Question

I have the following query:

-- CTE to remove outliers, e.g. remove the fastest and slowest results
;WITH MinMaxCTE AS 
(
  SELECT ServerName, CONVERT(VARCHAR(10), UpdatedOn, 101) AS [Date], Version, 
    MIN(JaguarStartupTime) AS MinStartTime, MAX(JaguarStartupTime) AS MaxStartTime
  FROM dbo.MachineConfiguration (NOLOCK)
  WHERE DomainLogin NOT IN (SELECT DomainLogin FROM dbo.SupportGroup)
  GROUP BY ServerName, CONVERT(VARCHAR(10), UpdatedOn, 101), Version
)

SELECT  AVG(mc.JaguarStartupTime) AS AverageTime
    , COUNT(*) AS NumEntries
    , mc.Version
FROM    #Eligible mc (NOLOCK)
JOIN MinMaxCTE cte ON mc.ServerName = cte.ServerName 
  AND CONVERT(VARCHAR(10), mc.UpdatedOn, 101) = cte.[Date] 
  AND mc.Version = cte.Version 
  AND mc.JaguarStartupTime <> cte.MinStartTime 
  AND mc.JaguarStartupTime <> cte.MaxStartTime
GROUP BY mc.Version
ORDER BY Version DESC, AVG(mc.JaguarStartupTime) ASC

The definition of the #Eligible temp table is

create table #Eligible (
    Version nvarchar(50), JaguarStartupTime int, 
    ServerName nvarchar(50), UpdatedOn datetime )

No matter what condition or aggregation I comment out, I always get the following error: Arithmetic overflow error converting expression to data type int.

Where can I go from here? How do I debug this further?

EDIT: Sample data

Version JaguarStartupTime    ServerName     UpdatedOn
6.4.6.082      16040         NewOrleansLA   2012-08-08 12:34:12.330
6.5.1.012      40390         BatonRougeLA   2012-08-08 18:33:17.440
6.5.1.012      48379         HonoluluHI     2012-08-09 04:42:50.453

Answer 1

Have you tried casting your jaguarstartup times to a bigint in your avg aggregate like so…

AVG(CAST(mc.JaguarStartupTime AS BIGINT))

This should sort out the arithmetic overflow.

To calculate the mean average the server needs to be able to sum all of the ints first, so the datatype you are averaging on needs to able to store the sum of those values, even if the returned answer is within the range of of an int