Based on getting Query Execution Statistics using this extremely useful piece of SQL obtained from this post Most Executed Stored Procedure - Stack Overflow:
SELECT TOP 100
qt.TEXT AS 'SP Name',
SUBSTRING(qt.text, qs.statement_start_offset/2, CASE WHEN (qs.statement_end_offset = -1) THEN LEN(qt.text) ELSE (qs.statement_end_offset - qs.statement_start_offset)/2 END) AS actual_query,
qs.execution_count AS 'Execution Count',
qs.total_worker_time/qs.execution_count AS 'AvgWorkerTime',
qs.total_worker_time AS 'TotalWorkerTime',
qs.total_physical_reads AS 'PhysicalReads',
qs.creation_time 'CreationTime',
qs.execution_count/DATEDIFF(Second, qs.creation_time, GETDATE()) AS 'Calls/Second'
FROM sys.dm_exec_query_stats AS qs
CROSS APPLY sys.dm_exec_sql_text(qs.sql_handle) AS qt
WHERE qt.dbid = (SELECT dbid
FROM sys.sysdatabases
WHERE name = 'BSP')
ORDER BY qs.total_worker_time/qs.execution_count DESC
How would I completely clear out these execution statistics and start from scratch?
This would be particularly useful as development bugs and testing have caused routines to be called an usually large number of times thus invaliding the true usage levels.
DBCC FREEPROCCACHE
DBCC DROPCLEANBUFFERS