- The DBA: Revered or Reviled?
- Why Learn Database Administration?
- The Management Discipline of Database Administration
- Database, Data, and System Administration
- DBA Tasks
- Types of DBAs
- Staffing Considerations
- Multiplatform DBA Issues
- Test and Production
- New Technology and the DBA
- DBA Certification
- The Rest of the Book
Database, Data, and System Administration
Some organizations define separate roles for the business aspects and the technical aspects of data. The business aspects of data are aligned with data administration, whereas the more technical aspects are handled by database administration. Not every organization has a data administration function. Indeed, many organizations combine data administration into the database administration role.
Sometimes organizations also split up the technical aspects of data management, with the DBA responsible for using the DBMS and a system administrator or systems programmer responsible for installing and upgrading the DBMS.
Data administration separates the business aspects of data resource management from the technology used to manage data; it is more closely aligned with the actual business users of data. The data administrator (DA) is responsible for understanding the business lexicon and translating it into a logical data model. Referring back to the ADLC, the DA would be involved more in the requirements gathering, analysis, and design phase, the DBA in the design, development, testing, and operational phases.
Another difference between a DA and a DBA is the focus of effort. The DA is responsible for the following tasks:
Identifying and cataloging the data required by business users
Producing conceptual and logical data models to accurately depict the relationship among data elements for business processes
Creating an enterprise data model that incorporates all of the data used by all of the organization's business processes
Setting data policies for the organization
Identifying data owners and stewards
Setting standards for control and usage of data
In short, the DA can be thought of as the Chief Data Officer of the corporation. However, in my experience, the DA is never given an executive position, which is unfortunate. Many IT organizations state that they treat data as a corporate asset, a statement that is belied by their actions. Responsibility for data policy is often relegated to technicians who fail to concentrate on the nontechnical business aspects of data management. Technicians do a good job of ensuring availability, performance, and recoverability, but are not usually capable of ensuring data quality and setting corporate policies.
In fact, data is rarely treated as a true corporate asset. Think about the assets that every company has in common: capital, human resources, facilities, and materials. Each of these assets is modeled: charts of account, organization charts, reporting hierarchies, building blueprints, office layouts, and bills of material. Each is tracked and protected. Professional auditors are employed to ensure that no discrepancies exist in a company's accounting of its assets. Can we say the same thing about data?
A mature DA organization is responsible for planning and guiding the data usage requirements throughout the organization. This role encompasses how data is documented, shared, and implemented companywide. A large responsibility of the DA staff is to ensure that data elements are documented properly, usually in a data dictionary or repository. This is another key differentiation between a DA and a DBA. The DA focuses on the repository, whereas the DBA focuses on the physical databases and DBMS.
Furthermore, the DA deals with metadata, as opposed to the DBA, who deals with data. Metadata is often described as data about data; more accurately, metadata is the description of the data and data interfaces required by the business. Data administration is responsible for the business's metadata strategy. Examples of metadata include the definition of a data element, business names for a data element, any abbreviations used for that element, and the data type and length of the element. Data without metadata is difficult to use. For example, the number 12 is data, but what kind of data? In other words, what does that 12 mean? Without metadata, we have no idea. Consider this: Is the number 12
A date representing December, the twelfth month of the year?
A date representing the twelfth day of some month?
A shoe size?
Or, heaven forbid, an IQ?
And so on. However, there are other, more technical aspects of metadata, too. Think about the number 12 again.
Is 12 a large number or a small one?
What is its domain (that is, what is the universe of possible values of which 12 is but a single value)?
What is its data type? Is it an integer or a decimal number with a 0 scale?
Metadata provides the context by which data can be understood and therefore become information. In many organizations, metadata is not methodically captured and cataloged; instead, it exists mostly in the minds of the business users. Where it has been captured in systems, it is spread throughout multiple programs in file definitions, documentation in various states of accuracy, or in long lost program specifications. Some of it, of course, is in the system catalog of the DBMS.
A comprehensive metadata strategy enables an organization to understand the information assets under its control and to measure the value of those assets. Additional coverage of metadata is provided in Chapter 21.
One of the biggest contributions of data administration to the corporate data asset is the creation of data models. A conceptual data model outlines data requirements at a very high level. A logical data model provides in-depth details of data types, lengths, relationships, and cardinality. The DA uses normalization techniques to deliver sound data models that accurately depict the data requirements of an organization.
Many DBAs dismiss data administration as mere data modeling, required only because someone needs to talk to the end users to get the database requirements. However, a true DA function is much more than mere data modeling. It is a business-oriented management discipline responsible for the data asset of the organization.
Why spend so much time talking about data administration in a book about database administration? Well, very few organizations have implemented and staffed a DA role. The larger the organization is, the more likely that a DA function exists. However, when the DA role is undefined in an organization, the DBA must assume the mantle of data planner and modeler. Unfortunately, the DBA will usually not be able to assume all of the functions and responsibility of a DA as summarized in this section for a number of reasons:
The DBA has many other technical duties to perform that will consume most of his time.
The manager of the DBA group typically does not have an executive position enabling him to dictate policy.
The DBA generally does not have the skills to communicate effectively with business users and build consensus.
Frankly, most DBAs are happier dealing with technical issues and technicians than with business issues and nontechnicians.
When DA and DBA functions coexist within the organization, the two groups must work very closely with one another. It is not necessary that both have the same manager, though it would facilitate cooperation. At any rate, it is imperative that some skills cross-pollinate the two groups. The DA will never understand the physical database like a DBA, and the DBA will never understand the business issues of data like a DA, but each job function is more effective with some knowledge about the other.
In short, organizations that are truly concerned about data quality, integrity, and reuse will invariably implement and staff the DA function.
Database administration is the focus of this entire book, so I will not spend a lot of time defining it in this short section. The rest of the book will accomplish that nicely. This section will quickly outline the functions performed by the DBA group when the DA function exists. The first duty of the DBA is to understand the data models built by the DA and to communicate the model to the application developers and other appropriate technicians. The logical data model is the map the DBA will use to create physical databases. The DBA will transform the logical data model into an efficient physical database design. It is essential that the DBA incorporate his knowledge of the DBMS to create an efficient and appropriate physical database design from the logical model. The DBA should not rely on the DA for the final physical model any more than a DA should rely on a DBA for the conceptual and logical data models. Figure 1-4 depicts this relationship.
Figure 1-4 DBA vs. DA
The DBA is the conduit for communication between the DA team and the technicians and application programming staff. Of course, the bulk of the DBA's job is ongoing support of the databases created from the physical design and management of the applications that access those databases. An overview of these duties is provided in the DBA Tasks section of this chapter.
Some organizations, usually the larger ones, also have a system administrator (SA) or systems programming role that impacts DBMS implementation and operations. The SA is responsible for the installation and setup of the DBMS. The SA typically has no responsibility for database design and support. Instead, the DBA is responsible for the databases and the SA is responsible for DBMS installation, modification, and support. (If this distinction is not clear to you, please refer to Appendix 1.)
Furthermore, the SA ensures that the IT infrastructure is implemented such that the DBMS is configured to work with other enabling system software. The SA may need to work with other technicians to configure transaction processors, message queueing software, networking protocols, and operating system parameters to enable the DBMS to operate effectively. The SA ensures that the IT infrastructure is operational for database development by setting up the DBMS appropriately, applying ongoing maintenance from the DBMS vendor, and coordinating migration to new DBMS releases and versions.
As with data administration, there must be cross-training of skills between the SA and DBA. The SA will never understand the physical database like the DBA, but the DBA is unlikely to understand the installation and in-depth technical relationships of system software like the SA. Each job function will be more effective with some knowledge of the other.
If no system administration group exists, or if its focus is not on the DBMS, the DBA assumes responsibility for DBMS-related system administration and programming. Figure 1-5 delineates the responsibilities of the DA, the DBA, and the SA.
Figure 1-5 DA, DBA, and SA responsibilities