Because a database is usually a model of a real-world situation, certain pieces of data are inevitably missing, unknown, or don’t apply. In the sample database, for example, the QUOTA column in the SALESREPS table contains the sales goal for each salesperson. However, the newest salesperson has not yet been assigned a quota; this data is missing for that row of the table. You might be tempted to put a zero in the column for this salesperson, but that would not be an accurate reflection of the situation. The salesperson does not have a zero quota; the quota is just “not yet known.”
Similarly, the MANAGER column in the SALESREPS table contains the employee number of each salesperson’s manager. But Sam Clark, the Vice President of Sales, has no manager in the sales organization. This column does not apply to Sam. Again, you might think about entering a zero, or a 9999 in the column, but neither of these values would really be the employee number of Sam’s boss. No data value is applicable to this row.
SQL supports missing, unknown, or inapplicable data explicitly, through the concept of a null value. A null value is an indicator that tells SQL (and the user) that the data is missing or not applicable. As a convenience, a missing piece of data is often said to have the value NULL. But the NULL value is not a real data value like 0, 473.83, or “Sam Clark.” Instead, it’s a signal, or a reminder, that the data value is missing or unknown. Figure 5-3 shows the contents of the SALESREPS table. Note that the QUOTA and REP_OFFICE values for Tom Snyder’s row and the MANAGER value for Sam Clark’s row of the table all contain NULL values.
In many situations, NULL values require special handling by the DBMS. For example, if the user requests the sum of the QUOTA column, how should the DBMS handle the missing data when computing the sum? The answer is given by a set of special rules that govern NULL value handling in various SQL statements and clauses. Because of these rules, some leading database authorities feel strongly that NULL values should not be used. Others, including Dr. Codd, have advocated the use of multiple NULL values, with distinct indicators for “unknown” and “not applicable” data.
Regardless of the academic debates, NULL values are a well-entrenched part of the ANSI/ISO SQL standard and are supported in virtually all commercial SQL products. They also play an important, practical role in production SQL databases. The special rules that apply to NULL values (and the cases where NULL values are handled inconsistently by various SQL products) are pointed out throughout this book.
Source: Liang Y. Daniel (2013), Introduction to programming with SQL, Pearson; 3rd edition.