Review the current state of the table and index statistics to determine if they need to be updated.
That turns out to make it quite different (because only the leading column in a statistic gets information in the histogram).
Looking at the properties of the scan, it figured out that Fake Birth Date Stamp would be there (because of the clustering key), and decided to scan this nonclustered index and output just that column: But … We didn’t allocate enough memory for our sort and had a little spill in tempdb. I don’t have a non-clustered index on this column for SQL Server to scan, but I was surprised that it wanted to re-compute every single row for it, because I mark this column as ‘persisted’ (I double-checked with a query). We’ve got another nonclustered index, and it has two key columns.
If you’re still reading, here’s where things get crazier than I expected. SQL Server uses the following query to gather data to update my column statistic… But this time, SQL Server really didn’t want to scan that clustered index again (we just did it, after all), so it decided to recompute every. Those two columns are in the auto-generated index statistic.
Designed to launch directly within SQL Server Management Studio (SQL 2012 and above) for optimal usability and efficient workflow.
Connect to database instances hosted in the cloud, such as SQL Server on Azure Virtual Machine (VM), SQL Server on Amazon Elastic Compute Cloud (EC2), and SQL Server on Amazon Relational Database Service (RDS).