From Database to Data WarehouseUnlike data warehouses, transactional databases are usually not suitable for business analysisbecause they contain only current, not historical, data. Often, data in transactional databases arealso scattered in different systems throughout an organization. The same data can be storeddifferently and under other names. For example, customer names might be recorded in a columncalled Name in one table and in two columns—First Name and Last Name—in another table.These discrepancies commonly occur when an organization uses both its own data and data itpurchases from other organizations, or if it has developed more than one database that containsthe same data under a different label. When management decides to build a data warehouse, theIT staff must carefully consider the hardware, software, and data involved in the effort.The larger the data warehouse, the larger the storage capacity, the greater the memory, andthe greater the processing power of the computers that are needed. Because of capacity needs,organizations often choose mainframe computers with multiple CPUs to store and manage datawarehouses. The computer memory must be large enough to allow processing of huge amountsof data at once. The amount of storage space and the access speed of disks are also important.Processing millions of records might take a long time, and variations in disk speed might meanthe difference between hours or minutes in processing time. And since a data warehouse is
considered a highly valuable asset, all data must be automatically backed up. Keep in mind that data warehouses grow continually, because their very purpose is to accumulate historical records.
Retail chains such as Wal-Mart and Costco record millions of sales transactions daily, all of which
are channeled into data warehouses. Some have data warehouses that hold tens or hundreds of
terabytes of data. In addition to retailers, banks, credit-card issuers, health-care organizations,
and other industries have augmented their hardware for large data warehouses. Many organiza
tions accumulate not only sales transactions but also purchasing records, so they can produce
information from which to make better purchasing decisions, such as which suppliers tend to
offer lower prices for certain items at certain times of the year.
đang được dịch, vui lòng đợi..