- Fixed Cartesian product bug introduced in 2.0
- Re-worked PBI dashboard to remove custom data retrieval functions to utilise Direct Queries and benefit from Query Folding. It should now be also possible to deploy dashboard to the PBI Service and schedule it.
- As part of this work, PBI schema was simplified and GUIDs are no longer required. I have therefore migrated keys from GUIDs to numerical values which saves the storage utilisation.
- Further schema improvements reducing storage requirements by ~40-60%
- Started adding views that serve pre-calculated data for reporting which can be fed into any dashboarding solution.
- To improve reporting times, delta calculations have been pushed to the collector so all data is available in tables and no expensive calculation happen during reporting.
- Some code cleansing and aligning with standard
- Added config procedure to enable index compression, on top of the table compression introduced in 2.0
- Removed unused indexes
- Tweaked clustered keys for better performance
- Normalised wait types
- Removed calculations that are not used
- Improved wait stats collection by the mean of capturing snapshots in stage tables to correctly calculate deltas between snapshots where the first snapshot had 0 wait time.
- Started moving most of the business logic from PBI to the database tier
- Re-worked SSIS to address schema changes
- Improved central repository SSIS import as in some cases, it would try and collect data from remotes without having valid header which would cause FK violation.
- Post deployment configuration is now applied automatically by a mean of triggering the relevant agent job
Due to GUID migration it is not possible to directly upgrade from version 2.0
Documentation can be found at https://docs.sqlwatch.io
As always, please feedback any issues via GitHub Bug or on our Slack channel