Department of Computer Engineering
S E M I N A R
Auto-parallelization in Data Stream Processing Systems
Buğra Gedik, Ph.D.
IBM T. J. Watson Research Center
In this talk, I will give a brief overview of my research on distributed data stream processing systems, with a particular focus on auto-parallelization. I will describe the basic premise of stream computing, give an overview of middle-ware and language support we have built for supporting stream computing applications, and do a deeper dive on auto-parallelization in streaming systems. The data flow graph representation of stream processing applications carries great potential for efficient runtime execution due to the inherent parallelism present in it. However, harnessing this parallelism requires reasoning about safety, which is a challenge in the presence of state, selectivity, and user-defined operators. Furthermore, the highly dynamic nature of streaming workloads requires adaptation. Such runtime adaptation in the presence of continuous flow of data requires special machinery to perform changes in the execution plan on-the-fly. In this talk, I will describe how we address these challenges in the context of data parallelism as well as pipelined parallelism and present some preliminary performance results. Time permitting, I will talk about future research directions in this area.
Bio: Buğra Gedik is currently a Research Staff Member at the IBM T. J. Watson Research Center. His research interests are in distributed data-intensive systems, with a particular focus on stream computing. In the past, he has served as the chief architect for IBM InfoSphere Streams. He is the co-inventor of the SPL and the SPADE stream processing languages. He is the recipient of the IEEE ICDCS 2003, IEEE DSN 2011, and ACM DEBS 2011 best paper awards. He has served as the PC chair for the ACM DEBS 2009 and IEEE CollaborateCom 2007 conferences. He has published over 50 peer-reviewed articles in the areas of distributed computing and data management. He is an IBM master inventor. Before joining IBM, he has received his Ph.D. degree from Georgia Institute of Technology, USA and prior to that, his B.S. degree from Computer Engineering Department at Bilkent University, Turkey.
DATE: 02 January, 2012, Monday @ 13:40