Google has released a new preview feature that enables metadata caching for SQL translation in BigQuery, significantly reducing latency for subsequent translation requests.
Users can now enhance translation quality by running interactive queries with additional translation configurations. This is accomplished by specifying configuration YAML files stored in a Cloud Storage folder. These configurations may include:
- SQL object metadata
- Object mapping information from source databases
- DDL information
- Schemas from source databases
How to Specify Translation Configurations
To specify translation configurations from source files:
- Navigate to the query editor and click More > Translation settings
- Enter the path to your translation configuration files in the Translation Configuration Source Location field
- Click Save
The BigQuery interactive SQL translator supports metadata ZIP files containing translation metadata and object name mapping. Users seeking information on uploading files to Cloud Storage can refer to the "Upload objects from a filesystem" documentation.
Enabling Metadata Caching
The new metadata caching feature provides significant performance improvements:
- In the query editor, select More > Translation settings
- Check the Enable metadata caching checkbox
- Click Save
For jobs with large metadata files, this process substantially reduces translation latency for subsequent requests. Cached metadata remains active for up to 7 days.
This feature is currently in preview. Users who wish to request support or provide feedback for this feature can contact bq-edw-migration-support@google.com.