How Data Replication and Early-fetch Policies Impact MapReduce Job Reliability and Energy Consumption
Currently, MapReduce has been widely employed to tackle data-intensive tasks over a large-scale MapReduce cluster. But node failure in such a system is an unavoidable problem. MapReduce used several policies, such as data replication policies and early-fetch policy, to prevent data loss, reduce job failure probability and shorten job turnaround time. However, how these policies impact the job reliability (JR) and energy consumption (EC) of a MapReduce cluster is not clear. Here JR is defined as the reliability with which MapReduce cluster can complete a job, while EC is the energy consumed by the cluster to complete the job. In this talk, we would like to show the JR and EC of a MapReduce cluster on their policy combinations (POCOs) derived from two task assignment policies and two data replication policies. The conclusions enable MapReduce managers to comprehend how these POCOs impact the JR and EC of a cluster so that they can select suitable POCOs for their own MapReduce jobs and clusters.
Biography of Prof. Fang-Yie Leu
Fang-Yie Leu received his bachelor, master and Ph.D. degrees all from National Taiwan University of Science and Technology, Taiwan, in 1983, 1986 and 1991, respectively. His research interests include wireless communication, network security, Grid applications and Security. He is currently a professor of Computer Science Department and the director of Database and Computer Security Laboratory, TungHai University, Taiwan, and one of the editorial board members of at least 7 international journals. Prof. Leu now organizes MCNCS and CWECS international workshops. He is an IEEE member and now serves as the TPC member of at least 10 international conferences. He was also a visiting scholar of Pittsburg University.
Prof. Hiroaki Nishino
Oita University, Japan
Sensing and Rendering: An Approach for Assisting Human Creative Activities
Availability of high-performance low-cost processors and ubiquitous ICT infrastructure rapidly advance the embodiment of Cyber Physical System (CPS) in various research fields. CPS is a framework for seamlessly integrating physical objects into digital processing in a cyber world. It has originally been evolved in manufacturing for connecting various producing machines and their controlling computers. Now it is broadly interpreted as a concept for orchestrating human activities in the building of cyber-world applications. I would like to introduce an approach for assisting human creative activities by showing human-centered CPS applications incorporating multisensory technologies such as visualization and haptization. Some example systems such as a network topology visualization system through mobile augmented reality (AR), a skill acquisition system via haptic modality, and a personal 3D modeler for digital fabrication will be demonstrated in the talk.
Biography of Prof. Hiroaki Nishino
Hiroaki Nishino is a professor in the Department of Computer Science and Intelligent Systems, Faculty of Engineering, Oita University, Japan. He received his Doctoral degree of Engineering from Kyushu University, Japan in 2002. His research interests include virtual reality, computer graphics, and human-computer interaction. He has served as a program committee (PC) member in many international conferences and as a guest editor in several international journals. He has received best paper awards in several international conferences such as ICOIN-15, VSMM-2006, NBiS-2009, ICCE-TW 2014, ICCE-TW 2015, and BWCCA-2015.
