title | description | keywords | services | ms.service | ms.topic | ms.subservice | author | ms.author | ms.reviewer | ms.date |
---|---|---|---|---|---|---|---|---|---|---|
Release notes for Azure SQL Edge |
Release notes detailing what's new or what has changed in the Azure SQL Edge images. |
release notes SQL Edge |
sql-edge |
sql-edge |
conceptual |
kendalvandyke |
kendalv |
jroth |
11/24/2020 |
This article describes what's new and what has changed with every new build of Azure SQL Edge.
SQL engine build 15.0.2000.1562
- Security bug fixes
SQL engine build 15.0.2000.1559
- PREDICT support for ONNX
- Improvements in handling of null data in PREDICT for ONNX
SQL engine build 15.0.2000.1557
- Upgrade ONNX runtime to 1.5.3
- Update to Microsoft.SqlServer.DACFx version 150.5084.2
- Miscellaneous bug fixes
SQL engine build 15.0.2000.1557
-
T-SQL streaming
- Fix in ownership and permissions for streaming objects
- Logging improvements with log rotation and log prefixing
- Azure Stream Analytics: Logging improvements, improve error code/ error messages in adapters
-
ONNX
- Bug fixes for parallel query scenario and model cleanup failures
- Upgraded ONNX runtime to 1.5.1
SQL engine build 15.0.2000.1553
- Allow Date_Bucket expressions defined in computed columns.
- Retention policy fix for dropping a table that has a retention policy enabled with an infinite timeout
- DacFx deployment support for streaming features and retention-policy features
- DacFx deployment fix to enable deployment from a nested folder in a SAS URL
- PREDICT fix to support long column names in error messages
SQL engine build 15.0.2000.1552
- Container images based on Ubuntu 18.04
- Support for
IGNORE NULL
andRESPECT NULL
syntax withLAST_VALUE()
andFIRST_VALUE()
functions - Reliability improvements for PREDICT with ONNX
- Support for cleanup based on the data-retention policy:
- Ring buffer support for a retention cleanup task for troubleshooting
- New feature support:
- Fast recovery
- Automatic tuning of queries
- Parallel-execution scenarios
- Power-saving improvements for low-power mode
- Streaming new-feature support:
- Snapshot windows: A new window type allows you to group by events that arrive at the same time.
- TopOne and CollectTop can be enabled as analytic functions. You can return records ordered by the column of your choice. They don't need to be part of a window.
- Improvements to MATCH_RECOGNIZE.
- Additional error messages and details for troubleshooting T-SQL streaming operations
- Improvements to preserve battery life in idle mode
- T-SQL streaming engine fixes:
- Cleanup for stopped streaming jobs
- Fixes for localization
- Improved Unicode handling
- Improved debugging for SQL Edge T-SQL streaming, allowing users to query job-failure errors from get_streaming_job
- Cleanup based on data-retention policy:
- Fixes for retention-policy creation and cleanup scenarios
- Fixes for background timer tasks to improve power savings for low-power mode
- The Date_Bucket T-SQL function can't be used in a computed column.
SQL engine build 15.0.2000.1549
- Support for custom origins in the Date_Bucket() function
- Support for BACPAC files as part of SQL deployment
- Support for cleanup based on the data-retention policy:
- DDL support for enabling the retention policy
- Cleanup for stored procedures and the background cleanup task
- Extended events to monitor cleanup tasks
- Additional error messages and details for troubleshooting T-SQL streaming operations
- Improvements to preserve battery life in idle mode
- T-SQL streaming engine:
- Fix for stuck watermark in the substreamed hopping window
- Fix for framework exception handling to make sure it's collected as a user-actionable error
SQL engine build 15.0.2000.1546
- Support for non root containers
- Support for the usage and diagnostic data collection
- T-SQL streaming updates:
- Support for Unicode characters for stream object names
- T-SQL streaming updates:
- Process cleanup improvements
- Logging and diagnostics improvements
- Performance improvement for data ingestion
SQL engine build 15.0.2000.1545
- Allowed the PREDICT-with-ONNX models to handle a CPUID issue in ARM
- Improved handling of the failure path when T-SQL streaming starts
- Corrected value of the watermark delay in job metrics when there's no data.
- Fix for an issue with the output adapter when the adapter has a variable schema between batches
SQL engine build 15.0.2000.1401
- Product name updated to Azure SQL Edge
- Date_Bucket function:
- Support for Date, Time, and DateTime types
- PREDICT with ONNX:
- ONNX requirement for the RUNTIME parameter
- T-SQL streaming support (limited preview)
- Issue: Potential failures with applying DACPAC on startup because of a timing issue.
- Workaround: Restart SQL Server. Otherwise, the container will retry applying the DACPAC.
You can request support on the support page. Select the following fields:
- Issue type: Technical
- Service: IoT Edge
- Problem type: My problem relates to an IoT Edge module
- Problem subtype: Azure SQL Edge
:::image type="content" source="media/get-support/support-ticket.png" alt-text="Screenshot showing a sample support ticket.":::
SQL engine build 15.0.2000.1331
- Date_Bucket function:
- Support for the DateTimeOffset type
- PREDICT with ONNX models:
- NVARCHAR support
SQL engine build 15.0.2000.1247
-
PREDICT with ONNX models:
- VARCHAR support
- Migration to ONNX runtime version 1.0
-
The following features are enabled:
- CDC support
- History table with compression
- A higher-scale factor for log read-ahead
- Batch mode ES filter pushdown
- Read-ahead optimizations
SQL engine build 15.0.2000.1147
- Azure IoT portal deployment:
- Support for deploying AMD64 and ARM images
- Support for streaming job creation
- DACPAC deployment
- PREDICT with ONNX models:
- Numeric type support
- The following features are enabled:
- Pushdown aggregate to column store scan
- Merry-go-round scans
- Footprint and memory-consumption reduction work