一个开发处理大数据软件的分析模型.pdf

DOT: 一个开发处理大数据软件的分析模型 张晓东 美国俄亥俄州立大学 Team Members 1 1 Yin Huai , Rubao Lee , 2 1 Simon Zhang , Cathy H. Xia 1Dept. of Computer Sci. Eng., The Ohio State University 2Department of Computer Science, Cornell University The Evolution of Computing Centralized computing era • The Von Neumann model (1945): A baseline model to guide centralized computer architecture design Parallel computing era • The Bulk Synchronous Parallel (BSP) model (Leslie Valiant, 1990): A “scale-up” model to guide parallel architecture improvement and software optimization software for HPC “The data center as a computer” era • Only Software frameworks for big data analytics available –MapReduce, Hadoop, Dryad and several others • No models yet 3 Von Neumann Model: Computer Architecture Design Before Von Neumann’s computer project, several operational computers were built: • 1936, Zuse’s Z1 (1st binary ditital computer) in Germany • 1937, Atanasoff and Berry’s ABC (1st electronic digital computer) in Iowa State University Von Neumann model • 1943, ENIAC based on ABC in UPenn •a memory containing both data and The most importa