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.

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

ADatabaseis the highest logical container for schemas and their objects in Snowflake.

Hierarchy:

Account → Database → Schema → Tables, Views, Stages, etc.

Schemas organize objects within a database; tables and views do not contain objects.

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

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.

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

PARSE_DOCUMENT supports two processing modes:OCRandLAYOUT. OCR mode performs Optical Character Recognition, extracting raw text from scanned documents, images within PDFs, or low-quality text-based documents. It is ideal for scenarios like contract ingestion, receipt processing, or older scanned documents. LAYOUT mode extracts structured layout elements—tables, paragraphs, lines, bounding boxes—preserving the original document’s spatial organization. This enables downstream analytical tasks such as table reconstruction or semantic segmentation of document content. OMR (Optical Mark Recognition) is not a supported feature, and “CONTENT” is not a valid mode. By supporting OCR and LAYOUT modes, Snowflake Cortex provides robust document intelligence capabilities directly within the Snowflake environment.

The SELECT * statement instructs Snowflake to returnall columnsfrom the referenced table or view. This is commonly used during data exploration, debugging, initial data profiling, and validation steps. It allows users to quickly view the complete dataset structure without manually specifying each column name.

However, while SELECT * retrieves all columns, it does not limit the number of rows. To restrict rows, developers must include aLIMITclause (e.g., SELECT * FROM table LIMIT 10;).

The query does not automatically apply DISTINCT or primary key filtering—Snowflake returns all rows exactly as stored unless additional filtering, WHERE conditions, or ordering are provided.

Though SELECT * is convenient, Snowflake best practices recommend explicitly selecting columns in production workloads to optimize performance and avoid unnecessary scanning of unused fields.

USERADMIN is one of Snowflake’ssystem-defined roles, created automatically in every account. It is responsible for managing users and roles, including CREATE USER, ALTER USER, and role assignment. It is part of Snowflake’s default RBAC hierarchy (SYSADMIN, SECURITYADMIN, USERADMIN, etc.).

SNOWFLAKE_ADMIN and SNOWFLAKE_DBA are not Snowflake system roles—they may exist in organizations as custom roles but do not appear by default. DATA_ENGINEER is also user-created and not a built-in role.

Therefore, USERADMIN is the only true system-defined role listed.

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

Snowflake Notebooks support cross-cell interaction between Python and SQL by using Jinja-style templating syntax. To reference a Python variable inside a SQL cell, you wrap the variable name in double curly braces, like {{myvar}}. During execution, the Notebook engine substitutes the Python variable’s value into the SQL statement before sending it to Snowflake.

This mechanism allows dynamic SQL generation, parameterization of queries, incorporating Python logic into SQL workflows, and building interactive analytics pipelines.

Other provided options are invalid in Snowflake Notebooks: $myvar resembles shell syntax and is not supported; ' myvar ' inserts a literal string rather than the variable ' s value; using myvar alone would cause SQL to interpret it as a column or object name.

Therefore, only {{myvar}} correctly represents Snowflake Notebook variable substitution syntax.

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

Snowflake supports several native data types for managing temporal information:

DATE – Stores dates only (year, month, day) without any time component.

TIMESTAMP – Represents date and time values, including several variants:

• TIMESTAMP_NTZ (no time zone)

• TIMESTAMP_LTZ (local time zone)

• TIMESTAMP_TZ (explicit time zone)

DATETIME – An alias for TIMESTAMP_NTZ, storing both date and time.

These data types enable powerful time-based analytics and comparisons.

❌TIMEDATE is not a valid data type.

❌Python and JavaScript are programming languages, not Snowflake data types.

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

COPY INTO < table > 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.

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

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.

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