>>>
http://social.technet.microsoft.com/wiki/contents/articles/25654.t-sql-find-the-peak-hour.aspx>>
>>Snip : "a simple solution"
>>
>>If you call that simple what do you call 'complicated' :-}
>>
>>I raised this with Kevin : given the option of a simple CLR method (a small piece of code with more flexible options over the number of periods in the table and the span of the 'peak' period why on earth would you use this T-SQL ?
>
>Did you compare CLR solution with T-SQL in terms of speed?
No. But I'd make a small bet that a CLR solution would be as fast if not faster.
Data access is minimal (you're only looking at one row) . The rest is essentially just processing a array.
Snip (
http://msdn.microsoft.com/en-us/library/ms131075(v=sql.100).aspx) :
"When Transact-SQL cursors must traverse data that is more easily expressed as an array, managed code can be used with significant performance gains."
If I find some time I'll test it.....
>Yes, for me that solution is simple, but the CLR solution would be more complex. I personally would only employ CLR solution when there is no simple T-SQL solution (or performant).
Our definitions of 'simple' differ. If it was simple Don wouldn't have raised the question in the first place (and, presumably, you would not have thought it worthwhile to post a solution on TechNet)