10.3 Types of Metadata

Team-Fly

---

 

	Internet-Enabled Business Intelligence By William A. Giovinazzo
	Table of Contents
	Chapter 10.  Common Warehouse Metadata

---

10.3 Types of Metadata

The example above raises an interesting point. If a number is a price, commission, or salary, we have a certain understanding of the context. We fill this in on our own. In the business world, much of the data's context is often left undocumented. It is assumed that the user will understand what is meant by the data. Often, system engineers and administrators content themselves with only a partial description of the data, the format. If we truly want to capture the context of the data contained within a system, be it a data warehouse or an operational system, we need to expand our view. Figure 10.3 presents a mapping of the many different types of metadata required for a complete view of the data's context.

Figure 10.3. Types of metadata.

The following list provides a more complete description of each metadata type:

• Static

  • Name Provides the name by which the data element is known to the system. For example, Employee_Name, Customer_Name, and Customer_ID.

  • Description Provides a full text description of the data element.

  • Format Provides the data presentation rules.

  • Data Type Defines the data that is stored within the data element. For example, integer, floating-point, and Boolean.

  • Relation Defines the relationships between objects within the system. For example, Customers buy product.

  • Domain Provides the domain or range of valid values.

  • Business Rules Provides the rules of the organization that govern the data element.

• Dynamic

  • Quality Describes the quality of the data within the system, such as the accuracy, completeness, consistency, and validity. An example of quality metadata within the data warehouse is the extraction log, which is discussed in Chapter 3. This log provides statistics on the data loaded into the data warehouse;

  • Statistical Describes usage and administrative characteristics of the system. Maintaining statistics on which data is most frequently accessed or the types of analysis performed on the data is extremely useful to systems developers in enhancing the performance of the system. It is also common to keep statistics on which users access the system, how often it is accessed, and for how long;

  • Status This metadata keeps track of the general health of the system. It is also beneficial to keep a record of backup statistics such as when a backup was last performed, how long the backup took, and what errors occurred. It is also helpful to keep track of disk utilization, system failures, Mean Time Between Failure (MTBF) and Mean Time To Repair (MTTR).

Figure 10.3 shows that there are two basic types of metadata: static and dynamic. Static metadata describes the structure that contains the data, and dynamic metadata describes the overall status of that structure. One of the most critical, yet least documented, static metadata types is the business rules. It is often assumed these rules are known by the user. With business rules, however, even the most obvious data elements are not always clear. What is a car? What distinguishes one car from another? Looking at two different cars , you could distinguish between car A and B. What if I take parts from car B and use them as replacement parts for car A? How many parts moved from A to B will it take to make A into B? This may look like a ridiculous example, but this question goes back to the time of Socrates. He asked these very same questions of a boat. If we continually replace the parts of a boat, when does the old boat become a new one? It is a very real issue. In the case of a car, the business rule is simple: The car's serial number is attached to the dashboard under the windshield . To whatever car A's dashboard is attached is car A. Although the rule is simple, it is not necessarily known unless documented.

Obviously, there will be changes to business rules, so one might object to the inclusion of business rules as static metadata. Static does not mean to imply the metadata never changes, but simply that it changes less frequently than dynamic metadata. The dynamic metadata primarily relates to the state and use of the data. Dynamic metadata has an operational flavor. It reflects the changing state of the system. Consider the different types of dynamic metadata. Dynamic metadata refers to such data as backup, data usage, space usage, and user access. All of this data is constantly changing.

---

Team-Fly

Top