Sequential File Processing

A sequential file stores records one after another in the physical order they were written, with no built-in index or key. To find a specific record, a program must read every record before it, starting from the beginning of the file. This organization is declared in the FILE-CONTROL paragraph of the ENVIRONMENT DIVISION using ORGANIZATION IS SEQUENTIAL, and it remains the backbone of batch processing on mainframes because it is fast for full-file passes such as month-end billing runs or payroll extracts.

The SELECT clause in FILE-CONTROL names the file and links it to an external dataset via ASSIGN TO, while ORGANIZATION IS SEQUENTIAL (often the default and omitted) confirms the layout. The FD entry in the DATA DIVISION describes the record layout, including RECORD CONTAINS and the 01-level record description with its fields. Before any I/O statement executes, the file must be opened with OPEN INPUT for reading, OPEN OUTPUT for creating a new file, OPEN EXTEND to append to an existing file, or OPEN I-O for update-in-place scenarios that are rarely used with pure sequential files.

The READ statement retrieves the next record into the record area defined by the FD; because sequential files have no random positioning, each READ automatically advances to the following record. The mandatory AT END clause (or NOT AT END) detects when the last record has been consumed, typically by setting a WS-EOF-FLAG to 'Y' so the PERFORM UNTIL loop can terminate cleanly. WRITE outputs the current contents of the record area to the file, and because sequential output is append-only within a single OPEN OUTPUT session, records must be written in the final desired order — there is no mechanism to insert a record in the middle later.

Because sequential files cannot be updated in place record-by-record without rewriting the whole file, COBOL batch systems traditionally use a match-merge pattern: an old master file and a transaction file, both sorted on the same key, are read in parallel, comparing keys to decide whether to copy the master record unchanged, apply an update, insert a new record, or skip a deleted one, writing everything to a brand-new master file. This pattern, still common in payroll and billing systems, avoids random access entirely and processes millions of records efficiently in a single pass, which is why sequential organization remains attractive even in an era of relational databases.