Explain COBOL Program Structure in IBM Mainframes

Every COBOL program has four divisions, always in the same order. IDENTIFICATION does nothing at runtime. It is still mandatory. Program name, author

All topics

1 / 60

COBOLbeginner

COBOL Program Structure

Understand the four mandatory divisions that every COBOL program is built from and what belongs in each.

Every COBOL program has four divisions, always in the same order.

Think of a COBOL program like a shipping manifest form with four locked sections in a fixed order — you can't write the cargo details before filling in the ship's registration, and you can't skip a section just because it feels redundant that day.

Key Concepts

    IDENTIFICATION DIVISION.   what the program is called
    ENVIRONMENT DIVISION.      which files it uses, and where
    DATA DIVISION.             every piece of storage it touches
    PROCEDURE DIVISION.        the code that runs

IDENTIFICATION does nothing at runtime. It is still mandatory. Program name, author, comments.

ENVIRONMENT is the bridge to JCL. This is the part worth understanding properly:

    SELECT CUSTOMER-FILE ASSIGN TO CUSTFILE.

CUSTOMER-FILE is the name the program uses. CUSTFILE is a DD name. The JCL says which dataset that is:

CUSTOMER-FILECUSTFILE

    //CUSTFILE DD DSN=PROD.CUSTOMER.MASTER,DISP=SHR

So the program never names a dataset. The same compiled program reads test data or production data, depending on the JCL. That is why batch systems are built this way — you change the file without recompiling anything.

DATA is where the verbosity comes from, and it buys something. Nothing exists unless it is declared, with a level number and a picture clause:

    WORKING-STORAGE SECTION.   the program's own variables
    FILE SECTION.              the record layout of each file
    LINKAGE SECTION.           parameters passed in by a caller

A record layout describes bytes at fixed offsets. That is what lets a program read a file written by another program thirty years ago. The layout is the contract, and both sides state it exactly.

PROCEDURE holds the code, organised into paragraphs:

    PROCEDURE DIVISION.
    MAIN-PARA.
        PERFORM OPEN-FILES
        PERFORM PROCESS-RECORDS UNTIL END-OF-FILE
        PERFORM CLOSE-FILES
        STOP RUN.

The mental shift, for anyone from a modern language. A paragraph is not a function.

It has no parameters, no return value and no local variables. Every variable it touches is declared in the DATA DIVISION and visible to the whole program.

So a paragraph is a named block you jump to, and all state is global. That explains a lot about how legacy COBOL reads, and why working out what a paragraph affects means reading it rather than reading a signature.

Being able to write this skeleton from memory is a baseline expectation. The ENVIRONMENT-to-JCL link is the piece candidates most often miss.

What the interviewer is probing.1. "Name the four divisions and what each is for." Probing: the foundation. Stalls: "There are divisions." Moves up: IDENTIFICATION names the program, ENVIRONMENT maps files to the system, DATA declares storage, PROCEDURE holds the logic.

2. "What is the difference between WORKING-STORAGE and LOCAL-STORAGE?" Probing: a real distinction. Stalls: "They are the same." Moves up: WORKING-STORAGE is allocated once and keeps its values between calls; LOCAL-STORAGE is fresh on every invocation, which matters for recursion and CICS.

3. "What does the FILE SECTION record describe?" Probing: the mechanism. Stalls: "The file." Moves up: the layout of one record in the buffer — it is not storage you own until a READ fills it, which is why referencing it before a successful read is a bug.

4. "Why do shops insist on structured paragraph naming?" Probing: the maintenance angle. Stalls: "Convention." Moves up: these programs are read far more often than written, and a numbered, ordered naming scheme is what makes a 10,000-line program navigable.