【正文】 se as well as database schemas, queries, and outputs via the Web or Excel for 60% to 90% of a typical project39。s decision making process . A more technical definition might be: the subset of a Business Intelligence tool set responsible for modeling, structuring, storing as well as extraction translation and loading (ETL) of the underlying data needed for analysis. So in summary, Business Intelligence software is the collection of applications needed to make sense of business data. The Data Warehouse, a ponent of this Business Intelligence tool set, is the more specific tool responsible for the cleanup, loading, and storage of the data needed by the business. Although we will address the overall BI tool set in the next lesson, this class focuses on the Data Warehouse ponent. A Data Warehouse can help to organize the data. It brings together all operative DataSources (these are mostly heterogeneous and have differing degrees of detail). The job of the warehouse is to provide this data in a usable form to the whole organization. The data can then be used for future requirements as the need arises. A warehouse has the following properties: 重慶 郵電 大學(xué)本科學(xué)生畢業(yè)設(shè)計(jì) （論文）附件 附件 C：譯文 C11 ? Readonly access: Users have readonly access, meaning that the data is primarily loaded into the Data Warehouse via the extraction, transformation and loading (ETL) process. ? Crossorganizational focus: DataSources from the entire organization (production, sales and distribution, controlling), and possibly external sources, make up the basis of the system. ? Data Warehouse data is stored persistently over a particular time period. ? Data is stored on a longterm basis. ? Designed for efficient query processing: The technical environment and data structures are optimized to answer business questions . not to quickly store transactions. R. Kimball, another guru of Data Warehousing, defines a Data Warehouse as .A copy of transaction data, specially restructured for queries and analyses.. (The Data Warehouse Toolkit, 1996, page 310). Business Intelligence Systems Objectives A modern Business Intelligence system must meet the following requirements: Standardized structuring and display of all business information: Decision makers urgently need reliable information from the production, purchasing, sales and distribution, finance, and human resources departments. They require an uptodate and prehensive picture of each individual business area and of the business as a whole. This results in high demand being put on the data collection process from the underlying DataSources. The data is defined uniquely across the entire organization to avoid errors arising through varied definitions in different sources. Simple access to business information via a single point of entry: Information must be bined homogeneously and consistently at a central point from which it can be called up. For this reason, modern Data Warehouses usually require a separate database. This database enables a standalone application environment to provide the required services. Highly developed reporting for analysis with self service for all areas: In terms of presentation, efficient analysis and meaningful multimedia visualization techniques are essential. The system must be able to cope with the information needs of varied user groups. Quick and costefficient implementation: When implementing the Data Warehouse, an influential cost factor is its integration into an OLTP system and the straightforward 重慶 郵電 大學(xué)本科學(xué)生畢業(yè)設(shè)計(jì) （論文）附件 附件 C：譯文 C12 loading of heterogeneous data. Alongside robust metadata management, delivered businessbased Business Intelligence content also has an important role here. High performance environment. Data modeling from heterogeneous sources: Data analyses can not be carried out via Data Warehouse without integrating heterogeneous sources. This is usually done with timeconsuming read processes. Scheduling tools are necessary to allow the data to be loaded in separate batch jobs at performancefriendly times. Relieving OLTP systems: In the past, OLTP systems were strongly overloaded by having to store data and analyze it at the same time. A separate Data Warehouse server now allows you to carry out data analysis elsewhere. Differences Between a BI/Data Warehouse System and an OLTP System ? Level of detail: The OLTP layer stores data with a very high level of detail, whereas data in the Data Warehouse is pressed for highperformance access (aggregation). ? History: Archiving data in the OLTP area means it is stored with minimal history. The Data Warehouse area requires prehensive historical data. ?Changeability: Frequent data changes are a feature of the operative area, while in the Data Warehouse, the data is frozen after a certain point for analysis. ? Integration: In contrast to the OLTP environment, requests for prehensive, integrated information for analysis is are very high. ? Normalization: Due to the reduction in data redundancy, normalization is very high for operative use. Data staging and lower performance are the reasons why there is less normalization in the Data Warehouse. ?Read access: An OLAP environment is optimized for read access. Operative applications (and users ) also need to carry out additional functions regularly, including change, insert, and delete. There are fundamentally different demands on an OLTP system pared with a Data Warehouse/ BI (OLAP) system. It is therefore most advantageous to technically separate all aggregated reportingrelated demands made on the Data Warehouse from the OLTP system. Note: Developments in technology and specific business cases are blurring the lines be