April 2022¶

New Account Usage View: COMPLETE_TASK_GRAPHS — Preview¶

With this release, a new view, COMPLETE_TASK_GRAPHS, has been added in preview to the Account Usage views (in the SNOWFLAKE shared database) to help you analyze the history of completed task graph runs. A graph is defined as a single scheduled task or a simple tree of tasks composed of a scheduled root task and one or more child tasks (i.e. tasks that have a defined predecessor task).

The output of this new view is identical to that of the new COMPLETE_TASK_GRAPHS table function in the Information Schema.

New Information Schema Table Functions: COMPLETE_TASK_GRAPHS and CURRENT_TASK_GRAPHS — Preview¶

With this release, two new table functions that return the history of task graph runs have been added in preview to the Snowflake Information Schema:

These table functions return the history of task runs for up to 14 days.

Information Schema Table Functions: New ERROR_ONLY Argument for TASK_HISTORY — Preview¶

With this release, you can filter the TASK_HISTORY function output and retrieve the history of failed or canceled task runs within a specified date range. Set the new ERROR_ONLY=>TRUE argument when querying the function to filter the results.

This functionality is currently in preview.

Resource Monitors: New Parameter Enabling Email Notifications for Non-administrators¶

With this release, Snowflake adds support to enable email notifications for resource monitors for non-administrator users (users who do not have the ACCOUNTADMIN role).

Support for this feature is implemented through a new parameter, NOTIFY_USERS, for resource monitors that can be set when the resource monitor is created (using CREATE RESOURCE MONITOR), or later (using ALTER RESOURCE MONITOR). You can add up to five non-administrator users to a resource monitor to receive email notifications.

There are some limitations for non-administrator users:

• Non-administrator users are notified by email but cannot see notifications in the web interface.

• Non-administrator users cannot create resource monitors.

• Non-administrator users cannot be notified of account level resource monitors.

• Non-administrator users cannot assign other users to be notified.

See Resource monitor notifications for more details.

New Account Parameter Sets Minimum Data Retention Time for Permanent Tables¶

With this release, Snowflake is pleased to announce a new parameter to set a minimum number of days for which Snowflake retains historical data for performing Time Travel operations. The MIN_DATA_RETENTION_TIME_IN_DAYS parameter can be set by account administrators at the account level and can be used to enforce a minimum data retention period on all permanent tables in the account.

Setting this parameter at the account level does not alter the DATA_RETENTION_TIME_IN_DAYS parameter value that is explicitly set on databases, schemas, or tables. It may, however, change the effective data retention period for an object. When this parameter is set on an account, the data retention period for an object is determined by MAX(DATA_RETENTION_TIME_IN_DAYS, MIN_DATA_RETENTION_TIME_IN_DAYS).

For more details, see Understanding & Using Time Travel.

Process Unstructured Data Using Java UDFs — Preview¶

We are pleased to announce an open preview of unstructured data extraction and processing using Java user-defined functions (UDFs) or tabular Java UDFs (UDTFs).

Unstructured data is data that lacks a predefined structure. It is often textual, such as open-ended survey responses and social media conversations, but can also be non-textual, including images, video, and audio. Java UDFs enable you to perform custom operations using the Java programming language to manipulate data and return either scalar or tabular results. Call your custom UDFs and compiled code to extract text, process images, and perform other operations on unstructured data for analysis.

You can either include the Java code inline in the function definition or package the code in a JAR file and copy the file to an internal or external stage. Call the UDF with the input as a scoped URL, file URL, or the string file path for one or more files located in an internal or external stage. The new SnowflakeFile class enables you to easily pass additional file attributes when calling the UDF, such as file size, to filter the results.

Previously, Snowflake customers were limited to processing unstructured files using external functions and remote API services.

This feature is currently in preview. Preview features are intended for evaluation and testing purposes, and are not recommended for use in production.

Access History Write Operations — General Availability¶

With this release, Snowflake is pleased to announce the general availability of write operations in the ACCESS_HISTORY view (Account Usage). Write operations are recorded in the objects_modified column of the ACCESS_HISTORY view.

This update provides a more unified picture of what data was accessed, when the data access took place, and how the accessed data moved from the data source object to the data target object.

For a complete list of supported write operations (e.g. CTAS, COPY … INTO TABLE), see the usage notes section in the ACCESS_HISTORY View.

Object Dependencies — General Availability¶

With this release, Snowflake is pleased to announce the general availability of object dependencies in the OBJECT_DEPENDENCIES view (Account Usage).

This update provides data stewards and data engineers a unified picture of the relationships between referencing objects and referenced objects. For example, when a table owner plans to modify a column, querying the OBJECT_DEPENDENCIES view based on the table name returns all of the objects (e.g. views) that will be affected by the modification.

For more details, see Object Dependencies.

Object Tagging: Allowed Values — General Availability¶

With this release, Snowflake is pleased to announce the general availability of allowed values for object tagging. This update allows tag administrators to specify the tag string values that can be used when a tag key is set on a supported object.

For example, the tag cost_center can now be configured to have the allowed string values of marketing and engineering. When this tag is set on a supported object, such as a warehouse or share, the string values that can be used are marketing and engineering only. Tag administrators can optionally add more allowed string values or drop existing allowed string values as needed. The maximum number of allowed values a tag can have is 50.

For details, see Object Tagging.

Access History: Support Added for TRUNCATE TABLE¶

With this release, Snowflake adds support to the ACCESS_HISTORY view (Account Usage) to record operations from the TRUNCATE TABLE command.

The objects_modified column in the ACCESS_HISTORY review records the table that was specified in the TRUNCATE TABLE command and all columns in the specified table.

Snowflake Documentation Available in Korean¶

Snowflake is committed to providing the highest levels of service to our global users. To support this commitment, we are pleased to announce the availability of Snowflake Documentation in Korean at https://docs.snowflake.com/ko/.

In line with the other documentation translated from English, the new Korean translated documentation will be updated weekly as we release new features, enhancements, and fixes, with a period of time, typically 5-7 business days (and sometimes longer), between the updates in English and the translated documentation.