Chapter 4: Exploring Relational Database Theory and Practice

Thursday Jun 21st 2007 by DatabaseJournal.com Staff

This excerpt from "Special Edition Using Microsoft Office Access 2007" discusses Moving from Spreadsheets to Databases, Defining the Structure of Relational Databases, Maintaining Data Integrity and Accuracy and more.

Special Edition Using Microsoft Office Access 2007
Roger Jennings
May 1, 2007, 1488 pages
ISBN10: 0-7897-3597-0
Publisher: Que
Buy this book

This excerpt from "Special Edition Using Microsoft Office Access 2007" discusses Moving from Spreadsheets to Databases, Defining the Structure of Relational Databases, Maintaining Data Integrity and Accuracy and more.

In this chapter

Moving from Spreadsheets to Databases

Reliving Database History

Defining the Structure of Relational Databases

Conforming to Table Design Rules

Choosing Primary Key Codes

Maintaining Data Integrity and Accuracy

Displaying Data with Queries and Views

In the Real World—When and Why Learn Relational Theory?

Moving from Spreadsheets to Databases

Word processing and spreadsheet applications were the engines that drove the fledgling personal computer market. In the early PC days, WordPerfect and Lotus 1-2-3 dominated the productivity software business. Today, most office workers use Microsoft Word and Excel on a daily basis. It's probably a safe bet that there's more data stored in Excel spreadsheets than in all the world's databases. It's an equally good wager that most new Access users have at least intermediate-level spreadsheet skills, and many qualify as Excel power users.

Excel's 2007 Data ribbon offers elementary database features, such as sorting, filtering, validation, and data entry forms. You can quickly import and export data in a variety of formats, including those of database management applications, such as Access. Excel's limitations become apparent as your needs for entering, manipulating, and reporting data grow beyond the spreadsheet's basic row-column metaphor. Spreadsheets basically are list managers; it's easy to generate a simple name and address list with Excel. If your needs expand to contact management and integrating the contact data with other information generated by your organization, a spreadsheet isn't the optimal approach.

The first problem arises when your contacts list needs additional rows for multiple persons from a single company. You must copy or retype all the company information, which generates redundant data. If the company moves, you must search and replace every entry for your contacts at the firm with the new address. If you want to record a history of dealings with a particular individual, you add pairs of date and text columns for each important contact with the person. Eventually, you find yourself spending more time navigating the spreadsheet's rows and columns than using the data they contain.

Contact lists are only one example of problems that arise when attempting to make spreadsheets do the work of databases. Tracking medical or biological research data, managing consulting time and billings, organizing concert tours, booking artist engagements, and a myriad of other complex processes are far better suited to database than spreadsheet applications.

Moving to a relational database management system (RDBMS), such as Access, solves data redundancy and navigation problems, and greatly simplifies updating existing information. After you understand the basic rules of relational database design, Access makes creating highly efficient databases quick and easy. Access 2007 has a collection of wizards to lead you step-by-step through each process involved in developing and using a production-grade database application. Unfortunately, there's no "Relational Wizard" to design the underlying database structure for you.

Tip - If your goal is learning relational database fundamentals, start with Access 2007. Access is by far the first choice of universities, colleges, trade schools, and computer-training firms for courses ranging from introductory data management to advanced client/server database programming. The reason for Access's popularity as a training platform is its unique combination of initial ease of use and support for advanced database application development techniques.

Reliving Database History

Databases form the foundation of world commerce and knowledge distribution. Without databases, there would be no World Wide Web, automatic teller machines, credit/debit cards, or online airline reservation systems. Newsgathering organizations, research institutions, universities, and libraries would be unable to categorize and selectively disseminate their vast store of current and historical information. It's difficult to imagine today a world without a network of enormous databases, many of which probably contain a substantial amount of your personal data that you don't want to be easily available to others.

The Early History of Databases

The forerunner of today's databases consisted of stacks of machine-readable punched cards, which Herman Hollerith used to record the 1890 U.S. census. Hollerith formed the Computing-Tabulating-Recording Company, which later became International Business Machines. From 1900 to the mid-1950s, punched cards were the primary form of business data storage and retrieval, and IBM was the primary supplier of equipment to combine and sort (collate) punched cards, and print reports based on punched-card data.

Note - Jim Gray's article "Data Management: Past, Present, and Future," which is available as a Microsoft Word document at http://research.microsoft.com/~gray/DB_History.doc, offers a more detailed history of data processing systems. Dr. Gray is a senior researcher and the manager of Microsoft's Bay Area Research Center (BARC).

The development of large computer-maintained databases—originally called databanks—is a post–World War II phenomenon. Mainframes replaced punched cards with high-capacity magnetic tape drives to store large amounts of data. The first databases were built on the hierarchical and network models, which were well suited to the mainframe computers of the 1950s. Hierarchical databases use parent-child relationships to define data structures, whose diagrams resemble business organization charts or an inverted tree with its root at the top of the hierarchy. Network databases allow relaxation of the rules of hierarchical data structures by defining additional relationships between data items. Hierarchical and network databases ordinarily are self-contained and aren't easy to link with other external databases over a network.

Note - Hierarchical databases remain alive and well in the twenty-first century. For example, data storage for Windows 2000's Active Directory and Microsoft Exchange Server is derived from the hierarchical version of Access's original relational Jet databases. The name Jet comes from the original Access database engine called Joint Engine Technology.

The Internet's Domain Name System (DNS) is a collection of hierarchical databases for translating character-based Internet domain names into numerical Internet Protocol (IP) addresses. The DNS database is called a distributed database, because its data is held by a global network of thousands of computers.

Early databases used batch processing for data entry and retrieval. Keypunch operators typed data from documents, such as incoming orders. At night, other operators collated the day's batch of punched cards, updated the information stored on magnetic tape, and produced reports. Many smaller merchants continue to use batch processing of customer's credit-card purchases, despite the availability of terminals that permit almost instantaneous processing of credit- and debit-card transactions.

The Relational Database Model

Dr. E. F. Codd, an employee of IBM Corporation, published "A Relational Model of Data for Large Shared Databanks" in a journal of the Association for Computing Machinery (ACM) in June 1970. A partial copy of the paper is available at http://www.acm.org/ classics/nov95/. Dr. Codd's specialty was a branch of mathematics called set theory, which includes the concept of relations. He defined a relation as a named set of tuples (records or rows) that have attributes (fields or columns). One of the attributes must contain a unique value to identify each tuple. The common term for relation is a table whose presentation to the user is similar to that of a spreadsheet.

Note - This book uses the terms field and record when referring to tables, and columns and rows when discussing data derived from tables, such as the views and query result sets described later in this chapter.

Relational databases solve a serious problem associated with earlier database types. Hierarchical and network databases define sets of data and explicit links between each data set as parent-child and owner-member, respectively. To extract information from these databases, programmers had to know the structure of the entire database. Complex programs in COBOL or other mainframe computer languages are needed to navigate through the hierarchy or network and extract information into a format understandable by users.

Dr. Codd's objective was to simplify the process of extracting formatted information and make adding or altering data easier by eliminating complex navigational programming. During the 1970s, Dr. Codd and others developed a comparatively simple language, Structured Query Language (SQL), for creating, manipulating, and retrieving relational data. With a few hours of training, ordinary database users could write SQL statements to define simple information needs and bypass the delays inherent in the database programming process. SQL, which was first standardized in 1985, now is the lingua franca of database programming, and all commercial database products support SQL.

Note - The most widely used SQL standard, SQL-92, was published by the American National Standards Institute (ANSI) in 1992. Few, if any, commercial relational database management systems (RDBMSs) today fully conform to the entire SQL-92 standard. The later SQL-99 (also called SQL3) and SQL-200n specifications add new features that aren't germane to Access databases.

RDBMS competitors have erected an SQL Tower of Babel by adding nonstandard extensions to the language. For example, Microsoft's Transact-SQL (T-SQL) for SQL Server, which is the subject of Chapter 21, "Moving from Access Queries to Transact-SQL," has many proprietary keywords and features. Oracle Corporation's Oracle:SQL and PL/SQL dialects also have proprietary SQL extensions.

Client/Server and Desktop RDBMSs

In the early database era, the most common presentation of data took the form of lengthy reports processed by centralized, high-speed impact printers on fan-folded paper. The next step was to present data to the user on green-screen video terminals, often having small printers attached, which were connected to mainframe databases. As use of personal computers gained momentum, terminal emulator cards enabled PCs to substitute for mainframe terminals. Mainframe-scale relational databases, such as IBM's DB2, began to supplement and later replace hierarchical and network databases, but terminals continued to be the primary means of data entry and retrieval.

Oracle, Ingres, Informix, Sybase, and other software firms developed relational databases for lower-cost minicomputers, most of which ran various flavors of the Unix operating system. Terminals continued to be the primary data entry and display systems for multiuser Unix databases.

The next step was the advent of early PC-based flat-file managers and relational database management systems. Early flat-file database managers, typified by Jim Button's PCFile for DOS (1981) and Claris FileMaker for Macintosh (1988) and Windows (1992), used a single table to store data and offered few advantages over storing data in a spreadsheet. The early desktop RDBMSs—such as dBASE, Clipper, FoxBase, and Paradox—ran under DOS and didn't support SQL. These products later became available in multiuser versions, adopted SQL features, and eventually migrated to Windows. Access 1.0, which Microsoft introduced in November 1992, rapidly eclipsed its DOS and Windows competitors by virtue of Access's combination of graphical SQL support, versatility, and overall ease of use.

PC-based desktop RDBMSs are classified as shared-file systems, because they store their data in conventional files that multiple users can share on a network. One of Access's initial attractions for users and developers was its capability to store all application objects—forms, reports, and programming code—and tables for a database application in a single .mdb file. FoxPro, dBASE, Clipper, and Paradox require a multitude of individual files to store application and data objects. Today, almost every multiuser Access application is divided (split) into a front-end .accdb file, which contains application objects and links to a back-end database .accdb file that holds the data. Each user has a copy of the front-end .accdb file and shares connections to a single back-end .accdb file on a peer Windows workstation or server.

Client/server RDBMSs have an architecture similar to Access's front-end/back-end shared-file multiuser configuration. What differentiates client/server from shared-file architecture is that the RDBMS on the server handles most of the data-processing activity. The client front end provides a graphical user interface (GUI) for data entry, display, and reporting. Only SQL statements and the specific data requested by the user pass over the network. Client/server databases traditionally run on network operating systems, such as Windows and Unix, and are much more robust than shared-file databases, especially for applications in which many users make simultaneous additions, changes, and deletions to the database. All commercial data-driven Web applications use client/server databases.

Since version 1.0, Access has had the capability to connect to client/server databases by linking their tables to an Access database. Linking lets you treat client/server tables almost as if they were native Access tables. Linking uses Microsoft's widely accepted Open Database Connectivity (ODBC) standard, and Access 2007 includes an ODBC driver for SQL Server and Oracle databases. You can purchase licenses for ODBC drivers that support other Unix or Windows RDBMSs, such as Sybase or Informix, from the database supplier or third parties. Chapter 19, "Linking Access Front-Ends to Access and Client/Server Tables," describes the process of linking Access and Microsoft SQL Server 2005 databases. Although Chapter 19 uses SQL Server for its examples, the linking procedure is the same for—or at least similar to—other client/server RDBMSs.

Note - Prior to Access 2000, Jet was Access's standard database engine, so the terms Access database and Jet database were interchangeable. Microsoft considered SQL Server to be its strategic RDBMS for Access 2000 and 2003. Strategic means that SQL Server gets continuing development funds and Jet doesn't. Jet 4.0, which was included with Access 2003 and is a part of the Windows XP and later operating systems, is the final version and is headed toward retirement.

Microsoft's Access team decided to enhance Jet 4.0 with the new features described in Chapter 1, "Access 2007 for Access 200x Users: What's New," change the file extension from .mdb to .accdb, and drop all references to Jet. To reflect this change, this edition uses the terms Access database and SQL Server database. Unless otherwise noted, SQL Server refers to all SQL Server 2005 editions except the Compact Edition 3.1.

Access data projects (ADP) and the Microsoft SQL Server 2005 Express Edition combine to make Access 2007 a versatile tool for designing and testing client/server databases, and creating advanced data entry and reporting applications. You can start with a conventional Access database and later use Access's Upsizing Wizard to convert the .mdb file(s) to an .adp file that holds application objects and an SQL Server 2005 back-end database. Access 2007's Upsizing Wizard has incorporated many improvements to the Access 2000 and earlier Wizard versions, but Access 2007's Wizard is the same as 2002's. Despite the upgraded wizardry, you're likely to need to make changes to queries to accommodate differences between Access and SQL Server's SQL dialects.

→ For an example of differences between Access and SQL Server SQL syntax that affects the upsizing process, see "Displaying Data with Queries and Views."

Defining the Structure of Relational Databases

Relational databases consist of a collection of self-contained, related tables. Tables typically represent classes of physical objects, such as customers, sales orders, invoices, checks, products for sale, or employees. Each member object, such as an invoice, has its own record in the invoices table. For invoices, the field that uniquely identifies a record, called a primary key [field], is a serial invoice number.

Figure 4.1 shows Access's Datasheet view of an Invoices table, which is based on the Northwind.mdb sample database's Orders table. The InvoiceNo field is the primary key. Values in the OrderID, CustomerID, EmployeeID, and ShipperID fields relate to primary key values in Northwind's Orders, Customers, Employees, and Shippers tables. A field that contains values equal to those of primary key values in other tables is called a foreign key [field].

Figure 4.1
This simple Invoices table was created from the Northwind Orders table and doesn't take advantage of Access's extended properties, such as the field captions, lookup fields, and subdatasheets in the Datasheet view of the Orders table.

→ To learn more about primary keys in Access tables, see "Selecting a Primary Key."

If you need information about a particular invoice or set of invoices, you open the Invoices table and search for the invoice(s) by number (InvoiceNo) or another attribute, such as a customer code (CustomerID), date (ShippedDate), or range of dates. Unlike earlier database models, the user can access the Invoices table independently of its related tables. No database navigation programming is needed. A simple, intuitive SQL statement, SELECT * FROM Invoices, returns all the data in the table. The asterisk (*) represents a request to display the contents of all fields of the table.

Removing Data Redundancy with Relationships

The Invoices table of Figure 4.1 is similar to a spreadsheet containing customer billing information. What's missing is the customer name and address information. A five-character customer code (CustomerID) identifies each customer to whom the invoice is directed. The CustomerID values in the Invoices table match CustomerID values in a modified version of Northwind's Customers table (see Figure 4.2). Matching a foreign key with a primary key value often is called a lookup operation. Using a key-based lookup operation eliminates the need to repeatedly enter name, address, and other customer-specific data in the Invoices table. In addition, if you change the customer's address, the change applies to all past and future invoices.

Figure 4.2
Foreign key values in the Invoices table must match primary key values in the Customers table.

Tip - Using derived key values, such as alphabetic codes for Customary, is no longer in favor among database designers. Most designers now use automatically generated numerical key values—called Access AutoNumber or SQL Server identity fields. The Northwind Orders and Products tables, among others, have primary keys that use the AutoNumber data type. The Employees, Shippers, Products, and Suppliers tables use AutoNumber keys to identify the persons or objects to which the table's records refer. Objects that are inherently sequentially numbered, such as checks, are ideal candidates for an AutoNumber key that corresponds to the check number, as mentioned in "Choosing Primary Key Codes" later in this chapter.

Another method of generating unique keys is by use of Globally Unique Identifiers (GUIDs), which also are called Universally Unique Identifiers (UUIDs). GUIDs are 16-byte computed binary numbers that are guaranteed to be unique locally and universally; no other computer in the world will duplicate a GUID. SQL Server's uniqueidentifier data type is a GUID. Because GUIDs can't represent a property of an object, such as a check number, GUID keys are called surrogate keys. You can't select a GUID data type in Access's Table Design mode.

The Invoices table also connects with other tables, which contain information on orders, sales department employees, and the products ordered. Connections between fields of related tables having common values are called relationships (not relations). Figure 4.3 shows Access's Relationships window displaying the relationships between the Invoices table and the other tables of the Northwind sample database.

Figure 4.3
Access's Relationships window displays the relationships between the tables of the Northwind sample database, plus the added Invoices table. Every relationship between these tables is one-to-many. The many-to-many relationship between Products and Orders is an indirect relationship.

Relationships come in the following three flavors:

  • One-to-many relationships represent connections between a single primary key value (the "one" side) and multiple instances of the same value in the foreign key field (the "many" side). One-to-many relationships commonly are identified by the number one and the infinity (x) symbol, as in Figure 4.3. All the direct relationships between the tables in Figure 4.3 are one-to-many. One-to-many—also called many-to-one—relationships are by far the most common.

  • One-to-one relationships connect primary key values in two tables. You might think that the relationship between the Orders and Invoices tables could be one-to-one, but an order requires more than one invoice if one or more items are backordered and then shipped later. One-to-one relationships are uncommon.

  • Many-to-many relationships require three tables, one of which is called a linking table. The linking table must have two foreign keys, each of which has a many-to-one relationship with a primary key in two related tables. In the example of Figure 4.3, the Order Details table is the linking table for the many-to-many relationship between the Orders and Products tables. Many-to-many relationships also are called indirect relationships.

There are many other indirect relationships between the tables shown in Figure 4.3. For example, there is a many-to-many relationship between the Suppliers and Orders tables. In this case, Products and Order Details act as linking tables between the Suppliers and Orders tables.

Note - Access 2007's new multivalue field feature automatically generates a hidden linking table "under the covers."

Tip - Don't add multivalued fields to tables of databases that you might want to upsize to SQL Server some day. SQL Server doesn't support the Allow Multiple Values option that enables or disables the feature.

The OrderID and ProductID fields comprise a composite primary key, which uniquely identifies an order line item. You can't repeat the same combination of OrderID and ProductID; this precaution makes sense for products that have only one stock-keeping unit (SKU), such as for Aniseed Syrup, which comes only in a carton of 12 550-ml bottles.

Note - The one-product-entry-per-order restriction prevents shared use of the Order Details table as an invoice line items table. If you short-ship an order item on one invoice, you can't add another record to the Order Details table when you ship the remaining quantity of the item. Microsoft didn't add an Invoices table for Northwind Traders, probably because of the complexity of dealing with backorders and drop-shipments.

The Oakmont.accdb sample database file in the \Seua2007\Oakmont folder of the accompanying CD-ROM has a structure that differs from that of Northwind.accdb, but the design principles of the two databases are similar. OakmontSQL.mdf is an SQL Server 2005 database for use with ADP. ADP uses a special set of tools—called the project designer or da Vinci toolset in this book—for designing and managing SQL Server databases.The Oakmont files are course enrollment databases for a college. Figure 4.4 shows the Database Diagram window for the OakmontSQL database. The SQL Server Diagram window is similar to the Relationships window for Access's traditional Access databases. The key and infinity symbols at the ends of each line represent the one and many sides, respectively, of the one-to-many relationships between the tables. Access and SQL Server databases store information on table relationships as an object within the database file.

Figure 4.4
The SQL Server Database Diagram window for the OakmontSQL database shows one-to-many relationships between primary key fields (identified by key symbols) and foreign key fields (infinity symbols).

This book uses the Access 2007 and SQL Server 2005 versions of the Northwind and Oakmont sample databases in almost all examples. The tables of the Oakmont database have many more records than the Northwind tables. The large number of records in the Oakmont database makes it better suited than Northwind for predicting the performance of production Access and SQL Server database applications.

Conforming to Table Design Rules

Designing tables for relational databases follows a formalized procedure called normalization. Dr. Codd described the complete normalization process in his 1972 paper "Further Normalization of the Data Base Relational Model." This paper isn't an easy read; it's steeped in the language of set theory and relational algebra. The sections that follow explain in common English the application of the normalization process to Access's Northwind database.

You normalize tables in a series of steps called normal forms. Applying the normalization process is necessary to move spreadsheet-style data to relational tables. You also employ the normalization rules when designing a new database or analyzing existing databases. In specific cases, however, you might need to depart from strict adherence to normalization rules to retain a history of data values that change over time or to improve performance of a large database.

First Normal Form

First normal form requires tables to be flat and have no repeating or potentially repeating fields or groups of fields. A flat table is one in which every record has the same number of fields. In addition, a single field cannot contain multiple data values. Repeating fields must be moved to a related table. The first normal form is the most important of the normalization steps. If all your tables don't meet the rules of first normal form, you are in big trouble.

Northwind's Customers and Suppliers tables violate the no repeating fields rule. If a customer or supplier has more than one person involved in the ordering process, which is likely, the table would need repeating pairs of fields with different names, such as ContactName2 and ContactTitle2 or the like. To conform the Customers and Suppliers tables to first normal form, you must create two new tables—CustPers(sonel) and SuppPers(sonel), for example—to hold contact records. Including contact names in the Customers and Suppliers tables also violates third normal form, which is the subject of the later "Third Normal Form" section.

The ContactName field also violates the rule against multiple data values in a single field by combining given and family names. This isn't a serious violation of first normal form, but it's a good database design practice always to identify persons by given and family names in separate fields. When you create the new CustPers and SuppPers tables, separate the ContactName field into two fields, such as LastName and GivenName, which can include initials. You can then use a code similar to that for CustomerID for the ContactID field. For this example, the ContactID code is the first character of GivenName and the first four characters of LastName. Alternatively, you could assign an AutoNumber value to ContactID.

Figure 4.5 shows the first 19 of the 91 records of the CustPers table generated from the Customers table. The CustomerID field is required for a many-to-one relationship with the Customers table. Additional fields, such as Suffix, TitleOfCourtesy, Email(Address), Phone, and Fax, make the individual contact records more useful for creating mailing lists and integration with other applications, such as Microsoft Outlook.

Figure 4.5
You extract data for records of the CustPers table from the ContactName and ContactTitle fields of the Customers table. Separating given and last names simplifies generating a ContactID code to identify each record.

You don't need to retype the data to populate the CustPers and SuppPers tables. You can use Access to import the data from an Excel worksheet or text file, or use Access action queries (append and update) to handle this chore.

→ For more information on importing from Excel, see "Importing and Linking Spreadsheet Files."

→ To learn how to use Access action queries, see "Creating Action Queries to Append Records to a Table."

Figure 4.6 shows the Relationships window with the CustPers and SuppPers tables added to the Northwind database and their many-to-one relationships with the Customers and Suppliers tables, respectively.

Figure 4.6
The Relationships window displays the many-to-one relationships between the Customers and CustPers tables and the Suppliers and SuppPers tables.

Second Normal Form

Second normal form requires that data in all nonkey fields be fully dependent on the value of a primary key. The objective of second normal form is to avoid data redundancy in your tables.

Only Northwind's Order Details linking table (see Figure 4.7) has a composite primary key (OrderID + ProductID). The UnitPrice field appears to violate the second normal form, because UnitPrice is a field of the Products table. UnitPrice values added to the Order Details table are dependent on the ProductID component of the composite primary key and not the OrderID component, so UnitPrice data is not fully dependent on the primary key. On first glance, the UnitPrice field appears to be redundant data. If you change the unit price of a product, it would appear that you would need to alter the UnitPrice value in every Order Details record for the product.

Figure 4.7
The Order Details linking table has a composite primary key consisting of the OrderID and ProductID fields.

The Order Details table is an example of a situation in which you must retain what appears to be redundant information to maintain the integrity of historical data. Prices of products vary over time, so the price of a particular product is likely to change for orders placed on different dates. If the price of a product changes between the order and shipping (invoice) dates, the invoice reflects a different amount than the order. Despite the "Prices are subject to change without notice" boilerplate, customers become incensed if the invoice price is greater than the order price.

Eliminating the UnitPrice field from the Order Details table and looking up its value from the current price in the Products table also can cause accounting errors and distortion of historical reports based on bookings and sales data. Removing the UnitPrice data also violates the rules for the fifth normal form, explained later in this chapter.

Third Normal Form

Third normal form requires that data in all nonkey fields of the table be fully dependent on the value of the primary key and describe only the object that the table represents. In other words, make sure that the table doesn't include nonkey fields that relate to some other object or process and includes nonkey fields for descriptive data that isn't contained in another related table.

As mentioned in the "First Normal Form" section, including contact information in the Customers and Products table violates third normal form rules. Contacts are persons, not customer or supplier organizations, and deserve their own related table that has attributes related to individuals.

Other examples of a common third normal form violation are the UnitsInStock and UnitsOnOrder fields of the Products table (see Figure 4.8). These fields aren't fully dependent on the primary key value, nor do they describe the object; they describe how many of the product you have now and how many you might have if the supplier decides to ship your latest order. In a production order entry database, these values vary over time and must be updated for each sale of the product, each purchase order issued to the product's supplier, and each receipt of the product. Purchases, receipts, and invoices tables are the most common source of the data on which the calculations are based.

Figure 4.8
The Products table's UnitsInStock and UnitsOnOrder values must be calculated from data in tables that record purchases, receipts, and shipments of products.

Including UnitsInStock and UnitsOnOrder fields isn't a serious violation of the normalization rules, and it's not uncommon for product-based tables of order entry databases to include calculated values. The problem with calculated inventory values is the need to process a potentially large number of records in other tables to obtain an accurate current value.

Tip - If you're designing an order entry database, make sure to take into account committed inventory. Committed inventory consists of products in stock or en route from suppliers for which you have unfulfilled orders. If you decide to include inventory information in a products table, add a UnitsCommitted field.

Fourth Normal Form

Fourth normal form requires that tables not contain fields for two or more independent, multivalued facts. Loosely translated, this rule requires splitting tables that consist of lists of independent attributes. The Northwind and Oakmont databases don't have an example of a fourth normal form violation, so the following is a fabricated example.

One of the objectives of Human Resources departments is to match employee job skills with job openings. A multinational organization is likely to require a combination of specific job skills and language fluency for a particular assignment. A table of job skill types and levels exists with entries such as JP3 for Java Programmer–Intermediate, as well as language/ fluency with entries such as TE5 for Telugu–Very Fluent. Therefore, the HR department constructs an EmplSkillLang linking table with the following foreign key fields: EmployeeID, SkillID, and LanguageID.

The problem with the linking table is that job skills and language fluency are independent facts about an employee. The ability to speak French has nothing to do with an employee's ability to write Java code. Therefore, the HR department must split (decompose) the three-field table into two two-field linking tables: EmplSkills and EmplLangs.

Fifth Normal Form

Fifth normal form involves further reducing redundancy by creating multiple two-field tables from tables that have more than two foreign keys. The classic example is identifying independent sales agents who sell multiple products or categories of products for different companies. In this case, you have a table with AgentID, CompanyID, and ProductID or CategoryID. You can reduce redundancy—at the risk of making the database design overly complex—by creating three two-field tables: AgentCompany, CompanyProduct (or CompanyCategory), and AgentProduct (or AgentCategory). Database developers seldom attempt to normalize designs to fifth normal form because doing so requires adding many additional small tables to the database.

Choosing Primary Key Codes

All Northwind and Oakmont tables use codes for primary key values, as do almost all production databases. The critical requirement is that the primary key value is unique to each record in the table. Following are some tips, many with online resources, to aid in establishing primary key codes:

  • Many types of tables—such as those for storing information on sales orders, invoices, purchase orders, and checks—are based on documents that have consecutive serial numbers, which are obvious choices for unique primary key values. In fact, most database designs begin with collecting and analyzing the paper forms used by an organization. If the table itself or programming code generates the consecutive number, make sure that every serial number is present in the table, even if an order is canceled or voided. Auditors are very suspicious of invoice and purchase order registers that skip serial numbers.

  • Tip - AutoNumber primary key values work well for serially numbered documents if you don't allow records to be deleted. Adding a true-false (Boolean) field named Deleted and setting the value to true is one approach. This technique complicates queries against the tables, so you might consider moving deleted records to another table. Doing this lets you write a query to reconstruct all records for audit purposes.

  • Packaged retail products sold in the United States have a globally unique 10-digit or longer Uniform Product Code (UPC). The UPC identifies both the supplier and the product's SKU. The Uniform Code Council, Inc., (http://www.uc-council.org/) assigns supplier and product ID values, which are combined into linear bar codes for automated identification and data capture (AIDC). The European Article Number (EAN) is coordinated with the UPC to prevent duplication. The UPC/EAN code is a much better choice than Microsoft's serially assigned number for the ProductID field.

  • Books have a 10-digit International Standard Book Number (ISBN) code that's unique throughout the world and, in North America, a UPC. ISBNs include a publisher prefix and book number, assigned to U.S. publishers by the U.S. ISBN Agency (http://www.bowker.com/standards/home/isbn/us/isbnus.html). ISBN Group Agencies assign codes for other countries. Canada has separate agencies for English- and French-language books. Either a UPC or ISBN field is suitable for the primary key of a North American books database, but ISBN is preferred if the code is for books only.

  • The North American Industry Classification System (NAICS, pronounced nakes) is replacing the U.S. Standard Industrial Classification (SIC) for categorizing organizations by their type of business. A six-digit primary key code for 18,000 classifications replaces the four-digit SIC code. Five of the six digits represent codes for classifications common to the United States, Canada, and Mexico. You can view a text file or purchase a CD-ROM of the NAICS codes and their SIC counterparts at http://www.naics.com/.

  • The U.S. Postal Service offers Address Information Systems (AIS) files for verifying addresses and corresponding ZIP/ZIP+4 codes. For more information on these files, go to http://www.usps.com and click the Address Quality link.

  • Social Security Numbers (SSNs) for U.S. residents are a possible choice for a primary key of an Employees table, but their disclosure compromises employee's privacy. Large numbers of counterfeit Social Security cards having identical numbers circulate in the United States, making SSN even less attractive as a primary key field. The Oakmont database uses fictitious nine-digit SSNs for EmployeeID and StudentID fields. Most organizations assign each employee a sequential serial number. Sequential EmployeeID numbers can do double-duty as seniority-level indicators.

Specifying a primary key for tables such as CustPers isn't easy. If you use the five-character code based on first and last names for the primary key, you encounter the problem with potential duplication of CustomerID codes discussed earlier. In this case, however, common last names—Jones, Smith, and Anderson, for example—quickly result in duplicate values. Creating a composite primary key from CustomerID and ContactID is a potential solution; doing this increases the number of new contacts you can add for a company before inevitable duplicates occur. In most cases, it's easier to use an AutoNumber key for all ID values.

Figure 4.9 shows the final design of the modified Northwind database with the added contact details tables. The tables of this database are included on the accompanying CD-ROM as Nwind04.mdb in the \Seua2007\Chaptr04 folder.

Figure 4.9
The final design of the expanded Northwind database with customer and supplier contact details tables added.

The modified Northwind database doesn't qualify as a full-fledged customer relationship management (CRM) system, but the design is sufficiently flexible to serve as the model for a sales and purchasing database for a small-sized wholesale or retail concern.

Maintaining Data Integrity and Accuracy

When you add, modify, or delete table data, it's important that the additions and changes you make to the data don't conflict with the normalization rules that you used to create the database. One of the most vexing problems facing users of large RDBMs is "unclean data." Over time, data entry errors and stray records accumulate to the point where obtaining accurate historical information from the database becomes difficult or impossible. Software vendors and database consultants have created a major-scale "data cleansing" business to solve the problem. You can avoid the time and expense of retroactive corrections to your data by taking advantage of Access and SQL Server features that aid in preventing errors during the data entry process.

Referential Integrity

Maintaining referential integrity requires strict adherence to a single rule: Each foreign key value in a related table must correspond with a primary key value in a base (primary) table. This rule requires that the following types of modifications to data be prevented:

  • Adding a record on the many side of a one-to-many relationship without the existence of a related record on the one side of the relationship (for example, adding a record to the Orders table with a CustomerID value of BOGUS when no such customer record exists in the Customers table)

  • Deleting a record on the one side of a one-to-many relationship without first deleting all corresponding records on the many side of the relationship (for example, deleting Around the Horn's Customers record when the Orders table contains records with AROUT as the CustomerID value)

  • Changing the value of a primary key field of a base table on which records in a related base or linking table depend, such as changing AROUT to ABOUT in the CustomerID field of the Customers table

  • Changing the value of a foreign key field in a linking table to a value that doesn't exist in the primary key field of a base table (for example, changing AROUT to ABOUT in the CustomerID field for OrderID 10355)

Note - You also must avoid changing the primary keys of or deleting one of two tables in a one-to-one relationship.

A record in a related table that doesn't have a corresponding foreign key value in the primary key of a base table is called an orphan record. For example, if the CustomerID value of a record in the Orders table is ABCDE and there's no ABCDE value in the CustomerID primary key field of the Customers table, there's no way to determine which customer placed the order.

Access and SQL Server databases offer the option of automatically enforcing referential integrity when adding or updating data. Cascading updates and deletions are optional. If you specify cascading updates, changing the value of a primary key of a table makes the identical change to the foreign key value in related tables. Cascading deletions delete all related records with a foreign key that corresponds to the primary key of a record in a base table that you want to delete.

→ To learn more about enforcing referential integrity in Access databases, see "Establishing Relationships Between Tables," and "Cascading Updates and Deletions."

Entity Integrity and Indexes

When you add new records to a base table, entity integrity assures that each primary key value is unique. Access and SQL Server ensure entity integrity by adding a no-duplicates index to the field you specify for the primary key. If duplicate values exist when you attempt to designate a field as the primary key, you receive an error message. You receive a similar error message if you enter a duplicate primary key value in the table.

→ For more information on Access indexes, see "Adding Indexes to Tables."

Indexes also speed searches of tables and improve performance when executing SQL statements that return data from fields of base and related tables.

Data Validation Rules and Check Constraints

Data entry errors are another major source of "unclean data." In the days of punched-card data entry, keypunch operators typed the data, and verifiers, who usually worked during the succeeding shift, inserted the cards in a punched-card reader and repeated the keystrokes from the same source document. This process detected typographical errors, which the verifier corrected. Keypunch operators had no visual feedback during data entry, so typos were inevitable; video display terminals didn't arrive until the mainframe era.

Note - Keypunch operators kept their eyes on the source documents, which gave rise to the term heads-down data entry. The term continues in common use to describe any data entry process in which the operator's entire working day is spent adding or editing database records as quickly as possible.

Rekeying data leads to low productivity, so most data entry applications support data validation rules designed to detect attempts to enter illegal or unreasonable values in fields. An example of a validation rule is preventing entry of a shipping date that's earlier than the order date. The rule is expressed as an inequality: ShipDate >= OrderDate, which returns False if the rule is violated. Similarly, UnitPrice > 0 prevents accidentally giving away a line item of an order.

Access tables and fields have a Validation Rule property that you set to the inequality expression. SQL Server calls validation rules check constraints. Both Access and SQL Server have a Validation Text property for which you specify the text to appear in an error message box when the entry violates the rule or constraint. It's a more common practice when working with client/server databases to validate data in the front-end application before sending the entry to the back-end server. Detecting the error on the server and returning an error message requires a roundtrip from the client to the server. Server roundtrips generate quite a bit of network traffic and reduce data entry efficiency. One of the objectives of client/server front-end design is to minimize server roundtripping.

→ To learn more about Access's validation methods, see "Validating Data Entry."


A database transaction occurs when multiple records in one or more tables must be added, deleted, or modified to complete a data entry operation. Adding an order or invoice that has multiple line items is an example of a transaction. If an order or invoice has five line items, but a network or database problem prevents adding one or more item records, the entire order or invoice is invalid. Maintaining referential integrity prevents adding line item records without a corresponding order or invoice record, but missing item records don't violate integrity rules.

Transaction processing (TP), also called online transaction processing (OLTP), solves the missing line item problem. Requiring TP for order entry, invoice processing, and similar multirecord operations enforces an all-or-nothing rule. If every individual update to the tables' records occurs, the transaction succeeds (commits); if any update fails, changes made before the failure occurs are reversed (rolled back). Transaction processing isn't limited to RDBMSs. Early mainframe databases offered TP and transaction monitors. IBM's Customer Information and Control System (CICS, pronounced kicks) was one of the first transaction processing and monitoring systems and remains in widespread use today.

Access and SQL Server databases offer built-in TP features. Access has a Use Transactions property that you set to Yes to require TP for updates. SQL Server traditionally requires writing T-SQL statements—BEGIN TRANS, COMMIT TRANS, and ROLLBACK TRANS—to manage transactions, but Access 2007's ADP forms have a new Batch Updates property that lets you enforce transactions without writing complex T-SQL statements.

Displaying Data with Queries and Views

So far, this chapter has concentrated on designing relational databases and their tables, and adding or altering data. SQL SELECT queries return data to Access, but you don't need to write SQL statements to display data in forms or print reports from the data. Access has built-in graphical tools to automatically write Access SQL for Access databases and T-SQL for SQL Server databases. Access's query tools use a modern implementation of query-by-example (QBE), an IBM trademark. QBE is a simple method of specifying the tables and columns to view, how the data is sorted, and rows to include or exclude.

Note - As mentioned earlier in the chapter, fields become columns and records become rows in a query. This terminology is an arbitrary convention of this book and not related to relational database design theory. The reason for the change in terminology is that a query's rows and columns need not—and often do not—represent data values stored in the underlying tables. Queries can have columns whose values are calculated from multiple fields and rows with aggregated data, such as subtotals and totals.

Linking related tables by their primary and foreign keys is called joining the tables. Early QBE programs required defining joins between tables; specifying table relationships automatically defines joins when you add records from two or more related Access or SQL Server tables.

Figure 4.10 is an example of Access's QBE implementation for Access databases, called Query Design View. You add tables to the query—in this case, Northwind's Customers, Orders, and Employees tables. As you add the tables, join lines indicate the relationships between them. You drag the field names for the query columns from the table lists in the upper pane to the Field row of the lower pane. You also can specify the name of a calculated column (Salesperson) and the expression to create the column values ([FirstName] & " " & [LastName]) in the Field row. The brackets surrounding FirstName and LastName designate that the values are field names.

Figure 4.10
Access's Query Design View for Access databases uses graphical QBE to create queries you can store in the database.

Selecting Ascending or Descending in the Sort column orders the rows in left-to-right column priority. You can restrict the display to a particular set of values by adding an expression in the Criteria column.

Running the query returns the result set, part of which is shown in Figure 4.11. You can save the query for later reuse as a named Access QueryDef(inition) object in the database.

Figure 4.11
These are the first 16 of the 408 rows of the query result set returned by executing the query design of Figure 4.10.

Access SQL - Access QBE automatically converts the query design of Figure 4.10 into the following Access SQL statement:

SELECT Customers.CompanyName, Orders.OrderID, Orders.OrderDate,
         [FirstName] & g g & [LastName] AS Salesperson
   FROM Employees
      INNER JOIN (Customers
         INNER JOIN Orders
         ON Customers.CustomerID = Orders.CustomerID)
      ON Employees.EmployeeID = Orders.EmployeeID
   WHERE ((Year([OrderDate])=2006))
   ORDER BY Customers.CompanyName, Orders.OrderID;

It's obvious that using QBE is much simpler than writing SELECT queries to concatenate field values, join tables, establish row selection criteria, and specify sort order. Access's QBE features are powerful; many developers use Access to generate the SQL statements needed by Visual Basic, C++, and Java programs.

The da Vinci QBE tool for creating T-SQL views is similar to the Access Query Design view, but has an additional pane to display the T-SQL statement as you generate it. You add tables to the upper pane and drag field names to the Column cells of the middle pane. An SQL Server view is the client/server equivalent of an Access QueryDef. As with Access QueryDefs, you can execute a query on an SQL Server view.

Despite their common ANSI SQL-92 heritage, SQL Server won't execute most Access SQL statements, and vice versa. Copying the preceding Access SQL statement to the Clipboard and pasting it into the SQL pane of the query designer for the NorthwindCS sample database doesn't work. The da Vinci designer does its best to translate the Access SQL flavor into T-SQL when you paste, but you receive errors when you try to run the query.

T-SQL - T-SQL uses + rather than & to concatenate strings, uses a single quote (') as the string delimiter, and requires a numerical instead of a string criterion for the YEAR function. Here's the T-SQL version of the preceding Access SQL statement after the SELECT and WHERE clauses have been tweaked:

SELECT TOP (2147483647) dbo.Customers.CompanyName,
      dbo.Orders.OrderID, dbo.Orders.OrderDate,
      dbo.Employees.FirstName + e e +
      dbo.Employees.LastName AS Salesperson
   FROM dbo.Employees
      INNER JOIN dbo.Customers
         INNER JOIN dbo.Orders
         ON dbo.Customers.CustomerID = dbo.Orders.CustomerID
      ON dbo.Employees.EmployeeID = dbo.Orders.EmployeeID
   WHERE (YEAR(dbo.Orders.OrderDate) = 2006)
   ORDER BY dbo.Customers.CompanyName, dbo.Orders.OrderID

The TOP modifier is needed to permit an ORDER BY clause in a view; prior to the addition of the TOP keyword in SQL Server 7.0, creating sorted views wasn't possible. The da Vinci query parser adds the TOP 100 PERCENT modifier if an ORDER BY clause is present. However, TOP 100 PERCENT ... ORDER BY doesn't sort SQL Server 2005 views. Replacing 100 PERCENT with a large integer (<= 2147483647) sorts the view.

The dbo. prefix to table and field names is an abbreviation for database owner, the default owner for all SQL Server databases you create as a system administrator. Figure 4.12 shows the design of the T-SQL query generated by pasting the preceding statement into the da Vinci query pane.

→ For more information on the da Vinci toolset, see "Exploring SQL Server Views."

Figure 4.12
Pasting an Access SQL statement into Access's version of the da Vinci query design tool and making a few minor changes to the T-SQL statement results in an SQL Server view equivalent to the Access query of Figure 4.10.

Note - Access 2007 automatically creates the NorthwindCS database the first time you open the NorthwindCS.adp file after installing SSX from the distribution CD-ROM.

→ For detailed instructions on installing SQL Server Express and NorthwindCS.adp, see "SQL Server 2005 Express Edition SP2 Setup," and "Exploring the NorthwindSQL Sample Project."

The Datasheet View of the SQL Server view generated by the preceding SQL statement is identical to the Access query's Datasheet View shown in Figure 4.11.

In the Real World—When and Why Learn Relational Theory?

A corollary of the Law of Preservation of Matter is, "Everything has to be somewhere." Most books about Microsoft Access deal with relational database design theory in the first few chapters, and this latest of the 11 editions of Special Edition Using Access is no exception. Most previous editions categorized this subject as an "Advanced Access Technique" that appeared toward the end of the book.

Understanding relational database design requires familiarity not only with the objects that make up the database, but with the use of these objects. After you're comfortable with table and query basics, have a feeling for form and report design, and gained an introduction to Web-based database applications, you're probably better prepared to delve into the arcana of relational algebra, such as normalization rules. That's why the previous editions didn't include this chapter in the first part of the book.

After you've gained experience working with the sample relational databases in Access and this book, their design appears intuitive and entirely logical. Experienced database designers envision even the most complex business processes as collections of related tables. So, it's somewhat surprising to many that Dr. Codd's relational database theory originated in 1970, well after the development of complex network and hierarchical architectures. If you're embarking on the road to relational database development, however, you're not likely to find database design topics at all intuitive.

An argument in favor of moving the relational database design topic to the beginning of the book is that many readers decide to use Access to accomplish a specific task that involves creating a special-purpose database. In this case, you advance through the book's chapters using the sample databases as examples rather than learning models. If you're using Access as a learning tool, starting with database design and implementation also makes the structure and relationships of the tables in the sample databases of the book more meaningful.

Increasing sales of desktop and client/server RDBMSs has spawned a multitude of books—ranging from introductory tutorials to graduate-level texts—on relational database design theory and practice. Michael J. Hernandez's Database Design for Mere Mortals (Addison-Wesley Developers Press, ISBN 0-201-75284-0, 2003), subtitled "A Hands-On Guide to Relational Database Design, Second Edition," is an excellent resource for folks who want more than this chapter offers in the way of database design guidance. Mike and John L. Viescas, a well-known Access writer and developer, are co-authors of SQL Queries for Mere Mortals (Addison-Wesley, ISBN 0-201-143336-2), which delivers thorough coverage of SQL SELECT queries for Access and SQL Server users.

If you're serious about getting the most out of Access 2007, consider purchasing a copy of Mike's book or browse the bookstore shelves for titles on relational database design. Your investment will pay handsome dividends when you're able to create the optimum design to start, instead of attempting to restructure a badly designed database after it's grown to 20 or 30 tables containing thousands—or millions—of rows.

Special Edition Using Microsoft Office Access 2007
Roger Jennings
May 1, 2007, 1488 pages
ISBN10: 0-7897-3597-0
Publisher: Que
Buy this book

Mobile Site | Full Site