Database administration is simply maintaining records of any type—customer lists, vendor histories, or addresses, for example—using computer software known as a database management system (DBMS). As Anne Kerven said in Colorado Business Magazine, "Database management means the transferring of file cabinet contents to an electronic file." Out of that simple statement has sprung a multimillion dollar computer software industry and a thriving database administration consulting niche. Almost every company has records of one type or another to maintain, which means that almost every company is affected by DBMS in some way or another.
Databases can range in size from a few hundred addresses maintained on a user's hard drive to hundreds of terabytes of data maintained on huge corporate mainframe computers. One of the benefits of using a database management system, however, is that even if the data is vast and is stored on a remote mainframe, end-users throughout a company can all access the data from their desktop using computer networking technology. Reports that in the past had to be requested days or even weeks in advance and created by computer technicians can be generated in minutes by the average user with today's database management systems.
The most common type of system is the relational database management system (RDBMS). It is found in almost every company data center. An RDBMS sorts data into unique fields and allows users to retrieve that data by each field and by linking fields between related records. Relational databases can sort the fielded data any number of ways and generate reports in a matter of minutes. Data can often be output in any form the end-user desires. In addition, a RDBMS can serve as the front-end program that brings data together from several individual databases and produces data tables that combine the information from the various databases.
Database management technology is improving every year, however, and relational databases are starting to be replaced by more sophisticated database management systems. This movement was spurred in large part by companies that realized that they had more than simple records to maintain—they had complicated files, with sounds and images; they had brochures, photographs, time-series inputs, and 3-D coordinates—all of which could be more easily maintained if they were organized and stored in a database. In response to this growing need, software developers have created new object-oriented database management systems (OODBMS) and object-relational database management systems (ORDBMS).
It is expected that ORDBMS will become the most popular type because the need to store disparate types of information is growing. An example of the type of data that might be stored in an object-relation system is a human resources file on an employee. In the past, the database record might have only included text information about the employee—birth date, address, starting date, etc. With an object-relational system, the record could also include the employee's photo or voice sample. Or, a company could maintain geospatial information that would allow it to query the database to locate all customers who made more than $50,000 and lived within 10 miles of the company's location.
SELECTING A SYSTEM FOR A SMALL BUSINESS
For small business owners who are entertaining thoughts about purchasing a database management system, experts say that the first thing they need to do is determine what they hope to get out of the system—what type of reports do they need, etc. Once the output is known, it is easier to know what type of database is needed, what information will be gathered, and what fields will be created. It is a good idea to start small—such as with a mailing list—to get used to the software. Once the first database is mastered, it is easy to set up additional ones for order tracking, inventory, or other purposes.
Most databases are one of two types—transactional or warehouse. Transactional databases are easier to build and are ideal for tracking simple things, such as the availability of a product or part. Warehouse databases collect company data of any type, such as sales histories or hiring statistics, and produce reports that can identify trends or group information in new and relevant ways. Small businesses use both types of databases.
Once you know what you hope to get out of a system and what type of database you will be building, you can move on to selecting the right software for the job. If you are a computer savvy user who is experienced with personal computers, and the database is simple, then you may be able to study the available software packages yourself and choose the one that is best for you. If you are not, or you expect the job to be complex, it is best to locate a consultant in your area and work with them to select your system.
The simpler the system you can get away with, the better. "If you can buy off-the-shelf, that's fantastic because it's less expensive," said Larry Skaff, owner of Junction Software Services, in Colorado Business Magazine. "But it will have tradeoffs. If they don't do business exactly the way that database is written, can they live with that?" The most common off-the-shelf system is Microsoft Access, which is sold separately and as part of Microsoft Office and Microsoft Small Office. Access allows users to build databases out of a number of templates so that they do not have to start from scratch. At the same time, it does allow advanced users to build custom databases, and it features fairly powerful sort and reporting options.
More adventurous users may opt for packages that are a notch or two above Access, such as Filemaker Pro, Borland's Paradox, or Microsoft's FoxPro. These are more robust than Access and allow users to define fields and create intricate databases. If you do not feel comfortable creating a database yourself but still need the additional features that these higher-level packages provide, then a consultant is once again the place to turn. Most will charge anywhere from $50 to $200 per hour to create a custom database. Be warned—this can be a time-consuming project and can cost tens of thousands of dollars.
For small businesses that have even more extensive database needs, consultants and computer professionals will undoubtedly be required. High-level products from one of the big three database management systems companies—Oracle, Informix, or Sybase—can manage huge amounts of complex data on either a stand-alone machine or a local-or wide-area network.
MAINTAINING A DATABASE MANAGEMENT SYSTEM
There is one important fact about databases that some people seem to forget—no matter how good the software is, no matter how expensive the computers are, they are only as good as the data that is put into them. Information must be loaded into the system via data entry work or some other form of input, and it is up to the business owner or manager of information systems to make sure that records are accurate and kept up to date.
Letting records slip from time to time may seem like a small thing, but especially for small business enterprises, poor database maintenance can reflect poorly on the business and make clients think twice about doing business there. Database managers estimate that more than half of small businesses do not maintain their databases once they are created. Examples of the kinds of mistakes that can occur include failing to bill an account, mailing literature to someone who is deceased, or indicating that there is plenty of a particular product in stock when in fact the supply was exhausted weeks ago. Such situations can give rise to business disasters for owners of small businesses.
Writing in Colorado Business Magazine, Anne Kerven outlined some of the other common maintenance mistakes that are made:
* Collecting too much or too little data. Collecting too much information slows down the system, clutters screens with unnecessary fields, and inflates the costs of gathering data. Recording too little data can render the database worthless for compiling reports that can help the business grow; instead, money spent on inputting the little data that is there is wasted.
* Poorly conceived data fields. The most common mistake is putting too much information in one field—the computer can only sort by field, not by what is in the field. The best rule of thumb is to put each unique record element—ZIP code, phone number, fax number, address—in its own searchable field.
* Try to avoid using personal names as the key identifier of a record or as a link between records. Instead, use numbers, assigning a unique number to each record. Personal names cause problems when two or more people have the same name; additionally, if the name is not entered exactly the same way every time it is used as a link—a middle initial is included in one instance and left out in another, for example—the records will not link properly.
* Check the database integrity at least once a month. Corrupted links or other problems can creep in over time. Utility programs are available for this function.
* Back up information and store it in a separate location, preferably someplace that is fireproof and waterproof.
* Set strict standards that must be followed whenever data is input into the system to ensure consistency. This is especially important if multiple departments will be adding data to the system.
* Periodically clean out the database, weeding out records that are inactive or no longer relevant. If you do not want to lose those records permanently, create an archive database and move the records into that file.
COMMON USES OF DATABASE MANAGEMENT SYSTEMS
As discussed, database management systems store company information and allow users to easily retrieve that information. But what does that mean in the business world? How exactly are database management systems being put to use so that companies get the most bang for their buck? Currently, there are two primary uses that are gaining in popularity—data marts and data warehouses, and the use of DBMS together with a company's Internet site or intranet.
DATA MARTS AND DATA WAREHOUSES Data marts and data warehouses refer to the information repositories that companies create with their database management software. Data marts are simply smaller versions of data warehouses, storing information on a department-by-department basis. Data warehouses are huge, centralized databases that unify information across an entire company. These huge databases can be used to improve customer service, profitability measurement, and product sales.
Data marts gained popularity before data warehouses. They were seen as a way for departments to achieve one of their main goals—getting information into the hands of all their users quickly and at the same time. However, as DBMS technology improves and larger databases become possible, the flaws in using data marts are being exposed. Data marts do not unify data across an organization—in fact, they can fragment it because every department might be doing things a little differently. Each department's data becomes an island that yields different answers to the same query.
That is not to say that data marts cannot work, however. They can, if they are built after a main data warehouse is built. Most people think it is better to start small with data marts and build up to the big data warehouse, but many experts contend that the opposite is true. If the data marts are built first, then fragmented data held in uniquely structured databases that cannot be accessed by all employees are created.
Instead, small business owners should build the warehouse first. Look at the types of information that are gathered in each department around the company and select the key data from each area. Use this as the starting point for the warehouse. Do not try to build an all-purpose warehouse right from the start. Do what makes sense, then add historical data and other information as time goes by. If information is simply gathered and stored with no allowance made for cross-departmental analysis, then the warehouse is useless. Those considering building a database should understand that if, upon examination, the current processes a company uses do not allow for such cross-departmental analysis, then fundamental changes will have to be made to those processes if the warehouse is to work. This is a significant point that too many managers or business owners do not understand when they decide to build a warehouse.
Small businesses also have to make sure that endusers from every department are actively involved in the design of the data warehouse. Without that type of feedback, the database may turn out to be useless because it does not store the right type of information, or it stores it in the wrong way. Those end-users involved in the design can learn how to use the system first and then serve as trainers in their department. Once the main warehouse is built, it then becomes easy for each department to build its own data marts by pulling out the fields from the main database that it needs.
The initial costs of building a data warehouse are high—software, hardware, and consulting fees add up quickly. However, most businesses, from supermarkets to banks, are finding that having a data warehouse is a competitive necessity. One example of how data warehouses are being used is a practice called "data mining." Data mining is the technique of creating statistical and predictive models of the real world based on patterns that are discovered as a result of complex queries performed on the huge amounts of data stored in a warehouse. When data mining is done right, it can produce amazing results, spotting trends before they happen or identifying new sales prospects, for instance. When done incorrectly, however, data mining can produce false correlations and misleading results. Companies should not rely too heavily on data mining and should ensure that they hire professionals who fully understand statistical analysis to perform the task.
DATABASE MANAGEMENT AND THE INTERNET OR INTRANETS
Data warehouses started out as internal projects, but now they are being seen as the next logical step on the Internet. As one analyst told Info World, "In about two years, 80 percent of queries will be run across the Web." Companies that need to gather data from customers and pass information down the line to business customers are finding it beneficial to make their data warehouse available over the World Wide Web. This means that either HTML or Java-based client servers need to be created to allow Web users to search the database. If the company desires to gather information on customers, it might make the warehouse available to the public over the Web. If the main purpose is to pass information on to business customers, then the company will make the warehouse available as part of its corporate intranet, which is available only to selected individuals.
At first, only basic queries could be run easily over the Web, but observers note that the situation has changed rapidly. Each of the major database management systems companies has scrambled to enable their databases to work closely with Web servers. Using new technology known as online analytical processing (OLAP), high-level, intricate queries of data warehouses will be possible. At the same time, consumers looking to data mine corporate information should be able to run simple queries.
There are risks associated with making such huge amounts of data available over the Web. Security is the paramount issue, since opening data warehouses to users around the world means that internal systems are exposed to outside interference or hacking. A second issue is the drain on system resources that unlimited access to the data warehouses would cause. Popular databases visited and searched by large numbers of users would need extremely powerful servers to keep up with demand. The servers would have to ensure that the employees and clients of a company would not be hindered by the excess traffic on the system. Finally, there is the issue of cost. As with any new technology, opening databases to the Web costs money. In addition to development costs associated with creating the search engines and OLAP tools, businesses will also have to weigh the cost of the powerful servers needed to meet the increased demand for information.