An interesting observation on Yukon SQL, sometimes an update statement’s performance is quite slow.  For example, let’s present a SQL table with 3+ million rows with a LogAuditID column , int data type, value of 0 for all columns; the table contains a nonclustered index not on the LogAuditID column.  If I wanted to update all of the rows within this table with a value of 1, running the following statement:

 

 

is quite slow. On both single proc and multi-proc boxes, this one simple statement can take a long time to run.

 

But, if you re-write the same statement (taking advantage of the fact that LogAuditID originally is 0), you can run the following statement:

 

 

This statement will take ~4:15 (mm:ss) on a single proc box (faster on a multi-proc box).

 

Even shorter still, you can specifically indicate to the SQL Server engine to not use the existing indices when performing the audit:

 

This particular statement runs in ~3:30 (mm:ss).