SOL-C01 SnowPro Associate: Platform Certification Exam Questions and Answers

Cortex LLM functions integrate large language model capabilities directly inside Snowflake. These functions enable:

• Sentiment analysis

• Summarization

• Document question answering

• Text classification

• Embedding generation for vector search

They operate directly on Snowflake data, require no external infrastructure, and support secure in-platform AI workloads.

====================================================

TheLISTcommand displays all files available in a stage—internal, external, user, or table stages.

It is typically used prior to COPY INTO to confirm which files are available for loading.

SHOW TABLES lists tables; SHOW GRANTS shows privileges; GRANTS queries show user-role assignments—not LIST.

==================

SHOW GRANTS is used to list privileges, including roles assigned to a user.

SHOW GRANTS TO USER <user> reveals all the roles that user inherits.

SHOW ROLES lists roles, SHOW USERS lists users, SHOW DATABASES lists databases—not SHOW GRANTS.

==================

Standard Edition provides all essential Snowflake capabilities:

• SQL data warehousing

• Secure data sharing

• Time Travel (1 day)

• Internal/external stages

• Basic governance features

It does not include HIPAA/PCI compliance (Business Critical), nor unlimited Time Travel, nor advanced private connectivity.

====================================================

Snowflake Notebooks support both SQL and Python as executable cell types. This means that users can create notebook cells written in Python for programmatic data processing and modeling, and in SQL for declarative querying and transformation. Because of this, the statements “Notebook cells can be written in Python” and “Notebook cells can be written in SQL” are both correct.

Snowflake databases do not natively execute queries written in Scala; Scala is supported via Snowpark APIs for application code, not as a direct query language. Worksheets in Snowsight are primarily SQL-based and also support procedural constructs (e.g., Snowflake Scripting), so the phrase “only in SQL” is not a precise or complete characterization. Dashboards in Snowsight are created using SQL-backed visualizations and built-in UI components; JavaScript is not a supported authoring language inside the native dashboarding layer. Therefore, options B, D, and E are not correct in this context.

==================

Auto-scale enables Snowflake to dynamically start or stop clusters in a multi-cluster warehouse depending on workload pressure. When concurrency rises, Snowflake adds clusters; when demand decreases, it scales back down. This provides seamless performance during peak workloads and efficient resource usage during off-peak times. Auto-scale does not require manual action—it is fully automated once configured. It is not limited to standard warehouses; multi-cluster mode is available in Enterprise Edition and above. Ultimately, auto-scale ensures elasticity and cost-efficient compute allocation.

=======================================

Aschemain Snowflake is a logical container inside a database that groups related objects. It can containtables,views,stages,file formats,sequences,streams,tasks, and other database objects. Schemas allow structured organization, privilege management, and separation of workloads.

Tablesstore structured data.

Viewsare virtual tables defined by queries on other objects.

Stagesstore data files for loading/unloading.

Virtual warehouses are compute resources, not database objects, and therefore exist outside schemas. Warehouses belong to the account-level namespace and are managed independently of databases and schemas.

Schemas play a key role in RBAC and object naming through fully qualified names: database.schema.object.

====================================================

A fully qualified name uniquely identifies Snowflake objects by specifyingdatabase.schema.object. This prevents ambiguity when multiple schemas or databases contain objects with identical names. Fully qualified names ensure that SQL statements operate on the intended object. They are not used for storing data, managing permissions, or configuring network settings. Their core purpose is precise object identification.

The LIMIT clause limits the number of rows Snowflake returns from a query. It is commonly used to preview datasets, improve performance when exploring large tables, and paginate results. LIMIT may be combined with OFFSET to skip a number of rows. Sorting is performed with ORDER BY, filtering with WHERE, and grouping with GROUP BY—not with LIMIT. Therefore, restricting returned rows is the primary purpose of LIMIT.

=======================================

Snowflake'sINFORMATION_SCHEMAcontains metadata views about databases, schemas, tables, columns, constraints, and other objects. To retrieve a list of tables, the correct view isINFORMATION_SCHEMA.TABLES, which returns:

Table name

Schema name

Table type (BASE TABLE, VIEW, MATERIALIZED VIEW)

Creation timestamp

Owner and retention properties

The query:

SELECT * FROM INFORMATION_SCHEMA.TABLES;

returns metadata for all tables in thecurrent databasecontext.

INFORMATION_SCHEMA.COLUMNS returns column-level metadata, not table listings.

PUBLIC.INFORMATION_SCHEMA is invalid—INFORMATION_SCHEMA is not inside PUBLIC schema.

PUBLIC.TABLES is not a system object unless created manually.

====================================================

The Query Processing layer of Snowflake is wherevirtual warehouses operate, so warehouse sizing parameters (X-Small to 6X-Large) fall under this layer. Warehouse size determines compute power, concurrency, and performance behavior for SQL workloads. Administrators configure warehouse size based on workload intensity, response time requirements, and cost considerations.

Serverless compute limits and micro-partition limits belong to storage and services layers. Table types (permanent, transient, temporary) are storage-level configurations, not part of Query Processing.

Thus, warehouse sizing is the correct parameter configured at the Query Processing layer.

==================

Snowflake supports theEXCLUDEkeyword to simplify queries when excluding certain columns from a SELECT * operation. SELECT * EXCLUDE (column1, column2) reduces verbosity and enhances maintainability, especially when table schemas evolve. Explicitly listing all columns is possible but inefficient. Snowflake does not support REMOVE functions for columns nor an OMIT clause. EXCLUDE is the correct and official mechanism.

=======================================

PARSE_DOCUMENT extracts text and, depending on mode, layout information from supported document formats. The function currently supportsPDF,PPTX, andDOCXfiles. For PDFs, it can perform OCR to extract scanned text or use LAYOUT mode to detect tables and structured regions. For PPTX files, it extracts slide text and content. For DOCX files, the function captures text and applies layout analysis. XLSX files are not supported—Excel content requires different extraction methods. PARSE_DOCUMENT enables automated document intelligence workflows entirely inside Snowflake, eliminating the need for external OCR or parsing services.

=======================================

Schemas in Snowflake serve asorganizational containersfor database objects such as tables, views, sequences, stages, file formats, and functions. They existwithin a database, forming Snowflake’s logical hierarchy: account → database → schema → object.

Schemas are not limited to a single role; access depends on granted privileges. Schemas are not immutable—administrators can rename, alter, or drop them. Schemas do not contain virtual warehouses; warehouses exist independently at the account level.

Thus, the correct characteristics are that schemas organize objects and are contained within a database.

==================

Query History in Snowsight includes:

• Query ID – A unique identifier for each executed query.

• Execution Status – Indicates whether the query succeeded, failed, or was canceled.

• Duration – Total time taken for the query to run.

While warehouse name is shown,warehouse sizeis not explicitly displayed. Credit consumption appears in Account Usage views, not directly in Query History.

====================================================

Internal stages are fullySnowflake-managed storage locationsused to temporarily store data files for loading or unloading operations. Examples include:

User stages (@~)

Table stages (@%table_name)

Named internal stages (@my_internal_stage)

Because they resideinside Snowflake, users do not need AWS, Azure, or GCP credentials to access them. Internal stages support all file types Snowflake can process, including CSV, JSON, Parquet, Avro, ORC, images, PDFs, and other semi-structured formats.

Incorrect statements:

They arenot limitedto structured or semi-structured data—they accept all supported formats.

External cloud credentials are only needed for external stages (S3, Blob, GCS).

Internal stages simplify secure data ingestion and operational pipelines because Snowflake handles encryption, metadata, and lifecycle management.

====================================================

Snowflake’s object hierarchy maintains a structured namespace. At the top is thedatabase, which serves as a container for related schemas. Eachschemaorganizes logical groupings of tables, views, stages, file formats, and other database objects.

Tablesstore persistent, structured data.

Viewsprovide virtual representations of data through stored SQL queries. They do not store data themselves but reference underlying tables.

The hierarchical structure is:

.<schema>.<object>

Incorrect statements:

Views do not contain databases—they reference objects but do not encapsulate them.

Schemas do not manage network or cloud settings; those are defined at account or cloud service layers.

This hierarchy enables clear object governance, secure sharing, and proper namespace organization.

====================================================

Every query executed in Snowflake is assigned a uniqueQuery ID, a system-generated identifier that allows users to retrieve, analyze, and reference the query. Query IDs support debugging, profile retrieval, audit tracking, workload analysis, and support cases.

Transaction IDs are only assigned to transactional operations and do not uniquely identify all queries.

Session IDs identify user sessions—not individual queries. Many queries can occur in the same session.

Account ID is the identifier for the entire Snowflake account and does not relate to specific queries.

Thus,Query IDis the unique attribute used to identify each query in Query History.

==================

ThePARSE_DOCUMENTfunction offers two primary processing modes:OCRandLAYOUT, each designed for different document-intelligence tasks.

OCR (Optical Character Recognition)mode extracts raw text from scanned or image-based documents. This is ideal for contracts, scanned PDFs, receipts, manuals, and any scenario where the primary goal is text extraction. OCR focuses on textual content without preserving document structure.

LAYOUTmode extracts both text and structural context, including tables, section groupings, paragraph blocks, and visual positioning. This provides layout-aware data, making it vital for downstream use cases such as Retrieval-Augmented Generation (RAG) pipelines, structured extraction, and content summarization while retaining relationships between document components.

CONTENTis not a mode—it is a JSON key in output.OMR(Optical Mark Recognition) is not supported within PARSE_DOCUMENT.

====================================================

The PARSE_DOCUMENT function is part of Snowflake Cortex AI and is designed specifically to extract text, layout information, and structured elements from unstructured documents, especially PDFs. It supports OCR-based extraction for scanned files and layout-aware extraction to preserve tables, headings, and format structure.

Its purpose is not PII detection; Snowflake does not provide built-in automatic PII identification via PARSE_DOCUMENT. It does not identify candidate data for directory tables and is unrelated to JSON parsing—Snowflake uses PARSE_JSON for JSON data.

PARSE_DOCUMENT is primarily used for workflows such as contract analysis, invoice extraction, document classification, compliance automation, and downstream AI enrichment.

==================

GET_PRESIGNED_URL() is the official Snowflake function used to create atemporary, secure, time-limited pre-signed URL, which provides controlled-access retrieval of a file stored in a Snowflake internal or external stage. Pre-signed URLs are essential for secure file sharing without exposing credentials or granting direct stage access to external users. The function accepts the stage name and file path and optionally allows specifying an expiration period. GET_STAGE_LOCATION() retrieves metadata about where a stage points, not access URLs. BUILD_STAGE_FILE_URL() constructs a basic URL but does not sign it, meaning it cannot authorize secure download access. GET_RELATIVE_PATH() provides path-only details, not a secure download link. GET_PRESIGNED_URL ensures that external consumers can access staged files safely, making it an important feature for secure distribution workflows, data export pipelines, partner integrations, and temporary file sharing scenarios.

A virtual warehouse in Snowflake is acompute resourcethat supplies CPU, memory, and temporary storage used to process SQL queries, perform data loading, execute DML operations, and run procedural logic. Warehouses are elastic, can be resized on demand, suspended to save costs, and automatically resumed when needed.

Virtual warehouses donotstore data—data is kept in Snowflake’s centralized storage layer. They also are not user accounts; user accounts belong to Snowflake’s identity management system. Worksheets are UI components in Snowsight or Classic Console used to write and run SQL, not compute resources.

Thus, virtual warehouses are Snowflake's compute engine, enabling all processing without managing physical infrastructure.

==================

The DATA_RETENTION_TIME_IN_DAYS parameter controls the Time Travel retention period for Snowflake objects such as tables, schemas, and databases. It specifies the number of days that historical data (prior versions of rows or dropped objects) is retained and accessible via Time Travel. Within this retention window, users can query data “as of” a previous time, restore dropped objects, or clone objects at a historical point.

DATE_OUTPUT_FORMAT determines the display format for date values and is unrelated to historical retention. TIMEZONE affects how timestamps are interpreted and displayed, not how long data history is preserved. USE_CACHED_RESULT governs whether Snowflake may return cached query results, not Time Travel behavior.

==================

To create a new schema in Snowsight, users must follow the hierarchical navigation of Snowflake’s object structure. The correct sequence is:

Data → Database → Create Schema

Steps:

Open theDatatab in the left navigation.

Select aDatabasewhere the schema will be created.

ChooseCreate → Schema.

Enter the schema name and (optionally) description.

ClickCreate.

Incorrect options:

“Add Data” is used for loading files, not creating schema objects.

“Data → Create Schema” is incomplete because Snowflake must knowwhich databasethe schema belongs to.

“Data Product” is unrelated to object creation.

Thus, Snowflake requires explicit database context before schema creation.

====================================================

A Snowflake database is a top-level container that can include schemas, and within those schemas it can contain objects such astables,views,stages,functions,procedures, anddatabase roles. However,global account-level roles(such as SYSADMIN, SECURITYADMIN, PUBLIC, or custom account roles) do not reside inside a database. These roles are created and managed at the Snowflake account level, not within any specific database. In contrast,database rolesare scoped within a specific database and can exist inside it. Views and tables are standard database objects. Therefore, the only option that a Snowflake database cannot contain is an account-levelRole.

=======================================

SnowflakeData Exchangeenables secure, governed sharing oflive databetween Snowflake accounts. It lets providers publish datasets to specific consumers—partners, vendors, departments, or customers—without copying or transferring data. Consumers query shared data instantly, leveraging Snowflake’s secure data sharing architecture.

Key capabilities include:

Targeted sharing to selected accounts

Fine-grained access management and auditing

Live, real-time data sharing without data movement

Ability to customize listings and visibility

Data Exchange is different from theMarketplace, which supports public or broad distribution.

It does not store raw data—storage occurs in provider accounts.

It does not manage user permissions (that is the role of RBAC).

====================================================

Directory tables allow Snowflake to expose metadata about files stored in an external stage, enabling users to query file listings using SQL. To enable a directory table, the correct approach is to update an existing external stage using:

ALTER STAGE my_stage SET DIRECTORY = (ENABLE = TRUE);

This enables automatic synchronization of file metadata from cloud providers such as AWS S3, Azure Blob, or Google Cloud Storage.

Incorrect options:

Directory tables arenot enabled by default.

There is noCREATE DIRECTORY TABLEcommand in Snowflake.

Creating a view cannot enable directory table functionality.

Once enabled, users can query:

SELECT * FROM DIRECTORY(@my_stage);

to retrieve file names, sizes, MD5 hashes, and timestamps.

Snowsight supports two worksheet types:SQL worksheetsandPython worksheets. SQL worksheets allow users to execute queries, create objects, and perform data analysis using ANSI SQL and Snowflake-specific extensions. Python worksheets, powered by Snowpark, allow users to write Python code that interacts directly with Snowflake tables, data frames, and machine learning workflows.

Java, Scala, and JavaScript are supported via Snowpark APIs or UDF development, but they cannot be used as worksheet languages. Worksheets are designed for interactive analysis, visualization, and iterative development, with native runtimes only for SQL and Python.

Thus, only SQL and Python worksheets can be created within Snowsight.

==================

Streamlit inside Snowflake Notebooks enables users to buildinteractive, real-time visualizations and data applicationsdirectly within Snowsight. Because Streamlit is event-driven, users can create UI components such as sliders, dropdowns, date pickers, buttons, and filters that dynamically update charts, tables, and computed outputs.

Data remains inside Snowflake, processed through Python with Snowpark, eliminating data movement to external systems. This allows rapid prototyping of dashboards, analytical workflows, and AI-driven apps. Streamlit also supports layout customization, reactive components, and seamless integration with Snowflake compute for scalable analytics.

Incorrect options:

Streamlit doesnotmanage data storage.

Streamlit does not simplify code execution; it enhances visualization and interactivity.

Warehouse clusters are not fixed-size—Snowflake warehouses can scale.

Thus, its primary benefit is enablingrich interactive analytics directly within Snowflake.

====================================================

Snowflake supports two official stage types:

• Internal Stages – User stages, table stages, and named internal stages stored inside Snowflake.

• External Stages – Refer to cloud storage (S3, Azure Blob, GCS) using integration credentials.

Managed, permanent, and temporary stages are not official Snowflake classifications.

====================================================

COPY INTO <title> bulk-loads data from internal or external stages into Snowflake tables. It supports transformations, validation, and multiple file formats.

It does not replicate databases, manage roles, or create warehouses.

====================================================

When auto-suspend is enabled on a virtual warehouse, Snowflake automatically suspends (stops) the warehouse after it has been idle for the configured period of time. Once suspended, the warehouse stops consuming compute credits, thereby reducing costs, since Snowflake bills only for active compute usage. When new queries arrive, auto-resume (if enabled) restarts the warehouse automatically.

Auto-suspend does not change the warehouse size; scaling up or down is a separate configuration. The warehouse is not dropped—its metadata and settings remain intact. Costs may be reduced and made more efficient but are not guaranteed to “stabilize month-to-month,” as the bill still depends on actual usage and workload patterns.

==================

The default Time Travel retention period for most objects in Snowflake is1 day(24 hours). Time Travel enables access to historical versions of data after updates, deletes, or drops. It also allows cloning databases, schemas, and tables at previous points in time, and recovering dropped objects.

While Snowflake Enterprise Edition and higher tiers allow retention periods up to 90 days, this extended window is not the default—administrators must explicitly configure it for each table, schema, or database.

Retention periods of 7, 45, or 90 days are possible only with higher service editions; the default for all accounts and objects remains 1 day unless explicitly overridden.

==================

Snowflake is a fully cloud-native platform that supports deployment across all three major cloud providers:AWS, GCP, and Azure. Snowflake delivers consistent functionality across each provider, allowing organizations to choose the cloud best suited for regulatory, architectural, and business requirements. Although small feature differences may exist between cloud providers, Snowflake’s core capabilities—virtual warehouses, storage layers, security, data sharing, and governance—operate uniformly across all three platforms. This multi-cloud support gives Snowflake strong flexibility for hybrid, multi-region, and multi-cloud deployments.

=======================================

A view in Snowflake represents avirtual tablewhose contents are defined by a stored SQL query. Views allow users to encapsulate transformation logic, simplify complex joins, enforce column-level security, and provide curated datasets for downstream consumers.

Views donotstore data themselves; the underlying tables store the data, and Snowflake executes the view’s query each time the view is referenced. This makes views ideal for abstraction layers, semantic modeling, and separating compute costs across user groups.

Incorrect options:

Snowflake views do not store raw data.

Views have no role in authentication or network configuration.

Views help streamline business logic and control data access efficiently.

====================================================

Snowsight’s Query History view allows filtering queries bystatus,date range, andexecuting user. Additional filters include SQL text search, warehouse, duration, statement type, session ID, query ID, query tag, parameterized query hash, and whether a query was generated by a client or task. These filters help users identify long-running, failed, queued, or specific types of queries for troubleshooting and performance optimization. Network or hardware configurations have no role in Query History filtering.

=======================================

Snowflake is a fully managed, cloud-native platform that automates nearly all infrastructure tasks. Two defining characteristics are:

1. Snowflake handles platform maintenance, management, and upgrades.

Snowflake automatically performs software updates, scaling operations, tuning, optimization, and patching. Users never need to upgrade hardware, install software, or manage clusters.

2. There is no infrastructure for customers to provision and maintain.

All compute, storage, and services run on public clouds (AWS, Azure, GCP). Snowflake abstracts the underlying infrastructure completely, eliminating the need for customers to manage VMs, disks, clusters, or networking.

Incorrect options:

• Snowflake supportsbothcolumn-level security (via masking policies)androw-level security (via row access policies).

• Snowflake is not responsible foralldata security—customers must configure RBAC, masking, and governance.

• Snowflake doesnotrun on-premises; it is exclusively cloud-based.

Thus, options C and D correctly describe Snowflake platform characteristics.

==================

The SQL command DROP SCHEMA IF EXISTS marketing_schema; instructs Snowflake to drop the specified schema only if it already exists in the database. The inclusion of the IF EXISTS clause prevents Snowflake from returning an error when the schema is not found. If the schemadoesexist, Snowflake drops it along with all objects it contains, unless a RESTRICT clause is enforced (RESTRICT is the default when not explicitly specified, but Snowflake also supports CASCADE for forcing deletion of all objects). Importantly, this command never creates a schema; it strictly performs a conditional drop. In addition, Snowflake supports Time Travel, so the dropped schema can typically be restored—using UNDROP SCHEMA—within the defined retention window. Therefore, the correct interpretation is that the schema is dropped only if it exists, and no error is thrown otherwise.

=======================================

TheData Storage Layerin Snowflake stores all data in acompressed, encrypted, columnar formatdesigned for analytical efficiency. Snowflake automatically reorganizes ingested data into micro-partitions optimized for pruning and high-performance scans. The compression and structuring are fully automated, requiring no tuning or indexing by the user.

Another foundational feature iselastic storage, made possible by Snowflake’s cloud-native architecture. Storage automatically scales up or down based on data volume without user intervention, and is separated from compute. This ensures cost efficiency and immediate availability of capacity for large datasets.

Incorrect options:

In-memory storageis not a Snowflake feature; long-term data is persisted in cloud object storage.

Elastic Cloudis not a Snowflake architectural term.

The Data Storage Layer is designed for durability, scalability, and optimal query performance.

====================================================

On the Snowflake Marketplace, users can apply a Price filter when searching for data products. This filter helps distinguish between free offerings and paid listings, making it easier to align data acquisition with budget and commercial terms. Marketplace metadata includes pricing information that this filter uses to refine search results.

Dataset “Size” is not exposed as a primary Marketplace search filter. “Compatibility” is implied, as all Marketplace products are consumable within Snowflake. “Performance” is not a filterable attribute, because performance is influenced mainly by the consumer’s compute configuration and query design, rather than a fixed property of the Marketplace listing.

==================

The recommended approach in Snowflake’s RBAC model is togrant privileges to roles, not to individual users or through ownership transfers. Privileges such as SELECT, INSERT, USAGE, and MODIFY are granted to roles, and roles are then assigned to users—creating scalable governance.

Transferring ownership is a powerful action and should not be used simply to provide access. The PUBLIC role should not be used to grant sensitive privileges since it is granted to all users by default. Cloning objects is unnecessary and creates new objects, not access to the original.

Thus, granting privileges to the role is the correct RBAC practice.

==================

A virtual warehouse is the compute engine of Snowflake. It provides CPU, memory, and temporary storage needed to execute SQL queries, data loading operations, and DML actions. Warehouses can be sized dynamically and suspended or resumed to optimize cost.

Warehouses donotstore data; Snowflake’s storage is independent and centralized. Warehouses do not manage roles—access control is handled through Snowflake’s RBAC system. Encryption is performed automatically by Snowflake’s storage and cloud services, not by warehouses.

Thus, the correct characteristic is that virtual warehouses supply compute.

==================

If you'd like, I can provideQuestions 42–55 next, with the same 150–200-word explanations.

You said:

Questions 42–55 next, with the same 150–200-word explanations

ChatGPT said:

Below areQuestions 42–55, all formatted exactly per your requirements, with150–200-word explanationsandevery original option preserved exactly.

The Database Storage Layer is responsible for persistently storing all customer data in Snowflake. This layer automatically manages micro-partitioning, compression, and metadata. Data is stored in optimized, columnar formats inside cloud object storage (AWS S3, Azure Blob Storage, or GCP Storage). Users cannot manipulate physical storage directly; access is only through SQL operations. Other layers serve different purposes: Cloud Services manages metadata and orchestration, Query Processing Layer executes workloads, and Virtual Warehouses provide compute but do not store data.

====================================================

Unstructured data such as PDF files, images, and other binary documents is stored in stages in Snowflake. These stages may be internal stages, which Snowflake manages directly, or external stages, which reference external cloud storage such as Amazon S3, Azure Blob Storage, or Google Cloud Storage. Stages are the designed mechanism for storing and accessing unstructured files so that they can be processed with functions like PARSE_DOCUMENT or accessed via directory tables.

External tables are used to query structured or semi-structured data (for example, Parquet or JSON) stored in external locations, not to store raw unstructured binary content. The Cloud Services layer coordinates metadata, security, and query services; it does not store user data. Tables with a single VARCHAR column might be used as an improvised approach for small text blobs, but this is not the native or recommended method for managing unstructured data at scale.

==================

The correct SQL syntax to create a zero-copy clone of an existing table is:

CREATE TABLE MY_TABLE_CLONE CLONE MY_TABLE;

This command instantly creates a new table that references the same underlying micro-partitions as the original. Because of Snowflake’smetadata-only cloning, no storage is consumed at the time of creation. Storage only increases when either the original or the clone diverges through DML operations, following acopy-on-writemodel.

Cloning is available for multiple object types—tables, schemas, databases, stages, streams, tasks, and more. This capability enables rapid creation of development sandboxes, QA environments, rollback copies, or controlled experimentation without duplicating data.

Incorrect options:

“COPY TABLE” is not a valid Snowflake command.

BACKUP/RESTORE are not Snowflake SQL commands.

RESTORE applies only to Time Travel or Fail-safe, not to cloning.

Thus, the CLONE keyword is the only correct method for zero-copy duplication.

====================================================

A multi-cluster warehouse raises Snowflake’s capability to handle concurrent workloads efficiently. When large numbers of users or queries hit the system, Snowflake can automatically start additional clusters (scaling out) to distribute workloads, reducing queuing and improving performance. As query demand drops, Snowflake scales in by shutting down clusters to conserve credits. This dynamic auto-scaling provides significantly increased compute capacity during peak usage periods and enhances responsiveness. Importantly, multi-cluster warehouses affect only compute, not storage, so they do not reduce storage costs. Their primary benefits revolve around performance stability, concurrency handling, workload isolation, and compute elasticity.

=======================================

The PARSE_JSON() function ingests a string containing JSON text and converts it into Snowflake’sVARIANTdata type, enabling the JSON to be queried, navigated, and transformed using SQL. Snowflake does not store JSON in its raw textual representation; instead, VARIANT allows Snowflake to apply optimized parsing, indexing, and querying operations against semi-structured content. This function is particularly useful when JSON arrives inline (e.g., supplied directly within SQL statements or loaded from CSV files containing JSON strings). PARSE_JSON does not perform data loading from stages—that is handled through COPY INTO—nor does it convert JSON into XML. Once JSON is converted to VARIANT, Snowflake allows access to nested structures using dot notation, bracket notation, and functions like FLATTEN(). Thus, the function serves as a bridge between raw JSON strings and Snowflake’s relational and analytical capabilities.

=======================================

Snowflake provides multipledate and time data typesto handle various temporal workloads. TheDATEtype stores only the calendar date (year, month, day), suitable for dimensional modeling, slowly changing dimensions, and calendar-related analytics.

TIMESTAMPrepresents a point in time and includes sub-second precision. Snowflake supports multiple variants:

TIMESTAMP_NTZ(no time zone)

TIMESTAMP_LTZ(local time zone)

TIMESTAMP_TZ(explicit time zone)

These options allow flexibility for global operations, event logs, and time-series analytics.

DATETIMEis an alias forTIMESTAMP_NTZ, meaning it stores both date and time but without time-zone awareness. It is commonly used in ETL, application logs, and system-generated events.

TIMEDATEis not a valid Snowflake data type and does not exist in Snowflake’s type system.

These temporal types support extensive built-in datetime functions, automatic casting, and integration with semi-structured data through VARIANT.

====================================================

Snowflake uses the COPY INTO <title> command as the primary mechanism for bulk loading semi-structured data, including JSON, from internal or external stages. COPY INTO is optimized for large file ingestion and supports file format specifications, such as TYPE=JSON, enabling Snowflake to parse, validate, and store JSON into the VARIANT data type. Unlike INSERT INTO, which is suitable for row-level inserts, COPY INTO supports large-scale ingestion with parallelization and error-handling parameters. Commands such as LOAD DATA or UPLOAD JSON are not part of Snowflake’s SQL interface. COPY INTO works seamlessly with both internal stages (USER, TABLE, or NAMED stages) and external cloud storage stages (S3, Azure Blob, GCS). Once the JSON is loaded, users can query nested fields using dot or bracket notation, flatten arrays for relational use, and build transformations directly in SQL. Therefore, the COPY INTO command is the correct and supported method for loading JSON files from a Snowflake stage.

=======================================

Snowflake’s database storage layer is responsible for encrypting data at rest. All persisted data—whether structured, semi-structured, or unstructured—is encrypted using strong encryption algorithms such as AES-256. This process is automatic and transparent to users, ensuring that files and micro-partitions stored in Snowflake-managed cloud storage are encrypted by default. Data in transit is protected separately through TLS.

Virtual warehouses provide compute resources to execute queries and do not perform storage-level encryption. Data loading utilities (such as COPY INTO and client tools) orchestrate data movement but do not handle at-rest encryption. The Cloud Services layer manages metadata, session management, security policies, and query optimization, but the actual encryption of stored data is part of the storage subsystem.

==================

Snowflake uses a hierarchical Role-Based Access Control (RBAC) model. When Role A is granted to Role B, the higher-level role (Role B) inherits all privileges of the lower-level role (Role A). This upward inheritance allows broad administrative roles (such as SYSADMIN) to automatically inherit permissions granted to subordinate roles.

Key principle:privilege inheritance flows upward, not downward.

• Parent roles inherit from child roles.

• Child roles donotinherit from parent roles.

Sibling roles do not share privileges unless explicitly granted.

Thus, roles above a given role in the hierarchy inherit privileges, making option C correct.

Options A and D incorrectly describe downward inheritance, which Snowflake does not perform.

Option B (same-level inheritance) also does not occur unless privileges are explicitly granted.

This hierarchical model simplifies governance by allowing higher-level administrative roles to centrally manage access across multiple subordinate roles.

==================

Snowflake notebooks use the.ipynbfile extension, the standard format for Jupyter notebooks. This format stores executable code, markdown, metadata, and cell outputs in a structured JSON layout. Snowflake adopts this format to ensure compatibility with the broader Python ecosystem, thereby enabling seamless migration between Snowflake and external notebook environments.

The .ipynb structure allows mixed SQL and Python cells, visualizations, Streamlit components, documentation, and stepwise development within Snowsight. It supports reproducibility, collaboration, and integration with Snowpark and Cortex.

Incorrect formats:

.ipnbis a misspelling and invalid.

.sqlis used for SQL scripts only.

.txtcannot represent notebook metadata or cell structure.

Thus, .ipynb is the correct and only supported notebook format.

====================================================

In Snowflake’s Secure Data Sharing model, the accountsharingdata is known as thedata provider. Providers create shares that include selected objects (tables, views, secure views, etc.) and grant access to consumer accounts.

Adata consumerreceives the shared data.Data readeranddata administratorare not defined terms in Snowflake’s sharing architecture.

==================

The Cloud Services Layer is responsible for generating and managing metadata, including object definitions, table schemas, micro-partition statistics, column-level profiles, access control information, and query optimization metadata. Metadata plays a central role in Snowflake’s performance and functionality because it informs pruning, query planning, and efficient execution.

Dashboards are created in Snowsight or external BI tools. Virtual warehouses belong to the Compute Layer, providing processing resources. Micro-partitions are created in the Storage Layer, where Snowflake automatically organizes compressed columnar data for efficient access.

Consequently, the Cloud Services Layer is where metadata—not data, not compute resources—is created and managed.

==================

As a fully managed cloud data platform, Snowflake is responsible for infrastructure provisioning, hardware, software installation, platform upgrades, scaling, and internal metadata management such as micro-partitions and statistics. Customers do not manage physical hardware or install Snowflake software.

Customers are responsible for their data and its lifecycle within Snowflake. This includes loading data into tables from internal and external sources, unloading data when required, organizing data structures (databases, schemas, tables), defining access controls, and managing how data is used, transformed, and governed. They design schemas and workloads but do not manage the underlying engine. Therefore, “Loading, unloading, and managing data” correctly describes the customer’s responsibility.

==================

A virtual warehouse provides compute resources in Snowflake and can be resized (scaled up or down) at any time. Scaling up increases compute power for intensive workloads, while scaling down reduces cost for lighter workloads.

Virtual warehouses do not store data—that is handled by Snowflake’s independent storage layer. They are not on-premises and do not provide unlimited concurrency; Snowflake supports high concurrency using multi-cluster warehouses but not boundless concurrency.

==================

The COPY INTO <title> command is Snowflake’s primary and optimized mechanism for loading large volumes of structured data from either internal or external stages. Unlike INSERT INTO, which is used for row-level inserts or small datasets, COPY INTO is designed for high-throughput operations, parallel loading, and performant ingestion from files stored in S3, Azure Blob Storage, GCS, or Snowflake-managed stages. The COPY INTO command supports detailed file formatting options (such as TYPE=CSV, FIELD_DELIMITER, SKIP_HEADER), error-handling rules (ON_ERROR), and transformations using SQL expressions inside the COPY statement. Snowflake does not support LOAD DATA or IMPORT DATA as SQL commands; these exist in other database systems but not within Snowflake. COPY INTO ensures efficient, resilient, fault-tolerant data ingestion and is heavily used for ETL/ELT, data lake ingestion, and large batch ingestion pipelines.

=======================================

When auto-resume is enabled, Snowflake automatically starts a suspended warehouse when a query or task is submitted. This ensures smooth operations without manual intervention.

It does not scale the warehouse, prevent suspension, or require manual restarting.

====================================================

In Snowflake’s logical hierarchy, tables reside within schemas, and schemas reside within databases. This structured arrangement allows clear organization, access management, and separation of workloads or business domains. A schema functions as a container that holds tables, views, functions, sequences, stages, and file formats. When users reference a table, they typically qualify it using the full three-part name: database.schema.table.

Stages are used to store or reference files, not to hold structured database tables. Virtual warehouses provide compute power for executing queries and performing DML operations but do not store data. File formats define how staged files should be interpreted (CSV, JSON, Parquet, etc.) and are unrelated to where tables physically reside.

Snowflake separates compute and storage, so while compute happens in virtual warehouses, all persistent data—including tables—is managed in Snowflake’s centralized storage layer, organized through schemas and databases. This organization ensures consistency, ease of management, and predictable access control.

==================

Semi-structured data like JSON stored in a VARIANT column can be queried directly usingpath notation, either dot notation or bracket notation.

Examples:

Dot notation:SELECT data:customer.name FROM table;

Bracket notation:SELECT data['customer']['name'] FROM table;

Snowflake automatically interprets the JSON structure, making relational extraction unnecessary. Complex fields can be accessed through functions such as FLATTEN, OBJECT_KEYS, ARRAY_SIZE, and TYPEOF.

Incorrect options:

Converting JSON to a relational table is optional, not required.

Stored procedures are not necessary for JSON querying.

WHERE on the full VARIANT cannot precisely extract fields.

Thus, path notation is the native, efficient method.

====================================================

Unstructured data refers to content that lacks a predefined schema or consistent structural pattern. It cannot be represented naturally in tabular or hierarchical formats without extraction or transformation. Snowflake supports unstructured data through external tables and native features allowing storage and machine learning analyses.Images,videos, andPDFsfall into this category because they contain raw multimedia or document content without intrinsic row/column structure, often requiring OCR, video/audio processing, or content extraction. In contrast, JSON is semi-structured, containing keys, hierarchical structures, and metadata that Snowflake’s VARIANT type can interpret directly. Unstructured data is becoming increasingly important for AI-driven workloads, enabling use cases such as OCR with PARSE_DOCUMENT, deep-learning analysis of images and video, and document intelligence tasks.

=======================================