editions, supported cloud platforms, and AI/ML capabilities such as Cortex AI, Snowpark, Iceberg tables, Notebooks, and Git integration.
- 2.Domain 2 — Account Management and Data Governance (20%): Covers Snowflake account configuration, resource monitors, virtual warehouse sizing and scaling policies, Role-Based Access Control (RBAC), user and role management, network policies, encryption, data masking, row access policies, object tagging, and compliance features supporting data governance frameworks.
- 3.Domain 3 — Data Loading, Unloading, and Connectivity (18%): Covers methods for ingesting data into and exporting data from Snowflake, including the COPY INTO command, internal and external stages, file format objects, Snowpipe for continuous/event-driven ingestion, Kafka and cloud storage connectors, Snowflake connectors for Python, Spark, JDBC/ODBC, and SnowSQL CLI.
- 4.Domain 4 — Performance Optimization, Querying, and Transformation (21%): Covers query profiling with the Query Profile tool, use of clustering and search optimization, warehouse sizing and multi-cluster warehouse configuration, result caching, materialized views, streams, tasks, stored procedures, Snowpark for data transformation in Python/Java/Scala, and interpreting EXPLAIN plans.
- 5.Domain 5 — Data Collaboration (10%): Covers Snowflake's data sharing capabilities including secure data shares, the Snowflake Marketplace, Data Clean Rooms, listing and consuming shared datasets, reader accounts, and collaboration models that allow sharing live data without copying or moving it.
Study Tips
- Download and thoroughly review the official COF-C03 Exam Study Guide from learn.snowflake.com — it outlines all five domains, their exact percentage weights, and the specific objectives tested, making it the single most authoritative preparation resource.
- Enroll in Snowflake University's 'SnowPro Core Certification Preparation' course and the 'Level Up: SnowPro Core' series, both available free on learn.snowflake.com, as they are purpose-built for this exam and cover the full domain blueprint.
- Activate a free Snowflake trial account and practice hands-on with COPY INTO, Snowpipe, virtual warehouse scaling, RBAC role hierarchies, and data sharing — the exam is scenario-based and rewards applied knowledge over rote memorization.
- Since Domain 1 (Architecture, 31%) and Domain 4 (Performance, 21%) together account for more than half the exam, prioritize understanding Snowflake's three-layer architecture, micro-partition internals, caching tiers, and query profiling with the Query Profile tool.
- Pay special attention to new COF-C03 topics not covered in COF-C02: Cortex AI functions, Snowflake Notebooks, Iceberg table support, and Git integration — these reflect Snowflake's AI Data Cloud positioning and are explicitly added to the updated exam.
- Use Snowflake's official documentation as your reference for precise command syntax, especially for COPY INTO options, file format parameters, stage types, resource monitor configurations, and warehouse properties — the exam tests specific parameter knowledge.
- Simulate exam conditions by taking timed practice tests under 115 minutes with the exam interface disabled; focus on eliminating incorrect multiple-select answers carefully, as partial credit is not awarded for multi-select questions.
Career Benefits
The SnowPro Core Certification is widely recognized as the foundational credential for data professionals working on the Snowflake platform, which has become one of the most widely adopted cloud data platforms across industries. Certified professionals qualify for roles including Snowflake Data Engineer, Cloud Data Architect, BI Engineer, Data Platform Consultant, and Solutions Engineer — roles that frequently list the SnowPro Core as a required or preferred qualification on job postings.
As of 2026, mid-level Snowflake data engineers typically earn $110,000–$140,000 annually, with senior engineers commanding $140,000–$180,000 base salaries, and contract rates ranging from $95–$135 per hour. Certification holders report salary increases of 20–40% compared to non-certified peers. The SnowPro Core also serves as the gateway to Snowflake's advanced-tier certifications (priced at $375 per attempt), including SnowPro Advanced: Data Engineer, Architect, and Administrator, which unlock higher-compensation specialist roles. Complementary skills in dbt, Snowpark, Python, Terraform, and AWS further amplify the market value of this certification.
Sample Questions
Preview — answers shown5 sample questions with correct answers and explanations. Start a practice session to test yourself across all 592 questions.
1. A Snowflake security administrator at Contoso Corp configures an account-level network policy with an ALLOWED_IP_LIST containing the corporate CIDR range 10.0.0.0/8. After discovering that a specific IP address 10.25.100.50 was used in a recent unauthorized access attempt, the administrator adds that exact address to the BLOCKED_IP_LIST. What happens when 10.25.100.50 attempts to connect to the Snowflake account? (Select one!)
AThe connection is allowed because 10.25.100.50 falls within the 10.0.0.0/8 CIDR range in ALLOWED_IP_LIST
BSnowflake raises a configuration error requiring the administrator to remove the IP from one of the lists
CThe connection is blocked because BLOCKED_IP_LIST takes precedence over ALLOWED_IP_LIST
DThe behavior is unpredictable when an IP address appears in both lists simultaneously
Explanation
In Snowflake network policies, BLOCKED_IP_LIST always takes precedence over ALLOWED_IP_LIST. When an IP address matches an entry in both lists — whether as an exact match or because it falls within an allowed CIDR range — the blocked list wins and the connection attempt is denied. This deterministic behavior is intentional: it allows administrators to carve out specific blocked addresses from a broader permitted range without restructuring the entire allowed list. No configuration error is raised; the conflict resolution rule is well-defined. No valid credentials will override a block imposed by BLOCKED_IP_LIST.
2. A data governance team at Litware Financial Services needs to implement column-level data protection for a TRANSACTIONS table. The SSN column must display 'XXX-XX-XXXX' for all roles except the DATA_STEWARD role, which must see the actual values. The solution must integrate with Snowflake's native governance framework and require no changes to the underlying table structure. Which feature should they implement? (Select one!)
AEnable Tri-Secret Secure on the account to encrypt the SSN column and configure decryption key access only for the DATA_STEWARD role
BCreate a Dynamic Data Masking policy that returns the actual SSN when IS_ROLE_IN_SESSION('DATA_STEWARD') is TRUE and returns 'XXX-XX-XXXX' otherwise, then apply the policy to the SSN column
CCreate a Row Access Policy that allows the DATA_STEWARD role to access rows containing SSN data and restricts all other roles from viewing those rows
DCreate a Secure View with a CASE expression on CURRENT_ROLE() to selectively display the SSN column and grant access to that view instead of the base table
Explanation
Dynamic Data Masking (DDM) applies column-level transformations at query runtime without altering the underlying data or table structure. A masking policy function can use IS_ROLE_IN_SESSION() to evaluate the active role and return either the original value or a masked substitute, allowing the DATA_STEWARD role to see real SSN values while all other roles receive the masked placeholder. This integrates directly with Snowflake's column-level security framework and requires no table structure changes. Row Access Policies control which rows are returned from a table but cannot transform or mask specific column values — they would hide entire rows rather than masking the SSN field for certain roles. Secure Views require managing a separate database object that users must query instead of the base table, adding complexity and bypassing the native column-level security governance model. Tri-Secret Secure is an at-rest encryption feature using customer-managed keys that protects data from Snowflake staff access; it does not provide runtime column masking based on role membership.
3. A security administrator at Northwind Financial is configuring a Snowflake network policy to restrict access to their corporate network IP range 192.168.0.0/16. After discovering that a compromised laptop assigned IP address 192.168.45.78 is attempting unauthorized access, they add that specific IP to the BLOCKED_IP_LIST while keeping the entire 192.168.0.0/16 range in the ALLOWED_IP_LIST. What will happen when a connection attempt comes from 192.168.45.78? (Select one!)
ASnowflake will return an error because the same IP address cannot logically exist in both the allowed and blocked lists simultaneously
BThe connection will be blocked because the BLOCKED_IP_LIST takes precedence over the ALLOWED_IP_LIST
CThe connection will trigger an MFA challenge before deciding whether to allow or block based on successful authentication
DThe connection will be allowed because 192.168.45.78 falls within the permitted 192.168.0.0/16 range in the ALLOWED_IP_LIST
Explanation
In Snowflake network policies, the BLOCKED_IP_LIST always takes precedence over the ALLOWED_IP_LIST. If an IP address matches any entry in the BLOCKED_IP_LIST, the connection is denied regardless of whether that address also falls within a range listed in the ALLOWED_IP_LIST. This design ensures that explicit denials cannot be overridden by overly broad allow rules. Snowflake does not validate for overlap between the two lists and will not generate an error if the same IP appears in both. Network policies are also evaluated independently of authentication mechanisms such as MFA — they control network-level access before any authentication challenge is initiated.
4. A data science team at Fabrikam Analytics is building ML inference pipelines in Snowflake using Snowpark Python. Their models require loading large feature matrices into memory for batch scoring, and they consistently encounter out-of-memory errors when running on a standard Large virtual warehouse. What change should they make to resolve the memory errors? (Select one!)
ASwitch to a Snowpark-optimized warehouse of size Medium or above
BIncrease the warehouse size to 2X-Large to provision additional compute nodes in the cluster
CEnable the Query Acceleration Service on the existing warehouse
DConfigure a multi-cluster warehouse to increase the total number of available compute nodes
Explanation
Snowpark-optimized warehouses provide 16x more memory per node compared to standard virtual warehouses and are specifically designed for memory-intensive operations such as ML model training and inference with large feature datasets. They are available starting at size Medium. Increasing to a 2X-Large standard warehouse provisions more nodes in the cluster but does not increase per-node memory, so individual memory-intensive Snowpark operations will still hit the same per-node memory ceiling. Query Acceleration Service helps parallelize portions of large analytical SQL queries but does not address per-node memory constraints for Snowpark Python workloads. Multi-cluster warehouses address query concurrency by spinning up additional clusters during peak load, which does not resolve the per-node memory requirement for a single ML inference operation.
5. A DevOps engineer at Contoso Corp manages key-pair authentication for automated ETL pipelines connecting to Snowflake. The security team requires periodic rotation of all authentication credentials. The engineer needs to rotate the RSA key pair for a service user without causing any downtime to the production pipeline. Which approach should the engineer take? (Select one!)
AGenerate a new RSA key pair, temporarily disable the service user account, update both the Snowflake user and pipeline configuration, then re-enable the account
BGenerate a new RSA key pair, create a new service user with the new key pair, update the pipeline to use the new credentials, then delete the old user
CContact Snowflake Support to perform an atomic key rotation that swaps both keys simultaneously to prevent authentication gaps
DGenerate a new RSA key pair, assign the new public key using the RSA_PUBLIC_KEY_2 property on the service user, update the pipeline to use the new private key, then remove the old public key
Explanation
Snowflake supports two simultaneous active RSA public keys per user through the RSA_PUBLIC_KEY and RSA_PUBLIC_KEY_2 properties, specifically designed for zero-downtime key rotation. By assigning the new public key to RSA_PUBLIC_KEY_2 while the pipeline continues authenticating with the existing RSA_PUBLIC_KEY, the engineer can update the application's private key, verify successful authentication, and then remove the old public key—all without any service interruption. Disabling the account during rotation creates an outage window that violates the zero-downtime requirement. Creating a new service user adds unnecessary credential management complexity and is not the intended mechanism. Snowflake does not offer an atomic key rotation service through Support; the dual-key mechanism is the standard self-service approach for seamless rotation.
$7
One-time access to this exam
Full access to all 592 questions
Or $15/mo for all 227 exams
Detailed explanations
Free preview stays available