About the Snowflake collector
Use this collector to directly harvest metadata from Snowflake data cloud including tabular objects, objects from Horizon (policies and tags), Snowflake data quality as well as functions and stored procedures. The collector allows you to harvest column statistics and table query counts. Additionally, the Snowflake collector can harvest metadata for Streamlit applications.
Important
The Snowflake collector can be run in the Cloud or on-premise using Docker or Jar files.
Note
The latest version of the Collector is 2.279. To view the release notes for this version and all previous versions, please go here.
Incremental runs for Snowflake Cloud collector
The Cloud version of the collector allows you to setup incremental collection. It harvests changed objects rather than performing a full scan from Snowflake. It identifies changes to objects in Snowflake and collects metadata at the schema level only for schemas containing changed Snowflake objects since the previous collector run. For details, see Setting incremental runs for Snowflake Cloud collector
Important
The collector uses metadata from the account_usage schema, which can have a latency of up to 3 hours before changes are reflected. If you don’t see recently modified or created objects in the collected metadata, it is likely due to this delay. We recommend waiting at least 3 hours after making changes in Snowflake before expecting them to appear in incremental collection results. For details, refer to the Snowflake documentation.
What is cataloged
The collector catalogs the following information.
Note
The collector harvests all versions of overloaded functions and stored procedures. Each version has its own title/name in the catalog, but a distinct identifier.
Object | Information cataloged |
|---|---|
Column | Name, Comments (saved as Description in data.world), Data Type, Is Nullable, Default Value, Key type (Primary, foreign) |
Table | Name, Comments (saved as Description in data.world), Primary key, Schema, External URL (Snowsight) Extended metadata: Is Iceberg, Is dynamic, Is transient, Created Date, Last Modified Date, Last Updated By (Snowflake LAST_DDL_BY), Owner (Snowflake role that owns table), Definition Altered Date, Table Size |
View | Name, Comments (saved as Description in data.world), Schema, External URL (Snowsight), Definition Altered Date, Last Modified By (Snowflake LAST_DDL_BY), Is Iceberg, Is dynamic, Is temporary, Created Date, Last Modified Date, Owner (Snowflake role that owns table), SQL definition |
Materialized View | Name, Comments (saved as Description in data.world), Schema, External URL (Snowsight), Definition Altered Date, Last Modified By (Snowflake LAST_DDL_BY), Is Iceberg, Is dynamic, Is temporary, Created Date, Last Modified Date, Owner (Snowflake role that owns table) |
Schema | Identifier, Name Extended metadata: Comments (saved as Description in data.world), Created date, Updated |
Database | Type, Name, Identifier, Server, Port, Environment, JDBC URL Extended metadata: Comments (saved as Description in data.world) |
User-defined function (this includes User-defined functions created in Snowpark) | Name Extended metadata: Function Definition, Comments (saved as Description in data.world), Created Date, Owner (Snowflake Role), Snowflake Function Type (Return Type), Function Signature, Function Language, Packages, Runtime version, Updated |
Stored Procedure (this includes Stored Procedures created in Snowpark) | Name Extended metadata: Function Definition, Owner (Snowflake Role), Comments (saved as Description in data.world), Return Type, Function Language, Packages, Created Date, Function signature, Snowflake Function Type, Runtime Version, Updated, Stored Procedure Type |
Cataloging Snowflake Tags and Policies
Following additional information is cataloged when you run the collector with the --policy-collection and --tag-collection parameters.
Object | Information cataloged |
|---|---|
Columns | Snowflake masking policies |
Snowflake Tags | Name, Allowed Values, Comments (saved as Description in data.world) |
Snowflake Tag Values | Name, Value |
Snowflake Masking Policies | Name, Comments (saved as Description in data.world), Date created, Snowflake tags, Owner, Type, Policy body, Schema, Database |
Snowflake Row access policies | Name, Comments (saved as Description in data.world), Date created, Snowflake tags, Owner, Type, Policy body, Schema, Database |
Cataloging Snowflake table query counts
Following additional information is cataloged when you run the collector with the --table-usage-collection and --table-usage-lookback-days parameters.
Object | Information cataloged |
|---|---|
Table | Query Count |
Cataloging Snowflake Data Metric Functions
You can harvest system data metric functions that capture data quality metrics. Following additional information is cataloged when you run the collector with the --data-metric-function-collection parameter.
Object | Information cataloged |
|---|---|
Data Metric Function |
|
Table |
|
Cataloging Snowflake Streamlit Applications
Following additional information is cataloged when you run the collector with the --streamlit-app-collection parameter.
Object | Information cataloged |
|---|---|
Snowflake Streamlit App |
|
Relationships between objects
By default, the harvested metadata includes catalog pages for the following resource types. Each catalog page has a relationship to the other related resource types. If the metadata presentation for this data source has been customized with the help of the data.world Solutions team, you may see other resource pages and relationships.
Resource page | Relationship |
|---|---|
Snowflake Row access policies | Tables |
Snowflake Masking Policies | Columns, Tag value |
Table | Columns, Tag value |
Snowflake Tags | Tag value |
Columns | Table, Tag Value, System Tag Value |
View | Schema that contains Views, Columns that are part of Views |
Materialized View | Schema that contains Materialized Views, Columns that are part of Materialized Views |
Schema | Database that contains Schema, Table that is part of Schema, View that is part of Schema, Materialized View that is part of Schema, Streamlit app that is part of schema |
Streamlit App | Schema that contains Streamlit App |
Database | Schema contained in Database |
Data Metric Function | Table or View analyzed by this Snowflake Data Metric Function, Table, Views, Tables, or Functions which the Function depends on |
Table | Data Metric Function that analyzes, Views, Tables, or Functions which the Table depends on |
View | Data Metric Function that analyzes View, Views, Tables, or Functions which the View depends on |
Lineage and dependencies for Snowflake
Lineage
The following lineage information is collected by the Snowflake collector. Note that any lineage for SQL Statements defined via variable statements are not supported
Important things to note:
Any lineage for SQL Statements defined via variable statements are not supported.
The collector traces these relationships from a View’s columns to ultimate source Table columns across SQL expressions and subqueries.
Object | Lineage available |
|---|---|
Column in View | The collector identifies the associated column in an upstream view or table:
|
User-defined function | A Function and:
A View and:
|
Stored Procedure | The collector identifies:
Note: The following stored procedures are not supported:
|
Dependencies
The Snowflake collector catalogs dependencies between tables, views, and functions using the Snowflake Account Usage view OBJECT_DEPENDENCIES. See the Snowflake documentation for more information about the OBJECT_DEPENDENCIES view.
Important details about Snowflake tags
The Snowflake collector harvests Snowflake tags allowing you to discover tags in their Snowflake instance.
When the collector is run with Collect Snowflake tag information (--tag-collection) option enabled, the collector harvests tags and tag-values (tag references) that meet one of the following criteria:
The tag is not linked to any object and resides in a database and schema specified in the collector run command.
The tag is linked with a database object, and the corresponding database is specified in the collector run command.
The tag is linked to a non-database object, and that object resides in a database and schema specified in the run command.
Important
Starting with version 2.268, the collector harvests system tags defined in SNOWFLAKE.CORE. Snowflake System tags are only applied at the column level.
If the collector is run with Collect all Snowflake tag information (--tag-collection-all) option, the collector does not filter based on whether the object of the tag is in one of the databases specified to be harvested.
Important details about Snowflake policies
The Snowflake Collector harvests Snowflake policies allowing you to discover policies in your Snowflake instance.
When the collector is run with Collect Snowflake policy information --policy-collection option, the collector harvests masking policies and row access policies that meet one of the following criteria:
The policy is not linked to any object and resides in a database and schema specified in the collector run command.
The policy is associated with a database object, and the corresponding database is specified in the collector run command.
The policy is linked to a non-database object, and that object resides in a database and schema specified in the run command.
If the collector is run with the Collect all Snowflake policy information (--policy-collection-all) option, the collector does not filter based on whether the object of the policy is in one of the databases specified to be harvested.
Sample view of Snowflake tag and policies in data.world
In the following example you can see how a Snowflake Tag-Based Masking Policy applied to sensitive data columns: routing numbers, bank name, and bank account number. In this view, you can also see the associated tag (Classification:confidential), and technical details about the policy, like the Policy Body which explains how the Policy works.
Feature | Details |
|---|---|
Snowflake object tags | Tags enable data stewards to track sensitive data for compliance, discovery, protection, and resource usage use cases through either a centralized or decentralized data governance management approach. For details, see Snowflake documentation. |
Snowflake masking policies | Masking policies define what data is masked and not shown to users allowing authorized users to access sensitive data at query runtime. For details, see the Snowflake documentation. |
Snowflake tag based masking policies | When a masking policy is applied to a Snowflake tag, the columns with the same tag are automatically protected based on the conditions set in the masking policy. For details, see the Snowflake documentation. |
Snowflake row access policies | Access policies are applied to rows of data, that is, who can see the rows of data, and masking policies define what data is masked and not shown to users. For details about this see the Snowflake documentation. |
Table query count | For each table, the Snowflake collector computes percentile popular and a query count based on the user access history table in Snowflake. This information can then be used to assign popularity ranking to Snowflake tables in data.world. For details about this see the Snowflake documentation. |
When this data is harvested and is displayed in data.world, you see the following information in your catalog: Snowflake masking policies, Snowflake row access policies, Snowflake tag values, and Snowflake tags.
When you browse to a table harvested from Snowflake, you will notice that the appropriate policies are applied to the tables. In the following example, the policy and tag values are applied to the Order table.
Users can click through to the tag values and policies to see their details.
Note
Note that users can also use the Search feature in the application to locate all these resources easily.
Authentication supported
The collector supports the following authentication method:
Username and Key-pair authentication. For details, see the Snowflake documentation.
Warning
Snowflake is transitioning away from username & password authentication soon. If you used this method of authentication, we advise that you update to key-pair authentication. For details, see this field notice.