¡¡	Chinese Journal of Computers   Full Text
Title	Notes on Classifying Network Computing Systems
Authors	XU Zhiª²Wei LIAO Huaª²Ming YU Haiª²Yan ZHA Li
Address	(Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190)
Year	2008
Issue	No.9(1509¡ª1515)
Abstract &   Background	Abstract Classifying computing systems is a basic component of computer science, directly impacting the design, programming and optimization of computer systems artifacts. The past 20 years have seen many network computing applications. However, an effective classification of network computing systems has not emerged in the academia. This negatively impacted scientific research, knowledge accumulation and sharing in network computing. This paper presents a classification based on three dimensions of (execution, control, and level), and shows that the classification method is able to cover representative network computing systems. The proposed method borrowed from successful experiences in classifying parallel computing systems. Keywords network computing; classification; grid; Web service; cloud computing Background In the past 20 years, many network computing systems have emerged due to rapid increases in market demands and enabling technologies. As these systems evolve by combining the best features of existing systems, differences among network computing systems are blurring. Research in classification is needed to assess new features and contributions of variant network computing systems, and to accelerate scientific research, knowledge accumulation and sharing in the field of network computing systems. In the applications and industry fields, network computing systems are grouped in an ad hoc fashion. Examples include single-site clusters, multisite clusters, peer-to-peer networks, grids, computing utilities, service-oriented computing, cloud computing, business Web, consumer Web, etc. There is a lack of commonly accepted classification or taxonomy. In academia, little research work has been done in classifying network computing systems. Only a four-way taxonomy has been found in the literature. Maheswaran et al proposed to differentiate network computing systems into four types, using the number of applications and number of owners as the grouping criteria. It is a good start, but still leaves many systems difficult to classify, such as grid, cloud computing, and decentralized systems. This paper presents a new classification method, based on three dimensions of execution, control, and level. It shows that the classification method can cover representative network computing systems. For instance, we now have a class of systems featuring distributed execution, central control, and at application level. Many Internet search services belong to this class. This research is supported in part by the National Basic Research Program (973 Program) of China Program of China (grant Noª±2005CB321807) and the National High Technology Research and Development Program (863 Program) of China (grant Nosª±2006AA01A106, 2006AA01A121). The former is research on Internet virtual computing environment for scientific activities, while the later aims at research and development of middleware level software as well as applications on the China National Grid environment. A similar part of those projects is to analyze the characteristics of corresponding applications and optimize the architecture of each system. The research in this paper is a basic component of application analysis and architecture study, directly impacting the design, programming and optimization of such systems.