With the rapid development of the Internet technology, the total data volumn ofInternet has increased from TB to PB level, and is still growing dramatically. Therefore,how to deeply analyze these big data and put forward more effective and more generaldistributed computing model, becomes the key point in the field of big data processing.This paper introduces parallel computing logic which is used for solvingmulti-objective decision-making and dynamically feedback problems according to boththe requirement mentioned above and some practical issues, such as data-intensiveinformation measurement. Moreover, a distributed computing model named MTDF(Multiobject Targets and Dynamiclly Feedback) has been proposed based onMapReduce, HBase, etc. With respect to the calculation and resource managementefficiency of current models, this model puts forward some solutions, which enabledevelopers to fully utilize available resources in the system. Finally, an stock trendprediction experiment illustrates how to develop applications based on this calculatingmodel, and the results show the efficiency and feasibility of this model. |