In many database designs, information is distributed to multiple tables. The following image shows two tables, one for employee information and one for employee addresses:

View full size graphic

Employees table

View full size graphic

Addresses tables

In this example, each table contains a column named EmpID. This column associates a row of the employees table with a row in the addresses table.

For example, to obtain all information about an employee, you request a row from the employees table and the row from the addresses table with the same value for EmpID.

One advantage of using multiple tables is that you can add tables containing new information without modifying the structure of your existing tables. For example, to add payroll information, you add a new table to the database where the first column contains the employee’s ID and the columns contain current salary, previous salary, bonus payment, and 401(k) percent.

Also, an access to a small table is more efficient than an access to a large table. Therefore, if you update the street address of an employee, you update only the addresses table, without having to access any other table in the database.

In many database environments, a database administrator defines the access privileges for users accessing the database, usually through user name and password. When a person attempts to connect to a database, the database ensures that the user name and password are valid and then imposes access requirements on the user.

Privileges can restrict user access so that a user can do the following:

• Read data.

• Read data and add rows.

• Read data, add rows, modify existing tables.

In ColdFusion, you use the ColdFusion Administrator to define database connections, called data sources. As part of defining these connections, you specify the user name and password used by ColdFusion to connect to the database. The database can then control access based on this user name and password.

For more information on creating a data source, see Configuring and Administering ColdFusion.

Before you access data stored in a database, it is important to understand several database concepts, including:

• Commit

• Rollback

• Transactions

A database commit occurs when you make a permanent change to a database. For example, when you write a new row to a database, the write does not occur until the database commits the change.

Rollback is the process of undoing a change to a database. For example, if you write a new row to a table, you can rollback the write up to the point where you commit the write. After the commit, you can no longer rollback the write.

Most databases support transactions where a transaction consists of one or more SQL statements. Within a transaction, your SQL statements can read, modify, and write a database. You end a transaction by either committing all your changes within the transaction or rolling back all of them.

Transactions can be useful when you have multiple writes to a database and want to make sure all writes occurred without error before committing them. In this case, you wrap all writes within a single transaction and check for errors after each write. If any write causes an error, rollback all of them. If all writes occur successfully, you commit the transaction.

A bank might use a transaction to encapsulate a transfer from one account to another. For example, if you transfer money from your savings account to your checking account, you do not want the bank to debit the balance of your savings account unless it also credits your checking account. If the update to the checking account fails, the bank can rollback the debit of the savings account as part of the transaction.

ColdFusion includes the cftransaction tag that lets you implement database transactions for controlling rollback and commit. For more information, see the CFML Reference.

From this basic description, the following database design rules emerge:

• Each record should contain a unique identifier as the primary key such as an employee ID, a part number, or a customer number. The primary key is typically the column used to maintain each record's unique identity among the tables in a relational database. Databases allow you to use multiple columns for the primary key.

• When you define a column, you define a SQL data type for the column, such as allowing only numeric values to be entered in the salary column.

• Assessing user needs and incorporating those needs in the database design is essential to a successful implementation. A well-designed database accommodates the changing data needs within an organization.

The best way to familiarize yourself with the capabilities of your database product or database management system (DBMS) is to review the product documentation.