COF-C03 SnowPro Core Certification 2026 Exam Question and Answers

 Automatic Clustering is a feature that allows users to control the organization of data within micro-partitions in Snowflake. By defining clustering keys, Snowflake can automatically reorganize the data in micro-partitions to optimize query performance1.

The query that contains a Snowflake hosted file URL in a directory table for a stage named bronzestage is select * from directory(@bronzestage). This query retrieves a list of all files on the stage along with metadata, including the Snowflake file URL for each file3.

 According to my knowledge, a security administrator can create multiple network policies, but only one network policy can be active for an account or specific user at any given time. This ensures that there is a clear and consistent policy being applied without conflicts. References: Based on my internal knowledge as of 2021.

The statistics that can be used to identify queries with inefficient pruning are ‘Partitions scanned’ and ‘Partitions total’. These statistics indicate how much of the data was actually needed and scanned versus the total available, which can highlight inefficiencies in data pruning34.

To address query concurrency issues, a user can add additional clusters to the virtual warehouse. This allows for the distribution of queries across multiple clusters, reducing the load on any single cluster and improving overallquery performance2.

Snowflake’s architecture is divided into three layers: database storage, query processing, and cloud services. The metadata, which includes information about the structure of the data, the SQL operations performed, and the service-level policies, is stored in the cloud services layer. This layer acts as the brain of the Snowflake environment, managing metadata, query optimization, and transaction coordination.

 To change permissions for new objects in a managed access schema, a user should be granted the MANAGE GRANTS global privilege. This privilege allows the user to manage access control through grants on all securable objects within Snowflake2. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The details section of a Query Profile in Snowflake provides users with various statistics and information about the execution of a query. One of the key pieces of information that can be determined from this section is the total duration of the query, which helps in understanding the performance and identifying potential bottlenecks. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The DATA_SHARING_USAGE schema includes views that display information about listings published in the Snowflake Marketplace or a data exchange, which includes DATA_TRANSFER_HISTORY and LISTING_TELEMETRY_DAILY2.

 An outer join, specifically a left outer join, will list all rows from the left table and match them with rows from the right table. If there is no match, the result will still include the row from the left table, with NULLs for columns from the right table. References: Based on general SQL knowledge as of 2021.

The Snowflake feature that allows for small volumes of data to be continuously loaded into Snowflake and incrementally made available for analysis is Snowpipe. Snowpipe is designed for near-real-time data loading, enabling data to be loaded as soon as it’s available in the storage layer3

When a database is cloned in Snowflake, it does not retain any privileges that were granted on the source object. The clone will need to have privileges reassigned as necessary for users to access it. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

 In Snowflake, a User-Defined Table Function (UDTF) can be accessed in the FROM clause of a query. UDTFs return a set of rows with one or more columns, which can be queried like a regular table

 A user with the OWNERSHIP privilege on a user object or the CREATE USER privilege on the account can create or manage other users in a Snowflake account56.

 To grant a role the privilege to select data from all current and future tables in a schema, two separate commands are needed. The first command grants the SELECT privilege on all existing tables within the schema, and the second command grants the SELECT privilege on all tables that will be created in the future within the same schema. 

 In a standard multi-cluster warehouse with auto-scaling, a cluster will shut down when, after 2-3 consecutive checks, the system determines that the load on the least-loaded cluster could be redistributed to other clusters. This ensures efficient resource utilization and cost management. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

In Snowflake, when queries begin to queue in a multi-cluster virtual warehouse, an additional compute cluster starts immediately if the warehouse is set to auto-scale mode. This mode allowsSnowflake to automatically add or resume additional clusters as soon as the workload increases, and similarly, shut down or pause the additional clusters when the load decreases

Clones in Snowflake have the advantage of being created almost instantly and saving space by not duplicating storage. This is due to Snowflake’s zero-copy cloning feature, which allows for the creation of object clones without the additional storage costs typically associated with data duplication23. Clones are independent of the original table and do not stay in sync with it, nor do they inherently have better query performance. However, they do inherit the time travel history from the original table at the time of cloning

Snowflake’s documentation recommends producing the largest files possible for data loading, as larger files reduce the number of files to process and the overhead associated with handling many small files. Thisapproach can maximize the load performance by leveraging Snowflake’s ability to ingest large files efficiently1. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The SnowSQL file that can store connection information is named ‘config’. It is used to store user credentials and connection details for easy access to Snowflake instances. References: Based on general database knowledge as of 2021.

 In a Query Profile, the statistical information that indicates a query is too large to fit in memory includes bytes spilled to local cache and bytes spilled to local storage. These metrics suggest thatthe working data set of the query exceeded the memory available on the warehouse nodes, causing intermediate results to be written to disk

Snowflake uses a technique called pruning to limit the number of micro-partitions scanned by each query. Pruning effectively filters out unnecessary micro-partitions based on the query’s filter conditions, which can significantly improve query performance by reducing theamount of data scanned1.

 SnowCD (Snowflake Connectivity Diagnostic Tool) is the best tool provided by Snowflake for troubleshooting network connectivity issues. It helps diagnose and resolve issues related to connecting to Snowflake services

https://docs.snowflake.com/en/user-guide/snowcd.html#:~:text=SnowCD%20(i.e.%20Snowflake%20Connectivity%20Diagnostic,their%20network%20connection%20to%20Snowflake.

 Running EXPLAIN and SHOW commands, as well as running a custom query, consume virtual warehouse credits in the Snowflake environment. These activities require computational resources, and therefore, credits are used to account for the usage of these resources. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

Pre-signed URLs in Snowflake allow users to access unstructured data without the need for authentication into Snowflake or passing an authorization token. These URLs are open and can be directly accessed ordownloaded by any user or application, making them ideal for business intelligence applications or reporting tools that need to display unstructured file contents

Efficient pruning in Snowflake is indicated by the number of partitions scanned out of the total available. If a small percentage of partitions are scanned, it suggests thatthe pruning process is effectively narrowing down the data, which can lead to improved query performance

The Activity area of Snowsight includes the Query History page, which allows users to monitor and view details about queries executed in their account, including performance data1. It also includesthe Automatic Clustering History, which provides insights into the automatic clustering operations performed on tables2.

The Cloud Services layer in Snowflake is responsible for various services, including security (like authentication and authorization) and metadata management (like query parsing and optimization). References: Based on general cloud architecture knowledge as of 2021.

To allow secure views the ability to reference data in multiple databases, the REFERENCE_USAGE privilege must be granted on each database that contains objects referenced by the secure view2. This privilege is necessary before granting the SELECT privilege on a secure view to a share.

 The OBJECT_CONSTRUCT function is used with the COPY INTO command to convert the rows in a relational table to a single VARIANT column, which can then be unloaded into a file. This is useful for transforming table data into a semi-structured JSON format

The VARIANT data type in Snowflake can store multiple types of data structures, as it is designed to hold semi-structured data. It can contain any other data type, including OBJECT and ARRAY, which allows it to represent various data structures

The OPERATE privilege is required for a role to resume a suspended warehouse if auto-resume is not enabled. This privilege allows the role to start, stop, suspend, or resume a virtual warehouse3.

 Snowflake allows data providers to share data with consumers across different cloud platforms and regions through database replication. The data provider must replicate the database to a secondary account in the target region or cloud platform within the same organization, and then create a share to the data consumer’s account. This process ensures that the data is available in the consumer’sregion and on their cloud platform, facilitating seamless data sharing. References: Sharing data securely across regions and cloud platforms | Snowflake Documentation

 Snowflake recommends using the COPY command for bulk loading data batches from files already available in cloud storage. This command allows for efficient and large-scale data loading operations from files staged in cloud storage into Snowflake tables3.

SnowCD (Snowflake Connectivity Diagnostic Tool) is used to diagnose and troubleshoot network connections to Snowflake. It runs a series of connection checks to evaluate the network connection to Snowflake

 Snowflake’s secure data sharing feature allows sharing of certain database objects with other Snowflake accounts. Amongthe options provided, external tables and secure UDFs can be shared

The Parquet file format is known for preserving the precision of floating-point numberswhen data is unloaded, preventing truncation of the values3.

 Periodic rekeying of micro-partitions is a feature that requires the Enterprise Edition of Snowflake or higher. This feature is part of Snowflake’s comprehensive approach to encryption key management, ensuring data security through regular rekeying1. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The RETURN_ERRORS value in the VALIDATION_MODE option of theCOPY command instructs Snowflake to validate the data files and return errors encountered across all specified files, including those that were partially loaded during an earlier load2. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

 An account administrator in the Data Exchange can perform tasks such as adding and removing members and approving or denying listing approval requests. These tasks are part of managing the Data Exchange and ensuring that only authorized listings and members are part of it12.

A clustering key in Snowflake is used to co-locate similar data within the same micro-partitions to improve query performance, especially for large tables where data is not naturally ordered or has become fragmented due to extensive DML operations. The appropriate use of a clustering key can lead to improved scan efficiency and better column compression, resulting in faster query execution times.

The indicators that it may be appropriate to use a clustering key for a table include:

D. Queries on the table are running slower than expected: This can happen when the data in the table is not well-clustered, leading to inefficient scans during query execution.

E. The clustering depth for the table is large: A large clustering depth indicates that the table’s data is spread across many micro-partitions, which can degrade query performance as more data needs to be scanned.

 In a CREATE PIPE ... AS COPY ... FROM (....) statement, the supported transformations include filtering data using an optional WHERE clause and omitting columns. The WHERE clause allows for the specification of conditions to filter the data that is being loaded, ensuring only relevant data is inserted into the table. Omitting columns enables the exclusion of certain columns from the data load, which can be useful when the incoming data contains more columns than are needed for the target table.

 Query compilation in Snowflake occurs in the Cloud Services layer. This layer is responsible for coordinating and managing all aspects of the Snowflake service, including authentication, infrastructure management, metadata management, query parsing and optimization, and security. By handling these tasks, the Cloud Services layer enables the Compute layer to focus on executing queries, while the Storage layer is dedicated to persistently storing data.

 Snowpipe is used for continuous, automated data loading into Snowflake. It uses a COPY INTO < table >  statement within a pipe object to load data from files as soon as they are available in a stage. Snowpipe does not execute UDFs, stored procedures, or insert statements. References: Snowpipe | Snowflake Documentation

 To view warehouse usage over time, the Query history view and the WAREHOUSE_METERING__HISTORY View can be utilized. The Query history view allows users to monitor the performance of their queries and the load on their warehouses over a specified period1. The WAREHOUSE_METERING__HISTORY View provides detailed informationabout the workload on a warehouse within a specified date range, including average running and queued loads2. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

Snowflake does not require source data files to be no larger than 16MB. In fact, Snowflake recommends that for optimal load performance, data files should be roughly 100-250 MB in size when compressed. However, it is not recommended to load very large files (e.g., 100 GB or larger) due to potential delays and wasted credits if errors occur. Smaller files should be aggregated to minimize processing overhead, and larger files should be split to distribute the load among compute resources in an active warehouse.

 User-level network policies in Snowflake can be created by roles with the necessary privileges to manage security and account settings. The ACCOUNTADMIN role has the highest level of privileges across the account, including the ability to manage network policies. The SECURITYADMIN role is specifically responsible for managing security objects within Snowflake, which includes the creation and management of network policies.

 Clustering is a technique used in Snowflake to improve the performance of queries. It involves organizing the data in a table into micro-partitions based on the values of one or more columns. This organization allows Snowflake to efficiently prune non-relevant micro-partitions during a query, which reduces the amount of data scanned and improves query performance.

When file format options are specified in multiple locations, the load operation applies the options in the following order of precedence: first, the COPY INTO TABLE statement; second, the stage definition; and third, the table definition1

 The kind of Snowflake table that stores file-level metadata for each file in a stage is a directory table. A directory table is an implicit object layered ona stage and stores file-level metadata about the data files in the stage3.

Masking policies in Snowflake are designed to protect sensitive information by dynamically hiding or obfuscating data based on the role of the user executing the query. This helps enforce data privacy and security by allowing only authorized users to see sensitive information. Masking policies do not encrypt data but apply rules to limit data visibility, ensuring sensitive data is protected during query execution without altering the underlying data.

Snowpipe uses key-based authentication for its REST endpoints. This involves generating and using a key pair (public and private keys) to securely authenticate API requests.

Generate Key Pair:Generate a public and private key pair.

Register Public Key:Register the public key with the Snowflake user that will be making the API requests.

Authenticate Requests:Use the private key to sign API requests sent to Snowpipe REST endpoints.

In Snowflake, a table ' s clustering depth indicates the degree of micro-partition overlap based on the clustering keys defined for the table. A clustering depth of 1 implies that the table has no overlapping micro-partitions. This is an optimal scenario, indicating that the table ' s data is well-clustered according to the specified clustering keys. Well-clustered data can lead to more efficient query performance, as it reduces the amount of data scanned during query execution and improves the effectiveness of data pruning.

The correct answer is D. The total number of micro-partitions that the operation will reference .

In Snowflake, EXPLAIN or EXPLAIN USING returns the logical execution plan for a SQL statement. The plan can include information about the operations Snowflake expects to perform, such as table scans, filters, joins, and estimated partitions involved in the query.

Why D is correct:

The explain plan can show partition-related information for table scan operations, including the number of partitions that may be scanned. This is useful for understanding how well Snowflake can prune micro-partitions.

Why the other options are incorrect:

A. EXPLAIN_PLAN does not provide a guaranteed predicted query runtime. Actual execution time depends on warehouse size, load, caching, pruning, and other runtime factors.

B. EXPLAIN_PLAN does not directly state the number of credits the query will consume. Credit consumption is based mainly on warehouse runtime and size.

C. EXPLAIN_PLAN describes the execution plan, not the actual result data type returned by the query.

Official Snowflake documentation reference:

Snowflake documentation explains that the EXPLAIN command displays the logical execution plan for a SQL statement. Query plans can include details about operations such as scans, joins, filters, and partitions.

Snowflake ' s governance features vary across different editions, but theOBJECT_DEPENDENCIESview is supported by all Snowflake editions. This feature is part of Snowflake ' s Information Schema and is designed to help users understand the dependencies between various objects in their Snowflake environment.

TheOBJECT_DEPENDENCIESview provides a way to query and analyze the relationships and dependencies among different database objects, such as tables, views, and stored procedures. This is crucial for governance, as it allows administrators and data engineers to assess the impact of changes, understand object relationships, and ensure proper management of data assets.

Object tags, masking policies, and row access policies are more advanced features that offer fine-grained data governance capabilities such as tagging objects for classification, dynamically masking sensitive data based on user roles, and controlling row-level access to data. These features may have varying levels of support across different Snowflake editions, with some features being exclusive to higher-tier editions.

Snowflake utilizes clustering information to enhance query performance by pruning unnecessary micro-partitions.

Clustering Metadata: Snowflake stores clustering information for each micro-partition, which includes data range and distribution.

Pruning Micro-partitions: When a query is executed, Snowflake uses this clustering metadata to identify and eliminate micro-partitions that do not match the query criteria, thereby reducing the amount of data scanned and improving query performance.

Snowflake Documentation on Clustering

Snowflake Documentation on Micro-partition Pruning

When loading data into Snowflake, the file format options specified in theCOPY INTO < table > statement take the highest precedence over other locations such as the stage or table definitions. This ensures that any specific settings for a particular load operation are applied correctly.

File Format Hierarchy:

Stage Definition:Specifies default file format options for files staged in the location.

Table Definition:Can specify default file format options associated with the table.

COPY INTO Statement:Overrides both the stage and table definitions with the file format options specified directly in the statement.

Example Usage:

COPY INTO my_table

FROM @my_stage

FILE_FORMAT = (FORMAT_NAME = ' my_format ' FIELD_OPTIONALLY_ENCLOSED_BY = ' " ' );

The correct answers are A. JSON null values and E. Timestamps within strings .

Important correction:

The provided answer A, B is not the best answer. Snowflake documentation specifically calls out that certain values in semi-structured data may not be extracted into separate columnar storage, which can result in slower queries and increased storage usage. Two commonly tested examples are JSON null values and date or timestamp values stored as strings.

Why A is correct:

When a semi-structured element contains JSON null values, Snowflake might not extract that element into a separate column. This can reduce pruning efficiency and require Snowflake to scan more of the VARIANT structure during query execution.

Why E is correct:

Date and timestamp values inside semi-structured data that are stored as strings may not be optimized in the same way as native typed values. Storing timestamps as strings can increase storage and slow query performance because Snowflake may need to parse or cast the values at query time.

Why the other options are incorrect:

B. Floating-point numbers are supported inside VARIANT; they are not the main documented issue in this context.

C. Arrays are supported in semi-structured data. While very complex nested structures can affect query patterns, arrays alone are not the best answer here.

D. Vectors are a Snowflake data type used for vector operations and are not the standard answer for this semi-structured data optimization question.

Official Snowflake documentation reference:

Snowflake documentation explains that for semi-structured data, elements containing JSON null values may not be extracted into separate columns. It also recommends storing date and timestamp values in native typed columns when possible rather than as strings inside VARIANT, to improve query performance and storage efficiency.

The correct answer is B. ALTER STAGE .

Directory tables are associated with stages. To manually refresh the metadata in a directory table, the stage is refreshed using the ALTER STAGE ... REFRESH command.

Why B is correct:

A directory table stores file-level metadata for files in a stage. Since the directory table is tied to the stage, the manual refresh operation is performed through the stage.

Example:

ALTER STAGE my_stage REFRESH;

This refreshes the directory table metadata for the stage.

Why the other options are incorrect:

A. TRUNCATE STAGE is not the command used to refresh directory table metadata.

C. ALTER TABLE is used to modify table properties, not refresh a directory table associated with a stage.

D. ALTER VIEW is used to modify views, not directory tables.

Official Snowflake documentation reference:

Snowflake documentation explains that directory table metadata can be refreshed manually using ALTER STAGE ... REFRESH.

The settings shown in the image indicate that the data retention time in days is configured at two different levels: the account level and the object level. At the account level, theMIN_DATA_RETENTION_TIME_IN_DAYSis set to 5 days, and at the object level, theDATA_RETENTION_TIME_IN_DAYSis set to 2 days. Since the object level setting has a lower value, it takes precedence over the account level setting for the specific object. Therefore, the data will be retained for 2 days at the object level.

Pruning in Snowflake refers to the process of filtering or disregarding micro-partitions that are not needed to satisfy the conditions of a query. This optimization technique helps reduce the amount ofdata scanned, thereby improving query performance

To provide a user with the necessary access to create a materialized view in a schema, the user must be granted a role that has the CREATE MATERIALIZED VIEW privilege on that schema. First, the role is granted to the user, and then the privilege is granted to the role

With the create data EXCHANGE listing privilege, a Snowflake user can modify the properties of listings and submit them for approval or publishing on the Data Exchange. This allows them to manage and share data sets with consumers effectively. References: Based on general data exchange practices in cloud services as of 2021.

The minimum privilege required on an external stage for any role to access unstructured data files using a file URL in the GET REST API is READ. This allows the role to retrieve or download data files from the stage.

The Snowflake edition that enables data sharing only through Snowflake Support is the Virtual Private Snowflake (VPS). By default, VPS does not permit data sharing outside of the VPS environment, but it can be enabled through SnowflakeSupport4.

The bulk data load history available after a COPY statement is stored in the metadata of the pipe and is retained for 64 days1.

Snowflake recommends using columns with very high cardinality and those that are most actively used in selective filters as clustering keys. High cardinality columns have a wide range of unique values, which helps in evenly distributing the data across micro-partitions. Columns used in selective filters help in pruning the number of micro-partitions to scan, thus improving query performance. References: Based on general database optimization principles.

The correct answer is D. To return the first non-NULL expression in an argument .

COALESCE evaluates its arguments from left to right and returns the first expression that is not NULL. If all expressions are NULL, it returns NULL.

Why D is correct:

SELECT COALESCE(phone_mobile, phone_home, phone_work, ' No phone ' ) AS contact_phone

FROM customers;

This returns the first available non-NULL phone number. If all phone columns are NULL, it returns ' No phone ' .

Why the other options are incorrect:

A. Splitting strings is handled by functions such as SPLIT.

B. Removing duplicates is handled with DISTINCT, grouping, or deduplication logic.

C. Joining column values is usually done with concatenation functions or operators, such as CONCAT.

Official Snowflake documentation reference:

Snowflake documentation describes COALESCE as a conditional expression that returns the first non-NULL expression among its arguments.

 To showall privileges granted to a specific schema, the command SHOW GRANTS ON SCHEMA < schema_name >  should beused3. In this case, it would be SHOW GRANTS ON SCHEMA ANALYTICS_DW.MARKETING. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

In Snowflake, the ACCOUNTADMIN role is required to create resource monitors. Resource monitors are used to manage and monitor the consumption of compute resources. The ACCOUNTADMIN role has the necessary privileges to create, configure, and manage resource monitors across the account.

Snowflake Documentation: Resource Monitors

The correct answer is C. To ensure the same sample is returned each time a query is run .

The REPEATABLE keyword is used with a seed value to make sampling deterministic. This means Snowflake can return the same sample when the query is rerun with the same seed and the underlying data has not changed.

Why C is correct:

SELECT *

FROM table_A

SAMPLE BERNOULLI (10) REPEATABLE (123);

This returns a repeatable sample based on the seed value 123.

Why the other options are incorrect:

A. Returning a fixed number of rows uses SAMPLE ( < n > ROWS), not REPEATABLE.

B. REPEATABLE does not remove duplicates.

D. Block sampling is controlled with BLOCK or SYSTEM, not REPEATABLE.

Official Snowflake documentation reference:

Snowflake documentation explains that REPEATABLE or SEED specifies a seed value to make sample results repeatable.

The correct answer is B. Loading and unloading data .

SnowSQL is Snowflake’s command-line client. It is used to connect to Snowflake, execute SQL statements, run SQL scripts, and perform data loading and unloading operations.

Why B is correct:

SnowSQL supports data loading and unloading workflows. For example, users can use SnowSQL to run PUT, GET, and COPY commands.

Example loading workflow:

PUT file:///local/path/data.csv @my_stage;

COPY INTO my_table

FROM @my_stage

FILE_FORMAT = (TYPE = CSV);

Example unloading workflow:

COPY INTO @my_stage/output/

FROM my_table;

GET @my_stage/output/ file:///local/path/output/;

Why the other options are incorrect:

A. Dashboards are usually developed in BI tools or applications, not SnowSQL.

C. Snowflake Native Apps are developed and managed using Snowflake Native App Framework tools and Snowflake CLI, not primarily SnowSQL.

D. Monitoring transformation activity is usually done through Snowflake views, query history, task history, event tables, or monitoring features, not as the primary purpose of SnowSQL.

Official Snowflake documentation reference:

Snowflake documentation describes SnowSQL as the command-line client for connecting to Snowflake and executing SQL queries, including data loading and unloading commands.

 Snowflake data providers can share data residing in different databases through secure views. Secure views allow for the referencing of objects such as schemas, tables, and other views contained in one or more databases, as long as those databases belong to the same account. This enables providers to share data securely and efficiently with consumers. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

 Snowflake’s column level security features include Dynamic Data Masking and External Tokenization. Dynamic Data Masking uses masking policies to selectively mask data at query time, while External Tokenization allows for the tokenization of data before loading it into Snowflake and detokenizing it at query runtime5.

 Once a database has been created in a consumer account from a share, the shared objects become accessible to users in that account. The shared objects are not transferred or copied; they remain in the provider’s account and are accessible to the consumer account

To find out which numeric values are present in both tables, the INTERSECT set operator should be used. This operator returns rows from one query’s result set which also appear in another query’s result set, effectively finding the common elements between the two tables45.

The first user assigned to a new Snowflake account, typically with the ACCOUNTADMIN role, should create at least one additional user with the USERADMIN administrative privilege. This role is responsible for creating and managing users and roles within the Snowflake account. References: Access control considerations | Snowflake Documentation

Query parsing and compilation in Snowflake occur within the cloud services layer. This layer is responsible for various management tasks, including query compilation and optimization

 The reader account compute will be billed to the provider account. Very Comprehensive Explanation: In Snowflake, when a provider creates a reader account for a consumer who does not have a Snowflake account, the compute resources used by the reader account are billed to the provider’s account. This allows the consumer to query the shared data without incurring any costs. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The Query Profile in Snowflake can identify common query problems, including:

B. Inefficient pruning: This refers to the inability of a query to effectively limit the amount of data being scanned, potentially leading to suboptimal performance.

D. Queries too large to fit in memory: This indicates that a query requires more memory than is available in the virtual warehouse, which can lead to spilling to disk and degraded performance.

The Query Profile helps diagnose these issues by providing detailed execution statistics and visualizations, aiding in query optimization and troubleshooting.

The TABLE_STORAGE_METRICS view can be used to determine if a table has frequent row updates or deletes. This view provides detailed metrics on the storage utilization of tables within Snowflake, including metrics that reflect the impact of DML operations such as updates and deletes on table storage. For example, metrics related to the number of active and deleted rows can help identify tables that experience high levels of row modifications, indicating frequent updates or deletions.

The Cloud Services layer in Snowflake is responsible for a wide range of services that facilitate the management and use of Snowflake, including:

D. User authentication: This service handles identity and access management, ensuring that only authorized users can access Snowflake resources.

E. Infrastructure management: This service manages the allocation and scaling of resources to meet user demands, including the management of virtual warehouses, storage, and the orchestration of query execution.

These services are part of Snowflake ' s fully managed, cloud-based architecture, which abstracts and automates many of the complexities associated with data warehousing.

The Fail-safe period for a transient table in Snowflake, regardless of the edition (including Enterprise edition and higher), is 0 days. Fail-safe is a data protection feature that provides additional retention beyond the Time Travel period for recovering data in caseof accidental deletion or corruption. However, transient tables are designed for temporary or short-term use and do not benefit from the Fail-safe feature, meaning that once their Time Travel period expires, data cannot be recovered.

TheOBJECT_CONSTRUCTfunction in Snowflake is used to convert rows in a relational table into a single VARIANT column that represents each row as a JSON object. This function dynamically creates a JSON object from a list of key-value pairs, where each key is a column name and each value is the corresponding column value for a row. This is particularly useful for aggregating and transforming structured data into semi-structured JSON format for further processing or analysis.

When creating a CSV file format in Snowflake and the requirement is to wrap columns by single quotes or double quotes, theFIELD_OPTIONALLY_ENCLOSED_BYparameter should be used in the file format specification. This parameter allows you to define a character (either a single quote or a double quote) that can optionally enclose each field in the CSV file, providing flexibility in handling fields that contain special characters or delimiters as part of their data.

The graphical representation of warehouse utilization indicates periods of significant queuing, suggesting that the current single cluster cannot efficiently handle all incoming queries. Configuring the warehouse to a multi-cluster warehouse will distribute the load among multiple clusters, reducing queuing times and improving overall performance1.

References = Snowflake Documentation on Multi-cluster Warehouses1

To manage compliance with Personal Identifiable Information (PII) requirements, the Virtual Private Snowflake and Business Critical Editions of Snowflake should be used. These editions provide advanced security features necessary for handling sensitive data

When publishing a Snowflake Data Marketplace listing into a remote region, it’s important to note that while the listing is replicated into all selected regions automatically, the data itself is not.Therefore, the data must be replicated separately. Additionally, the user must have the ORGADMIN role in at least one account to manage the replication of accounts1.

The parse_json() function in Snowflake interprets an input string as a JSON document and produces a VARIANT value containing the JSON document. This function is specifically designed for parsing strings that contain valid JSON information3.

When loading data into Snowflake, it is recommended to organize the data into single files with 100-250 MB of compressed data per file. This size range is optimal for parallel processing and can help in achieving better performance during data loading operations. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

 In Snowflake’s access control model, the default owner of an object is the role that was used to create the object. This role has the OWNERSHIP privilege on the object and can grant access to other roles1

The Enterprise edition of Snowflake allows for a dedicated metadata store, providing additional features designed for large-scale enterprises

 Snowflake’s architecture is described as a multi-cluster, shared data architecture. This design combines the simplicity of a shared-disk architecture with the performance and scale-out benefits of a shared-nothing architecture, using a central repository accessiblefrom all compute nodes2.

References = [ COF-C03 ] SnowPro Core Certification Exam Study Guide, Snowflake Documentation

A well-clustered table in Snowflake means that the data is organized in such a way that related data points are stored close to each other within the micro-partitions. This optimizes query performance by reducing the amount of scanned data. The diagram indicated by option C shows a well-clustered table, as it likely represents a more evenly distributed range of values across the micro-partitions1.

References = Snowflake Micro-partitions & TableClustering

 In Snowflake, both materialized views and transient tables will incur storage charges because they store data. They will also incur compute charges when queries are run against them, as compute resources are used to process the queries. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

Worksheets in Snowsight can be shared directly with other Snowflake users within the same account. This feature allows for collaboration and sharing of SQL queries or Python code, as well as other data manipulation tasks1.

 Row level security in Snowflake is available starting with the Enterprise edition. This feature allows for the creation of row access policies that can control access to data at the row level within tables and views

The default file size when unloading data from Snowflake using the COPY command is not explicitly stated in the provided resources. However, Snowflake documentation suggests that the file size can be specified using the MAX_FILE_SIZE option in the COPY INTO < location >  command2.

 Snowflake’s Fail-safe provides a 7-day period during which data in a permanent table may be recoverable, but only by Snowflake Support, not by user operations3.

 When a pipe is recreated using the CREATE OR REPLACE PIPE command, the load history of the pipe is reset. This means that Snowpipe will consider all files in the stage as new and will attempt to load them, evenif they were loaded previously by the old pipe2.

 In Snowflake, each size increase in virtual warehouses doubles the number of servers. Therefore, if a sizeSmall virtual warehouse is made up of two servers, a Large warehouse, which is two sizes larger, would be made up of eight servers (2 servers for Small, 4 for Medium, and 8 for Large)2.

Size specifies the amount of compute resources available per cluster in a warehouse. Snowflake supports the following warehouse sizes:

https://docs.snowflake.com/en/user-guide/warehouses-overview.html

An explain plan is beneficial because it allows for the evaluation of how a query will be processed without the need to actually execute the query. This can help in understanding the query’s performance implications and potential bottlenecks without consuming resources that would be used if the query were run

Operations that do not require compute resources are typically those that can leverage previously cached results. However, if no queries have been executed previously, all the given operations would require compute to execute. It’s important to note that certain operations like DDL statements and queries that hit the result cache do not consume compute credits2.

 The supported file formats for unloading data from Snowflake include JSON, Parquet, and CSV. These formats are commonly used for their flexibility and compatibility with various data processing tools

 Users are responsible for data storage costs in Snowflake until the data expires from the Fail-safe period. Fail-safe is the final stage in the data lifecycle, following Time Travel, and provides additional protection against accidental data loss. Once data exits the Fail-safe state, users are no longer billed for its storage

The query results cache can be used as long as the data in the table has not changed since the last time the query was run. If the underlying data has changed, Snowflake will not use the cached results and will re-execute the query1.

The features associated with Continuous Data Protection (CDP) that require additional Snowflake-provided data storage are Time Travel and Fail-safe. Time Travel allows users to access historical data within a defined period, while Fail-safe provides an additional layer of data protection beyond the Time Travel period. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The list @my_stage command in Snowflake can be used to verify if data has been uploaded to an external stage named my_stage. This command provides a list of files thatare present in the specified stage2.

 The COPY INTO SQL command with the option SINGLE=FALSE is used to unload data into multiple files. This option allows the data to be split into multiple files during the unload process. References: SnowPro Core Certification COPY INTO SQLcommand unload multiple files

 Organizing files into logical paths can significantly improve the efficiency of data loading from an external stage. This practice helps in managing and locating files easily, which can be particularly beneficial when dealing with large datasets or complex directory structures1.

Best practices for using the ACCOUNTADMIN role include ensuring that all users with this role use Multi-factor Authentication (MFA) for added security. Additionally, it is recommendedto assign the ACCOUNTADMIN role to at least two users to avoid delays in case of password recovery issues, but to as few users as possible to maintain strict control over account-level operations4.

The ResultSet cache is leveraged to quickly return results for repeated queries. Actions that prevent leveraging this cache include stopping the virtual warehouse that the query is running against (B) and executing the RESULTS_SCAN() table function (D). Stopping the warehouse clears the local diskcache, including the ResultSet cache1. The RESULTS_SCAN() function is used to retrieve the result of a previously executed query, which bypasses the need for the ResultSet cache.

 Snowpipe retains load history for 14 days. This allows users to view and audit the data that has been loaded into Snowflake using Snowpipe within this time frame3.

 In Snowflake, a task allows a user to execute a single SQL statement/command using a predefined schedule (B). Tasks are used to automate the execution of SQL statements at scheduled intervals.

The LOGIN_HISTORY view in the ACCOUNT_USAGE schema provides information about login attempts, including both successful and failed logins. This view can be used to review the failed login attempts of a specific user for the past 30 days. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

When cloning a database containing stored procedures and regular views with fully qualified table references, the cloned views and stored procedures will reference the cloned tables in the cloned database (A). This ensures that the cloned database is a self-contained copy of the original, with all references pointing to objects within the same cloned database. References: SnowPro Core Certification cloningdatabase stored procedures views

The feature in Snowflake that allows a user to substitute a randomly generated identifier for sensitive data before loading it into Snowflake is known as External Tokenization. This process helps to secure sensitive data by ensuring that it is not exposed in its original form, thus preventing unauthorized access3.

A row access policy can be applied to a table or a view within the policy DDL when defining the policy. Additionally, an existing row access policy can be applied to a table or a view using the ALTER < object > ADD ROW ACCESS POLICY < policy >  command

The SQL command Select * from table(validate(t1, job_id = > ' _last ' )); is used to return errors from the last executed COPY command into table t1 in the current session. It checksthe results of the most recent data load operation and provides details on any errors that occurred during that process1.

Data masking policies in Snowflake allow for the obfuscation of specific data within a field, enabling some users to see the full data while limiting others. This feature is particularly useful for handling PII, ensuring that sensitive information is only visible to authorized users1.

 In an auto-scaling multi-cluster virtual warehouse with the SCALING_POLICY set to ECONOMY, another cluster is started when the system has enough load for 2 minutes (A). This policy is designed to optimize the balance between performance and cost, starting additional clusters only when the sustained load justifies it2.

The maximum Time Travel retention period for a temporary Snowflake table is 1 day. This is the standard retention period for temporary tables, which allows for accessing historical data within a 24-hour window

 In Snowflake, a schema is a logical grouping of database objects, which can include streams and external tables. A stream is an object that allows users to query data that has changed in specified tables or views, and an external table is a table that references data stored outside of Snowflake. Roles, warehouses, users, and shares are not contained within a schema. References: SHOW OBJECTS, Database, Schema, & Share DDL

Shares in Snowflake are named objects that encapsulate all the information required to share databases, schemas, tables, secure views, and secure UDFs. These objects can be added to a share by granting privileges on them to the share via a database role

The Snowflake Search Optimization Service is designed to support improved performance for selective point lookup queries. These are queries that retrieve specific records from a database, often based on a unique identifier or a small set of criteria3.

 For a temporary table, the maximum total Continuous Data Protection (CDP) charges incurred are forthe duration of the session in which the table was created, which does not exceed 24 hours2.

References = [ COF-C03 ] SnowPro Core Certification Exam Study Guide, Snowflake Documentation2

Snowflake’s architecture is designed to operate entirely on cloud infrastructure. It uses the core compute and storage services of each cloud provider, which allows it to leverage the scalability and reliability of cloud resources. Snowflake’s services are deployed across multiple availability zones within a cloud provider’s region to ensure high availability and fault tolerance. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The responsibilities of Snowflake’s Cloud Service layer include authentication (A), which ensures secure access to the platform; resource management (B), which involves allocating and managing compute resources; and query parsing and optimization (D), which improves the efficiency and performance of SQL query execution3.

Network policies in Snowflake can be set at the user level and at the account level2.

 Protegrity is listed as a data tokenization integration partner for Snowflake. This partnership allows Snowflake users to utilize Protegrity’s tokenization solutions within the Snowflake environment3.

References = [ COF-C03 ] SnowPro Core Certification Exam Study Guide, Snowflake Documentation

 When the retention period for an object ends in Snowflake, Time Travel on the historical data is dropped ©. This means that the ability to access historical data via Time Travel is no longer available once the retention period has expired2.

Snowflake recommends file sizes between 100-250 MB compressed when loading data to optimize parallel processing. Smaller, compressed files can be loaded in parallel, which maximizes the efficiency of the virtual warehouses and speeds up the data loading process

 Snowflake stored procedures can be created to execute with the privileges of the role that owns the procedure (owner’s rights) or with the privileges of the role that calls the procedure (caller’s rights). This allows for flexibility in managing security and access control within Snowflake1.

Snowflake supports the use of external stages with Amazon Web Services (AWS), Microsoft Azure Cloud, and Google Cloud Platform (GCP). These platforms allow users to stage data externally and integrate with Snowflake for data loading operations

 Transient tables in Snowflake have a minimum Fail-safe retention time period of 0 days. This means that once the Time Travel retention period ends, there is no additional Fail-safe period for transient tables

To delete staged files from a Snowflake stage, you can specify the PURGE option in the COPY INTO < table >  command, which will automatically delete the files after they have been successfully loaded. Additionally, you can use the REMOVE command after the load completes to manually delete the files from the stage12.

References = DROP STAGE, REMOVE

In Snowflake, to extract a specific element from a JSON stored in a VARIANT column, the correct syntax is to use the dot notation. Therefore, the query select src.dealership from car_sales; will return the dealership information contained within each JSON object in the src column.

 Snowflake uses a Role-Based Access Control (RBAC) model to manage access to data and resources. Additionally, Snowflake ensures that all data is encrypted, both at rest and in transit, to provide a high level of security for data stored within the platform. References: [ COF-C03 ] SnowPro Core Certification Exam Study Guide

The ACCESS_HISTORY view in the SNOWFLAKE.ACCOUNT_USAGE schema contains information about the access history of Snowflake objects, such as tables and views, within thelast 365 days1.

 The GET command is used to download files from a Snowflake stage to a local folder on a client’s machine2.

When data is loaded into Snowflake, it undergoes a reorganization process where the data is stored in a columnar format and compressed. The columnar storage format enables efficient querying and data retrieval, as it allows for reading only the necessary columns for a query, thereby reducing IO operations. Additionally, Snowflake uses advanced compression techniques to minimize storage costs and improve performance. This combination of columnar storage and compression is key to Snowflake ' s data warehousing capabilities.

The layer of Snowflake ' s architecture associated with virtual warehouses is the Query Processing layer. Virtual warehouses in Snowflake are dedicated compute clusters that execute SQL queries against the stored data. This layer is responsible for the entire query execution process, including parsing, optimization, and the actual computation. It operates independently of the storage layer, enabling Snowflake to scale compute and storage resources separately for efficiency and cost-effectiveness.

The VALIDATE table function in Snowflake would typically use a unique identifier, such as a UUID_STRING, as an input argument. This function is designed to validate the data within a table against a set of constraints or conditions, often requiring a specific identifier to reference the particular data or job being validated.

Characteristics of reader accounts in Snowflake include:

A. Reader account users cannot add new data to the account: Reader accounts are intended for data consumption only. Users of these accounts can query and analyze the data shared with them but cannot upload or add new data to the account.

E. Reader accounts enable data consumers to access and query data shared by the provider: One of the primary purposes of reader accounts is to allow data consumers to access and perform queries on the data shared by another Snowflake account, facilitating secure and controlled data sharing.

The Virtual Private Snowflake (VPS) edition offers the highest level of security for organizations with the strictest security requirements. This edition provides a dedicated andisolated instance of Snowflake, including enhanced security features and compliance certifications to meet the needs of highly regulated industries or any organization requiring the utmost in data protection and privacy.

In Snowflake, stored procedures and User-Defined Functions (UDFs) have different invocation patterns within SQL:

By default, the ACCOUNTADMIN role in Snowflake has the ability to create a share from a shared database. This role has the highest level of access within a Snowflake account,including the management of all aspects of the account, such as users, roles, warehouses, and databases, as well as the creation and management of shares for secure data sharing with other Snowflake accounts.

By default, the standard retention period for Time Travel in Snowflake is 1 day across all Snowflake accounts. Time Travel enables users to access historical data within this retention window, allowing for point-in-time data analysis and recovery. This feature is a significant aspect of Snowflake ' s data management capabilities, offering flexibility in handling data changes and accidental deletions.

Network policies in Snowflake are defined as a set of rules that manage the network-level access to Snowflake accounts. These rules specify which IP addresses or IP ranges are permitted to connect to Snowflake, enhancing the security of Snowflake accounts by preventing unauthorized access. Network policies are an essential aspect of Snowflake ' s security model, allowing administrators to enforce access controls based on network locations.

When a Snowflake user increases the size of a suspended virtual warehouse, the changes to compute resources are queued but do not take immediate effect. The provisioning of additional compute resources occurs only when the warehouse is resumed. This ensures that resources are allocated efficiently, aligning with Snowflake ' s commitment to cost-effective and on-demand scalability.

The worksheet and database explorer feature in Snowsight allows users to tag frequently accessed worksheets for ease of access. This functionality helps users organize and quickly navigate to the worksheets they use most often, enhancing productivity and streamlining the data exploration and analysis process within Snowsight, Snowflake ' s web-based query and visualization interface.

In Snowflake, semi-structured data is optimally stored using specific data types that are designed to handle the flexibility and complexity of such data. TheVARIANTdata type can store structured and semi-structured data types, including JSON, Avro, ORC, Parquet, or XML, in a single column. TheARRAYdata type, on the other hand, is suitable for storing ordered sequences of elements, which can be particularly useful for semi-structured data types like JSON arrays. These data types provide the necessary flexibility to store and query semi-structured data efficiently in Snowflake.

A materialized view in Snowflake is a database object derived from a query specification, stored for later use, and can significantly speed up expensive aggregations on large data sets. Materialized views store the result of their underlying query, reducing the need to recompute the result each time the view is accessed. This makes them ideal for improving the performance of read-heavy, aggregate-intensive queries.

The VARIANT data type in Snowflake is used to store JSON key-value pairs along with other semi-structured data formats like AVRO, BSON, and XML. The VARIANT data type allows for flexible and dynamic data structures within a single column, accommodating complex and nested data. This data type is crucial for handling semi-structured data in Snowflake, enabling users to perform SQL operations on JSON objects and arrays directly.

In Snowflake, the maximum number of clusters that can be provisioned within a multi-cluster virtual warehouse is 10. This allows for significant scalability and performance management by enabling Snowflake to handle varying levels of query load by adjusting the number of active clusters within the warehouse.

TheSYSTEM$GET_PRIVATELINKfunction in Snowflake provides proxy information necessary for configuring PrivateLink connections, which can protect Snowsight as well as other Snowflake services. PrivateLink enhances security by allowing Snowflake to be accessed via a private connection within a cloud provider’s network, reducing exposure to the public internet.

Snowflake recommends enhancing the granularity and management of privileges by creating and utilizing custom roles. When additional privileges are needed beyond those provided by the system-defined roles for a virtual warehouse or any other resource, these privileges should be granted to a custom role. This approach allows for more precise control over access rights and the ability to tailor permissions to the specific needs of different user groups or applications within the organization, while also maintaining the integrity and security model of system-defined roles.

TheREVOKE ROLEcommand is used to remove a role from another role or a user in Snowflake. This command is part of Snowflake ' s role-based access control system, allowing administrators to manage permissions and access to database objects efficiently by adding or removing roles from users or other roles.

To temporarily bypass a network policy by configuring the user object property MINS_TO_BYPASS_NETWORK_POLICY, the USERADMIN role should be used. This role has the necessary privileges to modify user properties, including setting a temporarybypass for network policies, which can be crucial for enabling access under specific circumstances without permanently altering the network security configuration.

Temporary tables in Snowflake do not consume storage costs. They are designed for transient data that is needed only for the duration of a session. Data stored in temporary tables is held in the virtual warehouse ' s cache and does not persist beyond the session ' s lifetime, thereby not incurring any storage charges.

When querying semi-structured data in Snowflake, the behavior towards case sensitivity is distinct between column names and the names of elements within the semi-structured data. Column names follow the general SQL norm of being case-insensitive, meaning you can reference them in any case without affecting the query. However, element names within JSON, XML, or other semi-structured data are case-sensitive. This distinction is crucial for accurate data retrieval and manipulation in Snowflake, especially when working with JSON objects where the case of keys can significantly alter the outcome of queries.

Object Name Resolution: When unqualified object names (e.g., table name without schema) are used in a UDF, Snowflake follows a specific hierarchy to resolve them. Here ' s the order:

Current Schema: Snowflake first checks if an object with the given name exists in the schema currently in use for the session.

PUBLIC Schema: If the object isn ' t found in the current schema, Snowflake looks in the PUBLIC schema of the current database.

Note: TheSEARCH_PATHparameter influences object resolution for queries, not within UDFs.

The search optimization service in Snowflake supports use cases involving conjunctions (AND) of multiple equality predicates. This service enhances the performance of queries that include multiple equality conditions by utilizing search indexes to quickly filter data without scanning entire tables or partitions. It ' s particularly beneficial for improving the response times of complex queries that rely on specific data matching across multiple conditions.

When unloading data, the Parquet file format is known for its efficiency in preserving the data values for floating-point number columns. Parquet is a columnar storage file format that offers high compression ratios and efficient data encoding schemes. It is especially effective for floating-point data, as it maintains high precision and supports efficient querying and analysis.

In Snowflake, when setting up a network policy that specifies both allowed and blocked IP address lists, if an IP address appears in both lists, access from that IP address will be denied. The reason is that Snowflake prioritizes security, and the presence of an IP address in the blocked list indicates it should not be allowed regardless of its presence in the allowed list. This ensures that access controls remain stringent and that any potentially unsafe IP addresses are not inadvertently permitted access.

The ORGADMIN role in Snowflake is an organizational-level role that provides administrative capabilities across the entire organization, rather than being limited to a single Snowflake account. Users with this role can:

A. Create an account for an organization: The ORGADMIN role has the privilege to create new Snowflake accounts within the organization, allowing for the expansion and management of the organization ' s resources.

D. View usage information for all accounts in an organization: This role also has access to comprehensive usage and activity data across all accounts within the organization. This is crucial for monitoring, cost management, and optimization at the organizational level.

The Search Optimization Service in Snowflake is designed to improve the performance of specific types of queries, particularly those involving extensive sets of WHERE conditions. By maintaining a search index on tables, this service can accelerate lookup and analytical queries, making it a valuable feature for optimizing query performance and reducing execution times for complex searches.

A clustering key can be defined when a table is created by appending a CLUSTER Where each clustering key consists of one or more table columns/expressions, which can be of any data type, except GEOGRAPHY, VARIANT, OBJECT, or ARRAYhttps://docs.snowflake.com/en/user-guide/tables-clustering-keys

The default value for auto-suspending a Virtual Warehouse in the Snowflake Web Interface (UI) is 10 minutes. This setting helps manage compute costs by automatically suspending warehouses that are not in use, ensuring that compute resources are efficiently allocated and not wasted on idle warehouses.

Tri-secret secure encryption is a security model employed by Snowflake that involves combining a customer-managed key with a Snowflake-managed key to create a composite key for encrypting data. This model enhances data security by requiring both the customer-managed key and the Snowflake-managed key to decrypt data, thus ensuring that neither party can access the data independently. It represents a balanced approach to key management, leveraging both customer control and Snowflake ' s managed services for robust data encryption.

A materialized view in Snowflake denotes a pre-computed data set derived from a SELECT query specification and stored for later use. Unlike standard views, which dynamically compute the data each time the view is accessed, materialized views store the result of the query at the time it is executed, thereby speeding up access to the data, especially for expensive aggregations on large datasets.

UTF-8 is the only supported character set for loading and unloading data from all supported file formats in Snowflake. UTF-8 is a widely used encoding that supports a large range of characters from various languages, making it suitable for internationalization and ensuring data compatibility across different systems and platforms.

In Snowflake, overloading refers to the creation of multiple User-Defined Functions (UDFs) with the same name but differing in the number or types of their arguments. This feature allows for more flexible function usage, as Snowflake can differentiate between functions based on the context of their invocation, such as the types or the number of arguments passed. Overloading helps to create more adaptable and readable code, as the same function name can be used for similar operations on different types of data.

To easily transfer data from a production account to a non-production account in Snowflake within the same cloud provider region, creating a data share is the most efficient approach. Data sharing allows for live, read-only access to selected data objects from the production account to the non-production account without the need to duplicate or move the actual data. This method facilitates seamless access to the data for development, testing, or analytics purposes in the non-production environment.

When dealing with semi-structured data in Snowflake, and the specific types of operations to be performed on the data are not yet determined, Snowflake recommends using the VARIANT data type. The VARIANT type is highly flexible and capable of storing data in multiple formats, including JSON, AVRO, BSON, and more, within a single column. This flexibility allows users to perform various operations on the data, including querying and manipulation of nested data structures without predefined schemas.

Snowflake bills virtual warehouse usage in one-minute increments, rounding up to the nearest minute for any partial minute of compute time used. If a virtual warehouse runs for 61 seconds and then, after being shut down, restarts and runs for an additional 30 seconds, the total time billed would be 120 seconds or 2 minutes. The first 61 seconds are rounded up to 2 minutes, and the subsequent 30 seconds are within a new minute, which is also rounded up to the nearest minute.

Scalar functions in Snowflake (and SQL in general) are designed to return a single value for each invocation. They operate on a single value and return a single result, making them suitable for a wide range of data transformations and calculations within queries.

To utilize the search optimization service in Snowflake, the correct syntax for granting privileges to a role involves specific commands that include adding search optimization capabilities:

Option C: GRANT ADD SEARCH OPTIMIZATION ON SCHEMA < schema_name > TO ROLE < role > . This command grants the specified role the ability to implement search optimization at the schema level, which is essential for enhancing search capabilities within that schema.

Options A and B do not include the correct verb " ADD, " which is necessary for this specific type of grant command in Snowflake. Option D incorrectly mentions the database level, as search optimization privileges are typically configured at the schema level, not the database level.

Purpose:TheBUILD_STAGE_FILE_URLfunction generates a temporary, pre-signed URL that allows you to access a file within a Snowflake stage (internal or external). This URL requires authentication to use.

Key Points:

Security: The URL has a limited lifespan, enhancing security.

Use Cases: Sharing staged data with external tools or applications, or downloading it directly.

Snowflake Documentation (BUILD_STAGE_FILE_URL): https://docs.snowflake.com/en/sql-reference/functions/build_stage_file_url.html

When unloading floating-point number columns to CSV or JSON files, Snowflake truncates the values to approximately 15 significant digits with 9 digits following the decimal point, which can be represented as (15,9). This ensures a balance between accuracy and efficiency in representing floating-point numbers in text-based formats, which is essential for data interchange and processing applications that consume these files.

The Query Profile in Snowflake provides a graphical representation and statistics for each component of the query ' s execution plan. This includes details such as the execution time, the number of rows processed, and the amount of data scanned for each operation within the query. The Query Profile is a crucial tool for understanding and optimizing the performance of queries, as it helps identify potential bottlenecks and inefficiencies.

Query pruning in Snowflake refers to the optimization technique where the system reduces the amount of data scanned by a query based on the query conditions. This typically involves skipping unnecessary data partitions that do not contribute to the query result. The effectiveness of this technique can be observed through:

Option A: Partitions scanned. This statistic indicates how many data partitions were actually scanned as a result of query pruning, showing the optimization in action.

Option C: Bytes scanned. This measures the volume of data physically read during query execution, and a reduction in this number indicates effective query pruning, as fewer bytes are read when unnecessary partitions are skipped.

Options B, D, and E do not directly relate to observing the effects of query pruning. " Partitions total " shows the total available, not the impact of pruning, while " Bytes read from result " and " Bytes written " relate to output rather than the efficiency of data scanning.

To insert JSON formatted string data into a VARIANT field in Snowflake, the correct function to use isPARSE_JSON. ThePARSE_JSONfunction is specifically designed to interpret a JSON formatted string and convert it into a VARIANT type, which is Snowflake ' s flexible format for handling semi-structured data like JSON, XML, and Avro. This function is essential for loading and querying JSON data within Snowflake, allowing users to store and manage JSON data efficiently while preserving its structure for querying purposes. This function ' s usage and capabilities are detailed in the Snowflake documentation, providing users with guidance on how to handle semi-structured data effectively within their Snowflake environments.

In Snowflake, the query history and the results of queries executed by a user are accessible based on the roles and permissions. If User1 executed a query with the SYSADMIN role, User2 would be able to view the result set of that query executed by User1 only if User2 has the ACCOUNTADMIN role. The ACCOUNTADMIN role has the broadest set of privileges, including the ability to access all aspects of the account ' s operation, data, and query history, thus enabling User2 to view the results of queries executed by other users.

Query pruning in Snowflake is the mechanism used to limit the number of micro-partitions scanned by a query. By analyzing the filters and conditions applied in a query, Snowflake can skip over micro-partitions that do not contain relevant data, thereby reducing the amount of data processed and improving query performance. This technique is particularly effective for large datasets and is a key component of Snowflake ' s performance optimization features.

The Fail-safe retention period for transient and temporary tables in Snowflake is 0 days. Fail-safe is a feature designed to protect data against accidental loss or deletion by retaining historical data for a period after its Time Travel retention period expires. However, transient and temporary tables, which are designed for temporary or short-term storage and operations, do not have a Fail-safe period. Once the data is deleted or the table is dropped, it cannot be recovered.

To access a file URL from an external stage in Snowflake, the minimum permission required is USAGE on the stage object. USAGE permission allows a user to reference the stage in SQL commands, necessary for actions like listing files or loading data from the stage, but does not permit the user to alter or drop the stage.

Transient tables in Snowflake are designed for temporary or intermediate workloads with the following characteristics:

B. Transient tables can be cloned to permanent tables: This feature allows users to create copies of transient tables for permanent use, providing flexibility in managing data lifecycles.

C. Transient tables persist until they are explicitly dropped: Unlike temporary tables that exist for the duration of a session, transient tables remain in the database until explicitly removed by a user, offering more durability for short-term data storage needs.

By default, the ACCOUNTADMIN role in Snowflake has access to the SYSTEM$GLOBAL_ACCOUNT_SET_PARAMETER function. This function is used to set global account parameters, impacting the entire Snowflake account ' s configuration and behavior. The ACCOUNTADMIN role is the highest-level administrative role in Snowflake, granting the necessary privileges to manage account settings and security features, including the use of global account parameters.

Snowflake supports semi-structured data types that can store complex data structures within a single column. TheVARIANTdata type can hold structured and semi-structured data formats such as JSON, Avro, ORC, Parquet, and XML. TheARRAYdata type is also used for semi-structured data and can store an ordered list of elements. These data types enable users to work with semi-structured data directly in Snowflake without needing to flatten the data into a relational schema first.

The correct answer is D. The USAGE privilege on the database and the schema .

To query a table in Snowflake, a role needs the SELECT privilege on the table. In addition, the role must have USAGE on the containing database and schema so it can resolve and access the object path.

Why D is correct:

For a table named:

database1.schema1.table1

A role generally needs:

USAGE ON DATABASE database1

USAGE ON SCHEMA database1.schema1

SELECT ON TABLE database1.schema1.table1

Without USAGE on the database and schema, the role cannot access the namespace that contains the table, even if SELECT exists on the table.

Why the other options are incorrect:

A. There is no USAGE privilege on a table for this purpose. SELECT is the table-level privilege needed to read data.

B. OWNERSHIP on the schema is not required to query a table.

C. REFERENCES is used for foreign key/reference-related behavior and is not required for simple SELECT queries.

Official Snowflake documentation reference:

Snowflake documentation explains that access to objects requires privileges on the object and appropriate USAGE privileges on parent containers, such as the database and schema.

Snowflake ' s Data Exchange allows users to create and manage a group of accounts to which they can offer data shares. This platform facilitates secure and governed data sharing within and between organizations.

Data Exchange Setup:Users can set up a Data Exchange and invite other Snowflake accounts to join.

Sharing Data:Within the Data Exchange, users can offer shares to the entire group or to specific accounts.

Governance and Security:Data Exchange provides tools for data governance and security, ensuring that shared data is only accessible to authorized accounts.

TheTO_JSONfunction in Snowflake is used to convert a relational table or individual rows into JSON format. This function is helpful for exporting data in JSON format.

Using TO_JSON Function:

SELECT TO_JSON(OBJECT_CONSTRUCT(*))

FROM my_table;

Exporting Data:TheTO_JSONfunction converts the table rows into JSON format, which can then be exported to a file.

In Snowflake, permissions for accessing database objects, including tables, are not granted directly to users but rather to roles. A role encapsulates a collection of privileges on various Snowflake objects. Users are then granted roles, and through those roles, they inherit the permissions necessary to read a table or perform other actions. This approach adheres to the principle of least privilege, allowing for granular control over database access and simplifying the management of user permissions.

The correct answer is A. Clone the table .

The most cost-efficient way to back up a Snowflake table is to use zero-copy cloning . A clone creates a copy of the table’s metadata without immediately duplicating the underlying micro-partitions.

Why A is correct:

When a table is cloned, Snowflake creates a new independent table object that initially shares the same micro-partitions as the source table. Because the underlying data is not copied at clone creation time, cloning is fast and storage-efficient.

This is why cloning is commonly used for backups, development, testing, and point-in-time snapshots.

Important concept:

At the instant a clone is created, the original table and cloned table share the same micro-partitions. The clone becomes independent after creation, and changes to either object do not modify the other object. Additional storage is used only when data changes occur.

Why the other options are incorrect:

B. Downloading table data to a stage creates additional data files and increases storage and operational overhead.

C. Sharing data with another account in another region is not the simplest or lowest-cost table backup method. Cross-region replication or sharing can involve additional costs.

D. A CTAS statement physically creates a new table from query results and can duplicate data storage, making it less storage-efficient than cloning.

Official Snowflake documentation reference:

Snowflake documentation explains that zero-copy cloning creates a copy of a database, schema, or table without duplicating the underlying storage at the time the clone is created.

The correct answer is D. DataFrame .

In Snowpark, a DataFrame represents a lazily evaluated relational data set. It provides methods to operate on data, such as selecting columns, filtering rows, joining data sets, grouping data, and applying transformations.

Why D is correct:

A Snowpark DataFrame is the primary abstraction for working with data in Snowpark. It represents data in Snowflake and lets developers build transformations using supported programming languages such as Python, Java, or Scala.

Example concept:

df = session.table( " CUSTOMERS " )

filtered_df = df.filter(df[ " COUNTRY " ] == " USA " )

The DataFrame represents the data set, and operations on it are translated into SQL that runs in Snowflake.

Why the other options are incorrect:

A. A list is a general programming data structure, not the primary Snowpark data set abstraction.

B. A table stores data in Snowflake, but in Snowpark, the object used to represent and operate on a data set programmatically is a DataFrame.

C. An array is a data structure or semi-structured type, not the main Snowpark abstraction for data operations.

Official Snowflake documentation reference:

Snowflake documentation describes Snowpark DataFrames as objects that represent a lazily evaluated relational data set and provide methods for transforming and analyzing data.

The correct answer is C. VARCHAR(25) .

In Snowflake, CHAR, CHARACTER, STRING, and TEXT are synonyms for VARCHAR. Snowflake stores and displays these character string data types as VARCHAR.

Corrected typo:

Option D was written as CHARACTERS). The corrected option is CHARACTER(25) .

Why C is correct:

A column created as CHARACTER(25) is treated as VARCHAR(25) in Snowflake metadata and display output.

Why the other options are incorrect:

A. STRING is a synonym for VARCHAR, but Snowflake commonly displays the canonical type as VARCHAR.

B. TEXT is also a synonym for VARCHAR, but it is not the canonical display type.

D. CHARACTER is accepted when defining the column, but Snowflake displays it as VARCHAR.

Official Snowflake documentation reference:

Snowflake documentation explains that string data type synonyms such as CHAR, CHARACTER, STRING, and TEXT are equivalent to VARCHAR.

When a user is granted a custom role , they inherit only the privileges assigned to that role . This is aligned with the principle of least privilege in RBAC.

Network policies in Snowflake are used to control access to Snowflake accounts based on IP address ranges. These policies can be activated by specific roles that have the necessary privileges.

Role: ACCOUNTADMIN:

The ACCOUNTADMIN role has full administrative rights across the Snowflake account.

This role can manage all aspects of the Snowflake environment, including network policies.

Role with Global ATTACH POLICY Privilege:

Any role that has been granted the global ATTACH POLICY privilege can activate network policies.

This privilege allows the role to attach policies that control network access to the account.

Snowflake Documentation: Network Policies

The correct answer is A. To return a tabular value for each input row.

A User-Defined Table Function, or UDTF, returns a set of rows instead of a single scalar value. This is different from a scalar UDF, which returns one value for each input row.

Why A is correct:

A UDTF produces tabular output. It can return multiple rows and multiple columns, depending on how the function is defined.

Example concept:

SELECT *

FROM TABLE(my_udtf(input_column));

This syntax shows that the function result is treated like a table.

Why the other options are incorrect:

B. Returning a pandas Series is associated with vectorized Python UDF behavior, not the general definition of a UDTF.

C. Returning one output row as a single column or value describes a scalar UDF, not a table function.

D. Processing input rows as pandas DataFrames is associated with Snowpark Python and vectorized processing patterns, but it is not the basic Snowflake definition of a UDTF.

Official Snowflake documentation reference:

Snowflake documentation defines a UDTF as a user-defined function that returns tabular results. It can return multiple rows for each input row.

The correct answer is D. ON_ERROR = ABORT_STATEMENT .

ON_ERROR = ABORT_STATEMENT stops the load operation if an error is encountered. This helps avoid loading only part of a dataset when a file or row error occurs.

Why D is correct:

If the goal is to avoid a partial dataset, the load should fail rather than continue after errors. ABORT_STATEMENT prevents Snowflake from loading only the valid rows while skipping invalid ones.

Why the other options are incorrect:

A. FORCE = FALSE prevents reloading files that were already loaded, but it does not prevent partial loads caused by data errors.

B. RETURN_FAILED_ONLY affects command output reporting. It does not control whether partial data is loaded.

C. LOAD_UNCERTAIN_FILES = FALSE controls loading files with uncertain load status, but it is not the primary parameter for avoiding partial datasets caused by load errors.

D. ON_ERROR = ABORT_STATEMENT is the correct parameter because it stops the load when an error occurs.

Official Snowflake documentation reference:

Snowflake documentation for COPY INTO < table > describes ON_ERROR = ABORT_STATEMENT as the option that aborts the load operation when an error is encountered.