Skip to main content

What is Sorting?​

Sorting and data transformation are essential operations in enterprise batch processing. IBM's DFSORT utility is the backbone for sorting, merging, filtering, and formatting large datasets efficiently on z/OS.

It plays a critical role in automating business workflows—such as payroll, billing, and report generation—by organizing and manipulating data before further processing. With built-in support, high performance, and rich features, DFSORT is a reliable tool for handling data-heavy jobs in mainframe environments.

This handout introduces DFSORT's core concepts, its syntax, practical usage. In short, sorting means arranging records in a specific order—alphabetically, numerically, or by date.

IBM Docs - DFSORT Introduction

Why Sorting Matters in Enterprise Systems​

  • Makes large datasets easier to manage
  • Speeds up searches and report generation
  • Reduces data errors in billing, payroll, and reporting
  • Essential for tasks like file merging, deduplication, and analytics

💡 Without sorting, enterprise systems become slow and inefficient.

DFSORT is IBM’s built-in z/OS utility for:

  • Sorting, merging, and reformatting datasets
  • Filtering and summarizing data efficiently
  • Used in batch JCL to handle large volumes of data

DFSORT JCL example​

There are more than 20 control statements. Here are few commonly used.

IBM Docs - DFSORT Control Statement

Required DD Statements

  • SYSOUT – Log and messages
  • SYSIN – DFSORT control statements
  • SORTIN – Input dataset
  • SORTOUT – Output dataset

Execution Flow

INREC  →  INCLUDE  →  SORT / SUM  →  OMIT  →  OUTREC
tip
  • SORT FIELDS=COPY performs no sort, just filtering or formatting.
  • Use OPTION COPY when no sort is needed—faster than SORT FIELDS=COPY.
  • SUM FIELDS=NONE is a great way to deduplicate records.
  • Use INREC to filter early, reducing data before sorting.
  • OUTFIL SPLIT is handy for splitting one dataset into multiple outputs.

SORT​

SORT – Specifies the key fields and order (ascending or descending) used to sort the input records.

IBM Docs - SORT Statement

//SYSIN DD *
  SORT FIELDS=(start,length,format,A/D)
  SORT FIELDS=COPY,SKIPREC=5,STOPAFT=100
/*

OPTION​

OPTION – Used to set general processing options like copying without sorting, skipping or limiting records.

IBM Docs - Option Statement

//SYSIN DD *
  OPTION COPY,SKIPREC=10,STOPAFT=100
/*

INCLUDE​

INCLUDE – Filters records to include only those that match specific conditions.

IBM Docs - Include Statement

//SYSIN DD *
  INCLUDE COND=(start,length,format,OP,value)
/*

OMIT​

OMIT – Filters out records that match certain conditions, effectively the opposite of INCLUDE.

IBM Docs - Omit Statement

//SYSIN DD *
  OMIT COND=(start,length,format,OP,value)
/*

OUTFIL​

OUTFIL – Sends different output records to multiple files, or controls formatting and splitting of output.

IBM Docs - OUTFIL Statement

//SYSIN DD *
  OUTFIL FILES=(01,02),SPLIT
  OUTFIL FILES=01,INCLUDE=(...),STARTREC=5,ENDREC=20
/*

SUM​

SUM – Removes duplicates or totals up numeric fields for records with the same key values.

IBM Docs - SUM Statement

Used to add numeric fields or remove duplicates.

//SYSIN DD *
  SUM FIELDS=(start,length,format)
  SUM FIELDS=NONE             ← Removes duplicate records
  SUM FIELDS=NONE,XSUM        ← Move duplicates to another file (not supported here)
/*

INREC vs OUTREC​

INREC/OUTREC – Reformats records before (INREC) or after (OUTREC) sorting, merging, or copying.

IBM Docs - INREC Statement

IBM Docs - OUTREC Statement

//SYSIN DD *
OUTREC FIELDS=(1:1,10,11:C'**')
OUTREC FINDREP=(IN=C'OLD',OUT=C'NEW')
OUTREC OVERLAY=(5:C'ABCD')
OUTREC IFTHEN=(WHEN(conditions),OVERLAY=(...))
/*

📝 Same syntax applies to INREC and OUTREC.