A data warehouse is a system with its own database. It draws data from diverse sources and is designed to support query and analysis.
With the use of databases over the past decades, large volumes of data have been accumulated. To integrate and manage the data effectively and systematically, data warehouses have emerged. In addition, OLAP and data mining, which use the data warehouse, have become important research topics. OLAP allows users to easily analyze the data in the data warehouse in order to acquire information necessary for decision making.
Data mining extracts unknown useful knowledge from the data warehouse. Data warehousing is a collection of decision making techniques aimed at enabling the knowledge worker to make better and faster decisions. Data warehousing techniques can be classified into three categories: data warehouses, OLAP, and data mining. Research issues in the first are data cleaning, data warehouse refreshment, physical and logical design of a data warehouse, and meta data management.
Research issues in OLAP are multi dimensional data models, OLAP query languages, query processing, and system architectures --- ROLAP (Relational OLAP) using relational databases, MOLAP (Multi dimensional OLAP) using multi dimensional indexes, and HOLAP (Hybrid OLAP) combining ROLAP and MOLAP.
Data mining involves various techniques such as association rules, classification, clustering, and similarity search.
The objective of data warehousing is to analyze data from diverse sources to support decision making. To achieve this goal, we face two challenges:
- Poor system performance.
A data warehouse usually contains a large volume of data. It is not an easy job to retrieve data quickly from the data warehouse for analysis purposes. For this reason, the data warehouse design uses a special technique called a star schema
- Difficulties in extracting, transferring, transforming, and loading (ETTL) data from diverse sources into a data warehouse.
Data must be cleansed before being used. ETTL has been frequently cited as being responsible for the failures of many data warehousing projects. You would feel the pain if you had ever tried to analyze SAP R/3 data without using SAP BW.
SAP R/3 is an ERP (Enterprise Resources Planning) system that most large companies in the world use to manage their business transactions. Before the introduction of SAP Business Warehouse in 1997, ETTL of SAP R/3 data into a data warehouse seemed an unthinkable task. This macro-environment explained the urgency with which SAP R/3 customers sought a data warehousing solution.
The result is SAP
Business Warehouse born from SAP, the developer
of SAP R/3.
The basic concept of data warehousing.
First, we use sales analysis as an example to introduce the basic concept of data warehousing.
Sales Analysis - A Business Scenario
Suppose that you are a sales manager, who is responsible for planning and implementing sales strategy. Your tasks include the following:
- Monitoring and forecasting sales demands and pricing trends
- Managing sales objectives and coordinating the sales force and distributors
- Reviewing the sales activities of each representative, office, and region
In the real world, you might have years of data and millions of records.
To succeed in the face of fierce market competition, you need to have a complete and up-to-date picture of your business and your business environment. The challenge lies in making the best use of data in decision support.
In decision support, you need to perform many kinds of analysis. This type of online analytical processing (OLAP) consumes a lot of computer resources because of the size of data. It cannot be carried out on an online transaction processing (OLTP) system, such as a sales management system. Instead, we need a dedicated system, which is data warehouse such as the SAP Business Warehouse.
