SQL and Relational Databases Oracle Database Tips by Donald Burleson

Dr. E. F. Codd first introduced the concept of the Structured Query Language (SQL) as part of his early work on the relational database model.  Dr. Codd was famous for his creative names for his concepts and his rationale for naming SQL is no exception.  For example, Codd named the process of relational database design "normalization" because President Nixon was normalizing relations with China at the time, and Dr. Codd reasoned that if Nixon could normalize relations between countries then he could normalize relations between data relations.  The naming of the relational query language as SQL is another great misnomer.  One should note that SQL is NOT a query language.  SQL performs much more than queries (SQL allows updates, deletes, and inserts), and SQL  is also not a language, (SQL is embedded within procedural languages such as Cobol or "C"). Consequently, the name of Structured Query Language seemed a logical name for Dr. Codd's new tool.

Regardless of the appropriateness of the name, SQL offers three basic classes of operators, SELECT, PROJECT and JOIN.  The SELECT operator serves to shrink a relational table vertically by eliminating unwanted rows.  The PROJECT operator serves to shrink the table horizontally, removing unwanted columns, and the JOIN operator allowed the dynamic linking of two tables which share a common column value.  Most commercial implementations of SQL do not support a PROJECT operation, and projections are  achieved by specifying the columns which are desired in the output.  The JOIN operation is achieved by stating the selection criteria for two tables, and equating them with their common columns. 

In its day, the SQL language was a revolution.  It was no longer necessary to manually navigate the database, one record at a time in order to resolve a database request.  The new features of the SQL language included:

Data Access Flexibility.    The data resides in freestanding tables, which are not hard-linked with other tables. Columns can be added to relational tables without any changes to application programs, and the addition of new data or data relationships to the data model seldom require restructuring of the tables.

Declarative Data Access.  Database navigation is hidden from the programmers.   When compared to a navigational languages such as CODASYL DML, in which the programmer was required to know the details of the access paths, relational access is handled with an "SQL" optimizer, which takes care of all navigation on behalf of the user.  Relational data access is a "state space" approach, whereby the user specifies the Boolean conditions for the retrieval, and the system returns the data which meets the  selection criteria in the SQL statement.

A Simple Conceptual Framework. The relational database is very easy to describe, and even naive users can understand the concept of tables.  Complex network diagrams which are used to describe the structure of network and hierarchical databases are not needed to describe a relational database.

Referential Integrity (RI).  Relational systems allow for the control of business rules with "constraints".  These RI rules are used to insure that one-to-many and many-to-many relationships are enforced within the relational tables.  For example, RI would insure that a row in the CUSTOMER table could not be deleted if orders for that customer exist in the ORDER table.

One of the greatest benefits of the relational databases is the concept of data independence.  Because data relationships were no longer hard-linked with pointers, systems developers were able to design  systems based upon business requirements with far less time being spend on physical considerations.

 

---