学术报告
Title:
Cache Management for In-Memory Big Data Analytics
Jun Zhang
Research Assistant Professor
Hong Kong University of Science and Technology (HKUST)
时间：2018年10月15日下午16:00
地点：物联网科技园大楼西5楼通信技术研究所会议室
Abstract
Today’s big data analytics systems are shifting towards in-memory computations. By caching input and intermediate datasets in the main memory, users can gain orders of magnitude improvement for I/O intensive jobs. However, the memory resource is limited and shared by many users, and thus effective cache management plays a pivotal role, which faces a few key challenges. First, prevalent systems employ rather simple caching policies, which cannot exploit the data dependency that commonly exists in big data analytics. In shared, multi-tenant clouds, to ensure both user performance and high utilization of cache, there is a pressing need to enforce fair cache allocation. Another challenge is the severe load imbalance across cache servers. In particular, the routinely observed file popularity skew and load imbalance create hot spots, which significantly degrade the benefits of in-memory caching.
This talk presents some recent results to address the above research issues. It first presents a new cache replacement policy, called Least Reference Count (LRC), which exploits application-specific data dependency to improve the cache efficiency. Then a fair cache sharing mechanism for shared cloud environments, termed OpuS, or Opportunistic Sharing, is introduced, which provides performance isolation between users and is strategy-proof against “free-riding” manipulations. Finally, SP-Cache, which selectively partitions files based on their popularity, is proposed to achieve load balancing without cache redundancy or encoding/encoding overhead.
Biography
Dr. Jun Zhang received the Ph.D. degree in Electrical and Computer Engineering from the University of Texas at Austin. He is currently a Research Assistant Professor in the Department of Electronic and Computer Engineering at the Hong Kong University of Science and Technology (HKUST). His research interests include wireless communications and networking, mobile edge computing and edge learning, and big data analytics systems.
Dr. Zhang co-authored the book Fundamentals of LTE (Prentice-Hall, 2010). He is a recipient of several best paper awards, including the 2016 Marconi Prize Paper Award in Wireless Communications, the 2014 Best Paper Award for the EURASIP Journal on Advances in Signal Processing, an IEEE GLOBECOM Best Paper Award in 2017, an IEEE ICC Best Paper Award in 2016, and an IEEE PIMRC Best Paper Award in 2014. One paper he co-authored received the 2016 Young Author Best Paper Award of the IEEE Signal Processing Society. He also received the 2016 IEEE ComSoc Asia-Pacific Best Young Researcher Award. He is an Editor of IEEE Transactions on Wireless Communications, and is a guest editor of the special section on "Mobile Edge Computing for Wireless Networks" in IEEE Access.
南京邮电大学通信与信息工程学院
物联网研究院 通信技术研究所
2018.10.13