David Betten wrote:
>I just found out you have a PMR open so no need to send anything to the DFSORT
>Hotline. The same people will be investigating this through the PMR process.
Good. I would be really interested in the result of that PMR.
Neil E. Ervin wrote:
> One calls the COPY function using an INCLUDE which eliminates all the records.
> The second calls the SORT function with the same INCLUDE eliminating all the
> records so the final SORT is never executed.
Are they using the same input or do these jobs run in sequence?
> The OPTION COPY does the same number of EXCPS but uses the CPU and does
> larger I/Os (3-4 times as large) as the SORT. It also runs 4 times as fast
> as the SORT.
Hmmm, very interesting. Unless I missed something obvious, but I see COPY just
as is: Copy records based on selection and NO sorting is involved, while the
second does SORT which involves CPU + I/O overhead.
So, something is indeed wrong. Good catch Neil!
> The problem with slow SORT appears to be related to the small number of
> buffers used for the SORT function versus the OPTION COPY
As David mentioned, we need more info about the input datasets.
> OPTION COPY
INCLUDE COND=(26,6,CH,EQ,C'NEVERF')
> And
> INCLUDE COND=(26,6,CH,EQ,C'NEVERF')
SORT FIELDS=(22,4,CH,A)
I hope you can get it SORTed out! ;-)
Groete / Greetings
Elardus Engelbrecht
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
