ECE Seminar Series
Electrical and Computer Engineering, and Computer Science Special Seminar Sponsored by ISTeC
Title: Efficient Usage of Two Emerging Memory/Storage Technologies: NVRAM and Shingled Write Recording Drives
Speaker: David Hung-Chang Du
Affiliation: University of Minnesota
Day: Monday, March 23, 2015
Time: 2:00 pm - 3:00 pm
Location: Engr. E105
Abstract: The past decade has witnessed tremendous advances in computing, wired and wireless communication and storage technologies. It is also important that remarkable cost reductions have made large computing and storage capacity available to increasing number of consumers. With the unprecedented connectivity provided by Internet, many new applications have emerged and are being developed. A huge amount digital data has been created and become available for accesses to satisfy the demand of these new applications. It becomes an increasing challenge to even store this huge amount available data for our daily use. To answer the challenges in both computing and storage in this big data era, we more heavily depend on emerging memory/storage technologies. In this talk, we will cover the usage of Non-Volatile memory (NVRAM) and a new storage technology called Shingled Writing Recording.
All computers today have DRAM (Dynamic Random Access Memory) as main memory due to its fast access speed. Main memory in a computer is used for holding executable programs as well as caching data from slower storage devices. Since DRAM is volatile, the data updated has to be quickly flushed to storage devices like hard disks. Otherwise, the computer may run the risk of losing data if the computer or power failed. Because of this, the caching algorithms in the past only consider read cache since writes will be written to disks within a very short duration like less than 30 seconds. With the emerging non-volatile memory (NVRAM) becoming possible main memory, we have to consider caching for both reads and writes. In this talk, we will discuss a new caching algorithm called H-ARC. H-ARC is an extension of ARC caching algorithm and H-ARC will consider both read and write cache together. Our evaluation shows H-ARC can improved both read and write performance at the mean time it can dramatically reduce the traffic between main memory and storage devices like disk drives.
The big data era has brought tremendous potentials and benefits to the society. At the same time, it creates a crisis of drowning in data. Individuals cannot identify and locate the right information for their needs and organizations are struggling to manage and preserve the existing data. A complete solution to meet this challenge calls for new storage architectures, different storage systems design, new data models, new information access methods and new ways to deliver information. In this talk, we introduce a promising technology called Shingled Magnetic Recording (SMR) for storing large volume of data in magnetic disk drives (called Shingled Write Disks). In a SMR drive, data to be written sequentially with current track overlapped with some of the previous tracks to increase areal density of the drive. We will discuss the usage and several design issues of SMR drives.
Bio: David Hung-Chang Du: Dr. Du is currently the Qwest Chair Professor of Computer Science and Engineering at University of Minnesota, Minneapolis. He has served as a Program Director (IPA) at National Science Foundation (NSF) CISE/CNS Division from March 2006 to September 2008. At NSF, he was responsible for NeTS (networking research cluster) NOSS (Networks of Sensor Systems) Program and worked on Cyber Trust (Internet Security) Program. He is also the Director of a NSF I/UCRC Center on Intelligent Storage (CRIS). CRIS has been sponsored by more than 10 companies including Seagate, NetApp, Symantec, HP, Dell, SGI, LSI, Xyratex, NEC Labs, HGST, Los Alamos National Lab, etc. Dr. Du received a B.S. degree from National Tsing Hua University in 1974, an M.S. and Ph.D. degree from University of Washington (Seattle) in 1980 and 1981 respectively.
He joined University of Minnesota as a faculty since 1981. Dr. Du has a wide range of research expertise including multimedia computing, mass storage systems, high-speed networking, sensor networks, cyber security, high-performance file systems and I/O, database design, and CAD for VLSI circuits. He has authored and co-authored over 260 technical papers including 110 referred journal publications in these research areas. He has graduated 54 Ph.D. and more than 100 M.S. students in the last 30 years. Dr. Du is an IEEE Fellow (since 1998) and a Fellow of the Minnesota Supercomputer Institute. He is currently serving on the Editorial Boards of several international journals. He has also served as Conference Chair and Program Committee Chair for several major conferences in multimedia, networking, database, parallel/distributed computing and security areas. Currently he is the General Chair of the 30th IEEE Symposium on Security and Privacy (2009) and Program Committee Co-Chair for the 37th International Conference on Parallel Processing (2009), General Chair of ICDCS (2011) and General Chair for ICPP (2014). He has had research grants from many federal funding agencies including NSF, DARPA, ONR, and DOE. He has a strong tie with many industrial researchers and has collaborated with a number of companies.