IBM Mainframe Forum

by **sharma_deepu** » Thu Feb 03, 2011 3:30 pm

I have 2 datasets that contain duplicate and non-duplicate elements so I need two output datasets: all duplicates in one and all non-duplicates in the other.

by **skolusu** » Thu Feb 03, 2011 10:42 pm

sharma_deepu,

Use the following DFSORT JCL which will give you the desired results

Select all

//STEP0100 EXEC PGM=ICETOOL
//TOOLMSG DD SYSOUT=*
//DFSMSG DD SYSOUT=*
//IN DD *
A
A
B
C
C
C
//DUPS DD SYSOUT=*
//UNQ DD SYSOUT=*
//TOOLIN DD *
SELECT FROM(IN) TO(UNQ) DISCARD(DUPS) NODUPS ON(1,1,CH)
//*

by **hailashwin** » Sat Feb 05, 2011 10:21 am

Hi Kolusu,
It makes me wonder why you had gone in for the ICETOOL step instead of the traditional SORTXSUM (atleast for me

) for this particular scenario. I am asking this because, from the time I have started to read abt Sort and appreciate it...I have been hearing from ppl that a ICETOOL step is performance intensive than a normal sort. But I havent had the chance to do the compare myself.
Please correct my understanding..

Thanks,
Ashwin.

by **dick scherrer** » Sat Feb 05, 2011 10:48 am

Hello,

I have been hearing from ppl that a ICETOOL step is performance intensive than a normal sort

One must use care in choosing who to listen to. . . Many people have opinions that have no basis in reality - but they sometimes appear quite knowledgable. . .

The biggest causes of "intensive performance" is the number of records to be processed and how many passes over the data are used to arrive at the solution. Some of the worst performing "solutions" to be seen are when the developer was determined to use the smallest possible set of control statements rather than put in a bit of effort to get to a more efficient process.

by **Frank Yaeger** » Mon Feb 07, 2011 11:51 pm

I have been hearing from ppl that a ICETOOL step is performance intensive than a normal sort.

People say a lot of things. That doesn't necessarily make them true. A blanket statement like that is not true.

DFSORT does NOT support XSUM. The job Kolusu showed is DFSORT's equivalent of the XSUM function. For more information, see the "Keep dropped duplicate records (XSUM)" Smart DFSORT Trick at:

http://www.ibm.com/support/docview.wss? ... g3T7000094

IBM Mainframe Forum

Dupes in one file, rest in another

Dupes in one file, rest in another

Re: Dupes in one file, rest in another

Re: Dupes in one file, rest in another

Re: Dupes in one file, rest in another

Re: Dupes in one file, rest in another