Section 2.9. Profitable Database Accesses

2.9. Profitable Database Accesses

When you plan a visit to several shops, the first step is to decide what purchases have to be made at each shop. From this point, a trip is planned that will ensure minimum repetitive walking backward and forward between different shops. The first shop is then visited, the purchase completed, and then the next closest shop is visited. This is only common sense, and yet the principle underlying this obvious approach is not seen in the practical implementation of many database programs.

When several pieces of information are required from a single tableeven if it appears as if they are "unrelated" (which in fact is unlikely to be the case)--it is highly inefficient to retrieve this data in several separate visits to the database. For example, do not fetch row values column by column if multiple columns are required: do the work in one operation.

Unfortunately, good object-oriented (OO) practice makes a virtue out of defining one method for returning each attribute. But do not confuse OO methods with relational database processing. It is a fatal mistake to mix relational and object-oriented concepts and to consider tables to be classes with columns as the attributes.

Maximize each visit to the database to complete as much work as can reasonably be achieved for every visit.