CS700:Graduate Seminar in Computer Science & Informatics

Big Data Management and Systems Design
Weikuan Yu, Associate Professor, Auburn University

The explosive growth of big data prompts many organizations to deploy various MapReduce systems to meet their needs of data analytics. This talk will cover several recent research efforts on big data management and systems design. Three techniques have been designed to speed up data analytics on top of MapReduce. These include Hadoop acceleration, fast completion scheduler and cross-layer task coordination, that can accelerate the shuffling of intermediate data, increase fairness and throughput of job scheduling, and enhance task coordination, respectively. The work on Hadoop Acceleration has been transferred to industry as a software product called Unstructured Data Accelerator.
Bio Weikuan Yu is an Associate Professor at Auburn University where he directs the NSF-funded TigerSphere project for campus-wide network renovation and the Parallel Architecture and System Laboratory (PASL) that hosts a 98-node CPU+GPU cluster for interdisciplinary data analytics and systems research. Yu has research interests on big data, high-performance computing, computer architecture, file and storage systems, and interdisciplinary research on climate modeling and computational biology. His research has been sponsored by NASA, NSF, DOE, ORNL, LLNL, Intel, Mellanox, NVIDIA, SolarFlare, Scitor, Huawei, Alabama Department of Commerce, and Auburn University, and has won the 2012 Alabama Innovation Award and the First Prize in the 2012 Grand Finals of ACM Student Research Competition.