Relational databases differ from indexed file systems in several significant ways. These are the logical associations between database concepts and COBOL indexed file concepts: 
| Indexed File Concept     | 
Database Concept  | 
| Directory  | 
Database  | 
| File  | 
Table  | 
| Record  | 
Row  | 
| Field  | 
Column  | 
 
 
Or, put another way: 
- Database operations are performed on columns. Indexed file operations are performed on records 
 
- A COBOL indexed file is logically represented in database table format. Within a table, each column represents one COBOL field; each row represents one COBOL record 
 
- Database table columns are strictly associated with a particular data type, such as date, integer, or character. In COBOL, data can have multiple type definitions 
 
For example, for Oracle, a COBOL record that looks like this: 
01  terms-record.
    03  terms-code        pic 999.
    03  terms-rate        pic s9v999.
    03  terms-days        pic 9(2).
    03  terms-descript    pic x(15).
would be represented in the database as a table with a format similar to this: 
| Name  | 
Null  | 
Type  | 
| TERMS_CODE  | 
NOT NULL      | 
NUMBER (3)  | 
| TERMS_RATE  | 
–  | 
NUMBER (4, 3)  | 
| TERMS_DAYS  | 
–  | 
NUMBER (2)  | 
| TERMS_DESCRIPT      | 
–  | 
CHAR (15)  |