For machine learning, or in general any data analysis task, we need data. That’s not enough because data has to be in the right shape and data wrangling is usually tedious/time consuming/character shaping experience. QueryStore is one of the most important tools in context of Performance Intelligence (synonym for gaining performance insights and making automated decisions) because it provides consistent and unified view on queries’ performance by forming multi-feature timeseries for every query-plan-interval:
DR in Azure Sql DW Key component of Disaster Recovery Plan for Azure Sql DataWarehouse is a Geo-Backup Policy that must be enabled and working. Geo-Backup Azure Sql Datawarehouse uses simple recovery model and special storage level backups (consistent across multiple nodes) are taken every 4-8h. Backup history is stored in sys.pdw_loader_backup_runs DMV. Minimum once a day backup is copied to paired region so RPO is 24h assuming that Geo-Backup actually works.
Workload Generator tSql gist to generate all possible queries based on foreign key constraints.