Memory allocated for the execution of a query is based on the assumed row width. The assumed row width is not based on reality - it's based on the declared...

When moving to a new cluster, you will have to transfer the databases, logins, jobs, AGs, etc. to the new cluster and instances. If you plan to use Automatic Seeding...

Look at the Remote Object name in the Remote Scan. It's an object called 'IndexAnalysis'. It's not an actual table, but a chunk of memory. 

You're using a DMF...

If you partition the table on the @@SPID value, then you don't need any of that lock-checking mess; everything gets its own partition. (You'll need to align your indexes for...

When a sql:variable() value is substituted in the query, it is treated as a literal string, wrapped in double-quotes, instead of as an XML node test. 
Literal strings used by...

Adding...
If you are comparing node values to a variable, use sql:variable("@VariableName") in the XPath.
If you are comparing to another column in the query, use sql:column("ColumnName")

[code...

Leave the current DataMart online while you prepare the new one:
- Create a new database on the DataMart
 - ETL-load that database from your DWH while the current database...

Following up, I rebuilt the table with row-level compression and ran the tests again. The results were the same. The in-memory representation of the data is based not on the...

The counter argument is that while data stored on disk and persisted in the data cache uses actual size (if you use Unicode compression when nvarchar(columns) that hold ASCII...

When you deleted the column, the contents already on disk were not deleted. New rows will go in without the column, but the old data stays on disk until you...