Practice questions · Data & Analytics
SnowPro Core Certification (COF-C03): Practice Questions
Original practice questions for the SnowPro Core Certification (COF-C03). Each answer is explained, including why each other option is wrong. Filter by domain or difficulty. These are concept checks - not questions from the certification.
Answered 0 · Correct 0
-
Snowflake's architecture separates compute and storage into independent layers. Which component actually runs your queries?
Correct answer: A. A virtual warehouse is the compute cluster that executes queries and loads, and it can be scaled independently of storage. A micro-partition is a unit of stored data, not a compute engine; the storage layer holds data but does not process it; and a stage is a location for data files, not a query processor. -
How does Snowflake physically store the data in a table?
Correct answer: B. Snowflake automatically divides table data into many small, immutable, columnar micro-partitions, whose metadata enables pruning. It is not one large row-based file; it is not stored as a CSV in a stage (a stage holds load/unload files, not the table itself); and the warehouse cache only holds transient data for a running warehouse, not the permanent table. -
Which layer of Snowflake's architecture handles authentication, query optimisation and metadata?
Correct answer: C. The cloud services layer is the coordinating 'brain' that manages authentication, metadata, query optimisation and security. The storage layer only persists data; a virtual warehouse only provides compute for queries; and a reader account is a sharing mechanism, not an architectural layer. -
You need a single query to run faster because it processes a very large amount of data. What is the most direct action?
Correct answer: D. Scaling up to a larger warehouse size gives a single large query more compute resources, which is the direct fix for heavy individual queries. Adding clusters (scaling out) helps concurrency - many users at once - not one big query; Fail-safe is a data-recovery feature, unrelated to speed; and a reader account is for sharing data with non-customers. -
Many users are running queries at the same time and some are queuing. Which approach addresses this concurrency problem?
Correct answer: A. A multi-cluster warehouse scales out by adding clusters to serve more concurrent queries, which is exactly what queuing under load needs. Converting tables to external stages does not add compute; increasing Time Travel retention only affects data recovery history; and dropping a clustering key would, if anything, hurt query performance. -
A user re-runs the exact same query 10 minutes later with no data changes. Which Snowflake feature can return the result without using a warehouse?
Correct answer: D. The results cache returns the stored output of an identical query (within 24 hours, if nothing changed) without spinning up warehouse compute. Fail-safe is a data-recovery period, not a cache; a network policy controls access by IP address; and a sequence generates numbers - none of these serve cached results. -
What is the main purpose of defining a clustering key on a very large table?
Correct answer: C. A clustering key co-locates related rows across micro-partitions so the engine can prune more partitions and scan less data on filtered queries. It does not handle encryption (that is automatic and separate); it does not grant access (that is RBAC); and it has nothing to do with deleting the table. -
How is compute cost measured when a virtual warehouse is running?
Correct answer: B. A running warehouse consumes credits in proportion to its size and the time it runs, so suspending idle warehouses saves money. Stored micro-partitions relate to storage cost, not compute; there is no flat per-query rate; and the number of roles has no bearing on compute billing. -
Which command bulk-loads data files from a stage into a Snowflake table?
Correct answer: B. COPY INTO is the command that bulk-loads files from a stage into a table (and can also unload data out). FLATTEN expands semi-structured data into rows; CREATE SHARE sets up secure data sharing; and ALTER WAREHOUSE changes compute settings - none of these load files. -
In Snowflake, what is a stage?
Correct answer: A. A stage is a location - internal to Snowflake or external such as an S3 bucket - where data files sit before loading or after unloading. A compute cluster is a virtual warehouse; a role is part of RBAC; and cached results live in the results cache - none of these are a stage. -
You need data to load continuously, automatically, as small files arrive in a stage rather than in scheduled batches. Which feature fits?
Correct answer: C. Snowpipe loads data continuously and automatically as files land in a stage, ideal for near-real-time ingestion. A materialized view precomputes query results; Time Travel recovers historical data; and a clustering key improves query pruning - none of these perform continuous loading. -
An external stage in Snowflake refers to:
Correct answer: D. An external stage points to files held in your own cloud storage, such as an S3 bucket, rather than inside Snowflake. A stage inside Snowflake-managed storage is an internal stage; a warehouse in another region is unrelated to staging; and a role for an external user is a security concept, not a stage. -
When loading files, what does a named file format object define?
Correct answer: B. A file format object describes how Snowflake should parse files - the type (CSV, JSON and so on), delimiters, headers and compression - so loads are consistent. It does not control who can query the table (RBAC), does not set warehouse size, and has nothing to do with Time Travel retention. -
How do you export data out of a Snowflake table into files in a stage?
Correct answer: C. The same COPY INTO command unloads data from a table out to files in a stage. A network policy restricts access by IP; a clustering key improves query performance; and granting PUBLIC is a security action - none of these export data to files. -
Which Snowflake data type is designed to store semi-structured data such as JSON in a single column?
Correct answer: C. VARIANT stores semi-structured data such as JSON, Avro or Parquet within one column, which you can then query without a fixed schema. NUMBER holds numeric values, BOOLEAN holds true/false, and TIMESTAMP holds date-time values - none of them store nested semi-structured documents. -
You have a VARIANT column containing a JSON array and need each array element as its own row. Which function helps?
Correct answer: B. FLATTEN expands a nested array or object in semi-structured data into separate rows, which is exactly what is needed here. COPY INTO loads or unloads files; CREATE WAREHOUSE provisions compute; and a masking policy hides sensitive values - none of them unnest array elements. -
Which object presents data as a virtual table defined by a saved query, computing results when queried?
Correct answer: A. A view is a stored query that behaves like a virtual table, returning current results each time it is queried. A stage holds files; a warehouse provides compute; and a network policy controls access by IP - none of these present a query as a virtual table. -
Why might you choose a secure view over a standard view?
Correct answer: D. A secure view hides its SQL definition and prevents some query optimisations that could expose underlying rows, making it appropriate for sharing sensitive data. It does not speed up loading from a stage, does not scale warehouses, and does not change Time Travel retention. -
Which object generates unique, sequential numbers, often used for surrogate keys?
Correct answer: B. A sequence produces unique, increasing numbers that are commonly used to generate surrogate key values. A stage holds files; a reader account provides shared access for non-customers; and a clustering key improves pruning - none of them generate sequential numbers. -
In Snowflake's role-based access control (RBAC), how are privileges given to users?
Correct answer: A. In Snowflake RBAC, privileges are granted to roles, and roles are granted to users (and to other roles), so access flows through the role hierarchy. Privileges are not attached to micro-partitions, are not limited to warehouse settings, and have nothing to do with the results cache. -
Which of these is a Snowflake system-defined role with broad administrative power over the whole account?
Correct answer: C. ACCOUNTADMIN is the top system-defined role with account-wide administrative authority, so it should be used sparingly. PUBLIC is a system role but a minimal, default one with limited privileges; VARIANT is a data type; and SNOWPIPE is not a role at all - it is the continuous-loading feature. -
You want certain users to see only masked values in a sensitive column while others see the real data. Which feature applies?
Correct answer: D. A dynamic data masking policy masks column values at query time based on the querying role, so different roles see different results. Snowpipe loads data; a clustering key improves pruning; and zero-copy cloning makes instant copies - none of them mask column values by role. -
What does a network policy control in Snowflake?
Correct answer: A. A network policy restricts account access by allowing or blocking specific IP address ranges, strengthening account security. It does not determine micro-partition counts (automatic), does not set warehouse size, and is unrelated to flattening JSON. -
A table had rows deleted by mistake an hour ago. Which Snowflake feature lets you query or restore the data as it was before the delete?
Correct answer: A. Time Travel lets you query or restore data as it existed within the retention window, so you can recover from an accidental delete. Snowpipe is for loading; a network policy controls IP access; and a materialized view precomputes results - none of them recover historical data. -
How does Fail-safe differ from Time Travel?
Correct answer: B. Fail-safe is a distinct, Snowflake-managed seven-day window that begins after Time Travel ends and can only be used by Snowflake to recover data - users cannot query it. It is not a faster Time Travel, not a query cache, and not the same as a zero-copy clone, which is an instant copy of an object. -
What is the key advantage of a zero-copy clone when you duplicate a large table?
Correct answer: C. A zero-copy clone shares the original micro-partitions, so it is created almost instantly and consumes no additional storage until either copy is modified. It does not compress the table, does not mask columns, and does not make queries on the original faster - it simply makes a cheap, fast copy. -
With Snowflake secure data sharing, how does a consumer account access the shared data?
Correct answer: D. Secure data sharing gives the consumer live, read-only access to the provider's objects without copying or moving any data. It is not a nightly file copy, does not require rebuilding from a CSV, and does not involve emailing a backup - the whole point is no data movement. -
A data provider wants to share data with a business partner who is not a Snowflake customer. What can the provider set up?
Correct answer: B. A reader account is created and managed by the provider so a non-Snowflake party can read shared data without their own Snowflake contract. A clustering key improves query pruning; a results cache is automatic and internal; and a larger warehouse is just compute - none of these enable sharing with a non-customer. -
Why are Snowflake editions (such as Standard and Enterprise) important to know for the exam?
Correct answer: A. Editions define which features and service limits you get - for example, higher editions extend the maximum Time Travel retention and add security features. They do not change the SQL dialect, do not cap table row counts in that way, and have nothing to do with your web browser. -
What is the difference between scaling a virtual warehouse 'up' and scaling it 'out'?
Correct answer: D. Scaling up increases the warehouse size to give a single heavy query more power, while scaling out (a multi-cluster warehouse) adds clusters to handle more concurrent users. The description that has scaling up adding clusters and scaling out enlarging the warehouse reverses the two definitions; they are not the same thing; and neither one archives data or deletes micro-partitions - those are unrelated to warehouse scaling.
Practice questions FAQ
- Are these real SnowPro Core exam questions?
- No. These are original study questions written to test understanding. They are not real exam questions, exam dumps, or copied from any provider.
- How should I use these practice questions?
- Answer each one, read the explanation (including why the wrong options are wrong), and use the per-domain score below to focus your revision on weak areas. Revisit before exam day.
- How many questions should I do before the exam?
- Enough to score consistently across every domain, alongside full-length practice from official or reputable providers. Understanding why each answer is right matters more than raw volume.
- What score means I am ready?
- A good signal is consistently scoring around 80% or higher across all domains on questions you have not seen before, and being able to explain why the wrong options are wrong.
- Should I use exam dumps?
- No. Dumps (real or leaked questions) breach provider policy, can void your certification, and do not build the understanding the exam actually tests.