Re: Another interesting SQL result

[HankHeath@xxxxxxx]

Sounds like a difference from casting of the data types. In English, 
overhead.

In a message dated 8/12/04 3:19:34 PM Eastern Daylight Time, 
dbale@xxxxxxxxxxxxx writes:


> FWIW, don't know if this will surprise SQL experts at all or not, but...
> 
> Referring back to the SQL with three joins I was developing, you may
> remember that my Where clause looked like:
> 
>    Where     T01.CONOCV = 01
>          and T01.ICDTCV between 1040701 and 1040731
>          and T03.SCCDSS = '2'
> 
> Since the hard-coded date wasn't suitable for the production environment, I
> changed this to:
> 
>    Where     T01.CONOCV = 01
>          and T01.ICDTCV between :BeginDate and :EndDate
>          and T03.SCCDSS = '2'
> 
> T01.ICDTCV is defined as 7,0 packed
> BeginDate and EndDate are defined in the RPGLE program D-specs as 7,0 and
> are set at *INZSR based on *entry parameters.
> 
> I ran the two versions back-to-back for about 20 iterations and, clearly,
> the version that used variables :BeginDate and :EndDate completed in half
> the time as the one using the hardcoded dates.  (I collected timestamps at
> *INZSR and at end-of-program, then calculated the duration.)  Some
> statisticals (all durations in seconds):
> 
>   Using:   Variable  Hard-coded
> 
>   Average:   1.436      2.843
>   Maximum:   1.652      3.141
>   Minimum:   1.262      2.647
> Std. Dev.:   0.121      0.153
> 
> As always, YMMV.  But very interesting any way you look at it.