SQL Server MVP Deep Dives- P17: Each year Microsoft invites all the MVPs from every technology and country to Redmond for an MVP Summit—all top secret—“don’t tweet what you see!” During the MVP Summit, each product team holds a series of presentations where they explain their technologies, share their vision, and listen to some honest feedback. | 594 Chapter 47 How to use Dynamic Management Views queries that are relatively inexpensive for individual executions but are called very frequently which makes them expensive in aggregate or you may have individual queries that are more expensive CPU-wise but are not called as often. Looking at total worker time is a reliable method for finding the most expensive queries from an overall CPU perspective. Another similar DMV query shown in listing 6 sorts by average worker time. This will let you find expensive CPU queries that may be easier to improve at the database level with standard database tuning techniques. Listing 6 Finding expensive stored procedures sorted by average worker time -- Get Top 20 executed SP s ordered by Avg worker time CPU pressure SELECT TOP 20 AS SP Name AS AvgWorkerTime AS TotalWorkerTime AS Execution Count ISNULL DATEDIFF Second GetDate 0 AS Calls Second ISNULL 0 AS AvgElapsedTime DATEDIFF Minute GetDate AS Age in Cache FROM AS qs CROSS APPLY AS qt WHERE db_id -- Filter by current database ORDER BY DESC Finding I O pressure in SQL Server Most large-scale SQL Server 2005 2008 deployments sooner or later run into I O bottlenecks. This happens for several reasons. First systems engineers often just think about CPU and RAM when sizing big database servers and neglect the I O subsystem. Second many DBAs are unable to completely tune the SQL Server workload to minimize excessive I O requirements. Finally there are often budgetary issues that prevent the acquisition of enough I O capacity to support a large workload. Whatever your situation it helps if you know how recognize and measure signs of I O pressure on SQL .
