SnowPro Advanced Data Engineer key terms

A named location holding data files for loading into or unloading out of Snowflake. Internal stages are managed by Snowflake; external stages point at cloud storage such as S3, GCS, or Azure Blob.

An account-level object that stores the cloud identity and permissions Snowflake uses to read and write an external storage bucket, so individual stage definitions do not need embedded credentials.

A reusable named definition of how to parse staged files: type (CSV, JSON, Parquet and so on), delimiter, compression, header handling, and null treatment, referenced by stages and COPY statements.

The SQL command that loads staged files into a table or unloads table data to a stage, with options controlling file formats, error handling, file purging, and validation before a real load.

A Snowflake-managed service that loads new files automatically as they land in a stage, using cloud event notifications or a periodic file listing, with serverless compute billed per file loaded.

A low-latency ingestion API that writes rows directly into Snowflake tables from client applications using the Snowflake Ingest SDK, without staging files first.

A Snowflake object that captures change data, meaning inserts, updates, and deletes, on a source table so downstream jobs can process only the rows that changed since the last consumption.

A single SQL statement that inserts, updates, or deletes rows in a target table depending on whether each source row matches an existing key, commonly used to apply streamed change data.

A table function that turns arrays or objects stored inside a VARIANT column into individual rows, usually joined back to their parent row with a LATERAL clause.

A Snowflake clause that filters rows on the output of window functions inside the same query block, removing the need for an extra subquery or CTE.

Snowflake's procedural SQL dialect for stored procedures and anonymous blocks, providing variables, branching, loops, cursors, resultsets, and exception handling.

Snowflake's immutable, compressed storage unit holding a contiguous slice of table rows, with per-column metadata such as minimum and maximum values that the optimizer uses to skip data.

A table setting naming the columns Snowflake should physically co-locate data on, improving pruning for large tables that are frequently filtered or joined on those columns.

A measure of how well a table's micro-partitions are sorted by its clustering columns; lower depth means better pruning and typically faster scans.

A metadata-only copy of a database, schema, or table that shares the source's micro-partitions until either side is modified, making copies fast and cheap in storage.

A semi-structured Snowflake data type that holds JSON, Avro, ORC, Parquet, or XML documents in a single column, queried with dot and bracket notation.

A visual execution breakdown of a single query showing each operator's timing, rows and partitions processed, and memory or disk spilling, used to locate performance bottlenecks.

The optimizer's ability to skip micro-partitions whose metadata shows they cannot satisfy a query's filters, reducing the volume of data scanned.

Overflow of intermediate results to local disk and then remote storage when a warehouse runs out of memory, which slows queries and signals an undersized warehouse or an unfiltered join.

An account-wide cache that returns the stored output of a previously executed query when an identical query arrives and the underlying data has not changed, using no warehouse compute.

A warehouse setting that starts additional compute clusters automatically when queries begin to queue and shuts them down when demand falls, supporting high concurrency.

An optional service that maintains an auxiliary index structure so selective lookups, equality tests, and substring predicates on large tables avoid full table scans.

Snowflake's access model in which privileges are granted to roles, roles are granted to users or other roles, and a user's effective rights come from the roles active in the session.

A schema-level policy attached to a table or view that decides, at query time, which rows a given role is allowed to see.

A column-level policy that returns a substituted value, such as a hash or null, to unauthorized roles while authorized roles continue to see the original data.

Assigning a tag to sensitive columns and attaching a masking policy to that tag, so the protection is applied automatically everywhere the tag is set.

A view that hides its defining SQL and the objects it references from users, so they cannot inspect underlying tables or infer data they lack privileges for.

A retention window, up to 90 days for permanent objects, during which past table states can be queried with AT or BEFORE clauses or restored after a mistake.

A fixed seven-day period after Time Travel expires during which only Snowflake support can recover deleted or lost data; users cannot query it directly.

An account-level or user-level rule that permits or blocks session access based on the client's IP address range.

A Snowflake capability that exposes selected database objects to consumers through a share with no data copied, so consumers query the provider's live, read-only data.

A separate Snowflake account created and paid for by a provider so external consumers who do not own a Snowflake account can query shared data.

A private, invitation-only hub where members publish and subscribe to shares, typically used between an organization and its partners or internal business units.

A role defined inside a database that bundles privileges on that database's objects and can be granted to shares, accounts, or account-level roles.

Copying databases, shares, and account objects to another region or account so they stay available for disaster recovery or for readers located in that region.

A Snowflake object that runs a SQL statement or stored procedure on a schedule or when triggered, and can be chained to other tasks.

A set of tasks connected by dependencies so that one root task triggers a directed acyclic graph of child tasks in a defined order.

A table whose contents Snowflake keeps current automatically from a query you define, refreshing incrementally as base data changes instead of relying on scheduled jobs.

A task that runs on Snowflake-managed compute rather than a user warehouse, so no warehouse has to be sized or kept running for scheduled work.

An account-level integration with a cloud messaging service or email that lets tasks and pipes send alerts when a run fails.