Research On Business Intelligence Framework Based On Grid Computing

Withthe continuous development of science and technologyin recent years, thereare higher demands on the computing power of computers in many areas. For somelarge-scale scientific computing, it is not possible to achieve desired effects to relysolely on one or several ultra-high-performance computers. Under this circumstance,grid computing technology emerges and is developing rapidly. The goal of grid is toconnect and integrate through high speed internet variety of high performancecomputers, data servers, large-scale retrieval and storage systems, visualization andvirtual reality systems which are geographically and organizationally distributed andheterogeneous, to complete some major scientific researches or commercialapplicationsandeventuallyeliminateinformationandresourceisolatedislands.No matter whether grid computing technology is applied to the field of scientificresearch or it is applied to commercial applications, the introduction of businessintelligence system is necessary; In scientific researches, grid computingtechnologyismainly used for large-scale scientific computing, if the data generated in scientificcomputing is integrated by business intelligence systems, analyzed and visuallyshowed, it will undoubtedly help to improve the algorithms in scientific research,shorten the cycle of research and reduce the cost of IT. In commercial applications,grid computing is mainly used to collect resource information which is geologicallydistributed and may even come from different organizations. The collected data arethen converted into knowledge through OLAP or data mining technology, theknowledge can in turn be provided to the decision-makers in commercialdecision-making and let enterprises adapt to the increasing competitive marketenvironment.This paper studies the design and realization of business intelligence frameworkunder a grid operating system environment: Platform EGO. To improve theperformance of the data integration module in the existing BI: PERF, this paperproposes a flexible and generic Shell script-based data loading model. By providingunified programminginterfaces, the proposed data loadingmodel significantlyreducesthe time for developing new data loaders; it also reduces some intermediate stepsbetweenthedevelopmentandthedeploymentofdataloadersandhencereducesrisks.The design and implementation of this Shell scripts based generic data loader effectively reduces the requests time for enterprise users to load the new data, it helpsto improve the efficiency for enterprise user to convert data to knowledge, andprovides timely business decisions making. It is undoubtedly of great commercialvalueinanerainwhencompetitionsbetweenenterprises arebecomingmoreandmorefierce.