Information Technology Reference
In-Depth Information
To reduce telecommunications costs, some organizations build a replicated database. A
replicated database holds a duplicate set of frequently used data. The company sends a copy
of important data to each distributed processing location when needed or at predetermined
times. Each site sends the changed data back to update the main database on an update cycle
that meets the needs of the organization. This process, often called data synchronization , is
used to make sure that replicated databases are accurate, up to date, and consistent with each
other. A railroad, for example, can use a replicated database to increase punctuality, safety,
and reliability. The primary database can hold data on fares, routings, and other essential
information. The data can be continually replicated and downloaded on a read-only basis
from the master database to hundreds of remote servers across the country. The remote
locations can send back the latest figures on ticket sales and reservations to the main database.
replicated database
A database that holds a duplicate set
of frequently used data.
Online Analytical Processing (OLAP)
For nearly two decades, multidimensional databases and their analytical information display
systems have provided flashy sales presentations and trade show demonstrations. All you
have to do is ask where a certain product is selling well, for example, and a colorful table
showing sales performance by region, product type, and time frame appears on the screen.
Called online analytical processing (OLAP) , these programs are now being used to store and
deliver data warehouse information efficiently. The leading OLAP software vendors include
Microsoft, Cognos, SAP, Business Objects, MicroStrategy, Applix, Infor, and Oracle.
Lufthansa Cargo depends on OLAP to deliver up-to-the-minute company statistics that help
the company compete in the growing global air-freight market. 30 The market is growing by
six percent annually, and competitors are emerging all around the world to get a piece of the
action. Lufthansa Cargo uses OLAP to analyze its data to provide the fastest service to its
customers and the lowest rates.
The value of data ultimately lies in the decisions it enables. Powerful information-analysis
tools in areas such as OLAP and data mining, when incorporated into a data warehousing
architecture, bring market conditions into sharper focus and help organizations deliver
greater competitive value. OLAP provides top-down, query-driven data analysis; data mining
provides bottom-up, discovery-driven analysis. OLAP requires repetitive testing of user-
originated theories; data mining requires no assumptions and instead identifies facts and
conclusions based on patterns discovered. OLAP, or multidimensional analysis, requires a
great deal of human ingenuity and interaction with the database to find information in the
database. A user of a data-mining tool does not need to figure out what questions to ask;
instead, the approach is, “Here's the data, tell me what interesting patterns emerge.” For
example, a data-mining tool in a credit card company's customer database can construct a
profile of fraudulent activity from historical information. Then, this profile can be applied
to all incoming transaction data to identify and stop fraudulent behavior, which might oth-
erwise go undetected. Table 5.9 compares the OLAP and data-mining approaches to data
analysis.
online analytical processing
(OLAP)
Software that allows users to
explore data from a number of
perspectives.
Table 5.9
Data Mining
Characteristic
OLAP
Comparison of OLAP and Data
Mining
Purpose
Supports data analysis
and decision making
Supports data analysis and
decision making
Type of analysis
supported
Top-down, query-driven
data analysis
Bottom-up, discovery-driven data
analysis
Skills required
of user
Must be very knowledgeable
of the data and its business
context
Must trust in data-mining tools to
uncover valid and worthwhile
hypotheses
 
 
Search WWH ::




Custom Search