In January we implemented some improvements that added stability and increased reliability for SQL Data Warehouse.
Loading
- We fixed an issue where Polybase queries against external tables would sometimes fail with error messages indicating JVM out of memory. This could occur when reading flat files with a large number of values of date/time data type.
- We fixed an issue where bulk load operations could run for longer than expected or hang and eventually fail. This could occur specifically when concurrent loads went to the same destination table which is a common pattern when using Azure Data Factory to load data into Azure SQL Data Warehouse.
Monitoring
We fixed an issue with the sys.dm_pdw_exec_requests dynamic management view that was causing additional unexpected rows to appear that were not associated with the session.
Stability
We fixed an issue that intermittently caused increased unavailability for data warehouse instances during upgrades, initiated by the Azure SQL DW service team, and not related to any customer action. The service team monitors the health of all data warehouses worldwide. We detect and proactively resolve service health problems, but this fix prevents specific issues from even occurring.
Tell us what you think
It is thanks to you we’re able to make our product better and more relevant for your use cases; so keep the feedback coming, we’re listening! Please reach out on feedback.azure.com with feature requests or on StackOverflow for help or would like to see more examples.