This article provides a brief overview of the new features, enhancements, and other important changes introduced in this scheduled release of Snowflake. For more detailed information, see the Snowflake documentation.
Important: These changes may take 24 hours or more (after the completion of the release) to become available.
If you have any questions, please email email@example.com.
Virtual Warehouses: Billing Continuation
Previously, each time a warehouse was resumed (and servers were provisioned), the warehouse was charged for 1 hour of server usage, regardless of how much time had passed since the last time the warehouse was charged. As a result, if the same warehouse was resumed multiple times within the same hour, credits were charged each time. This caused some of our users to spend their time and efforts managing their warehouses to minimize credit charges. To address this, we introduced Warehouse Billing Continuation (WBC).
With WBC, we now track the last time each individual server in a warehouse was charged. If a warehouse was suspended and resumed within 60 minutes of the last charge, we do not charge again for the servers in the warehouse. The charge is continued from last time as if the warehouse had never been suspended. This eliminates additional credit charges, thereby reducing the need for strictly monitoring and controlling when warehouses (single-cluster or multi-cluster) are suspended/resumed and resized.
This enhancement doesn’t require any changes on your part. WBC has been running quietly in the background for some time now so you may have already seen some credit savings for warehouses that you suspend and resume frequently. However, you might want to revisit some of your warehouse practices and settings, particularly auto-suspend, which can now be set to short intervals (e.g. 5 or 10 minutes) without causing additional charges.
Note: When determining your practices to follow for suspending and resuming warehouses, there are some caveats to consider regarding server provisioning time and caching. For more details, see the Snowflake documentation.
Leading Comments Removed During Query Execution
As announced previously, in this release, the behavior of leading comments has changed as follows:
- Previous behavior: Snowflake included leading comments in the SQL text when processing queries. The comments were prepended to the query in the SQL Text column on the History page and in the QUERY_HISTORY function output.
- New behavior: To allow private notes and query instructions in leading comments without concern for exposing private information, Snowflake now strips any leading comments when executing a query, i.e. they are no longer included in the SQL text.
To intentionally add identifying information to a query, we suggest nesting the relevant comments within the query instead. Snowflake will continue to include nested comments in the SQL text when processing queries.
Alternatively, use the QUERY_TAG session parameter to associate a comment (i.e. "tag") with all queries executed after the parameter is set in the current session.
Note: If you encounter issues due to this change in behavior and need to switch back to the previous behavior, please contact Snowflake Support.
Table Clustering Keys: Support for Expressions
Clustering keys, currently in open preview, enable explicitly specifying one or more keys in a table to cluster on, which helps maintain optimal data clustering in the table, particularly for very large tables.
With this release, Snowflake adds support for specifying expressions (on table columns) as clustering keys. This enables more fine-grained control over the data in the columns used for clustering.
TIMESTAMP_TZ: Fix for Inconsistent Time Zone Offset Behavior
This release includes a fix for an issue that could result in incorrect TIMESTAMP_TZ calculations.
- In previous releases, calculations based on a date or time literal caused TIMESTAMP_TZ values to behave as though the time zone information were preserved. The values could be adjusted based on DST start or end times.
This behavior was inconsistent with other cases where TIMESTAMP_TZ values were used.
- In the current release, all TIMESTAMP_TZ calculations retain the time zone offset for the original value based on the TIMEZONE session parameter.
Client Package Validation: New Public Key
As part of the regular rotation of the GPG security key that we provide for validating our client packages, Snowflake has released a new public key for use with the following signed client packages (scheduled to be released this week):
- SnowSQL 1.1.26 and higher (on Linux)
- Spark Connector 2.1.3 and higher
- JDBC Driver 3.0.15 and higher
- ODBC Driver 2.12.97 and higher (on Linux)
For detailed instructions, see the Snowflake documentation for the relevant client.
Important: If you have automated the installation and configuration of new client versions in your local environment, to ensure your uninterrupted usage of these clients, make sure to import the new key before the corresponding client version is released. If you have any questions, please contact Snowflake Support.