The Advantages of Database Management

The serialization process, introduced in Chapter 17 and Chapter 18, ties a document object to a disk file. All the document's data must be read into memory when the document is opened, and all the data must be written back to disk when an updated document is closed. Obviously, you can't serialize a document that's bigger than the available virtual memory. Even if the document is small enough to fit in memory, you might not need to read and write all the data every time the program runs.

You could program your own random access disk file, thus inventing your own DBMS, but you probably have enough work to do already. Besides, using a real DBMS gives you many advantages, including the following:

• Use of standard file formats—Many people think of dBASE/Xbase DBF files when they think of database formats. This is only one database file format, but it's a popular one. A lot of data is distributed in DBF files, and many programs can read and write in this format. Lately, the Microsoft Access MDB format has become popular, too. With the MDB format, all of a database's tables and indexes can be contained in a single disk file.

• Indexed file access—If you need quick access to records by key (a customer name, for example), you need indexed file access. You could always write your own B-tree file access routines, but that's a tedious job that's been done already. All DBMS's contain efficient indexed access routines.

• Data integrity safeguards—Many professional DBMS products have procedures for protecting their data. One example is transaction processing. A transaction encompasses a series of related changes. If the entire transaction can't be processed, it is rolled back so that the database reverts to its original state before the transaction.

• Multiuser access control—If your application doesn't need multiuser access now, it might in the future. Most DBMS's provide record locking to prevent interference among simultaneous users. Some multiuser DBMS's use the client-server model, which means that most processing is handled on a single database server computer; the workstations handle the user interface. Other multiuser DBMSs handle database processing on the workstations, and they control each workstation's access to shared files.