New Techniques for Enhancing Performance and Efficiency in Dynamic Workload Driven Clusters
Large-scaled cluster systems like data centers and cloud computing become an important part of contemporary computing environments. Everybody is moving their computing and data from desktops to large cluster systems, spanning from scientific computing clusters to commercial and military data centers. How to effectively manage resources in such large systems is an inherently difficult problem as the complexity of these systems increases and the workloads to these systems are becoming dynamic and diverse. In particular, this requires new designs that are able to manage unplanned increases or bursts in user demands. Bursty workloads are often found in multi-tier architectures, storage systems, and communication networks. This feature is extremely important in system design because it can significantly degrade system performance and availability. This talk focuses on how to leverage the knowledge of burstiness or temporal dependence in workloads to develop new techniques and tools for modeling, prediction, and resource management. More specifically, we will present in this talk (1) new capacity planning models to well capture dependence and variability of the true service processes in systems, despite inevitable inaccuracies that result from inexact and limited measurements; (2) new load balancing algorithms for the application providers to deal with bursty user demands in order to minimize delay and interferences from other applications; and (3) a new VMware Flash Resource Manager (VFRM) which aims to maximize the utilization of Flash resources with minimal CPU, memory, and I/O cost for managing and operating Flash.
Ningfang Mi is an Assistant Professor in Department of Electrical and Computer Engineering (ECE) at Northeastern University. She graduated with a B.S. in Computer Science from Nanjing University, China in 2000 and a M.S. in Computer Science from the University of Texas at Dallas in 2004. She received her Ph.D in Computer Science from the College of William and Mary in 2009. She is a recipient of the 2010 IBM Faculty Award and the 2014 Air Force’s Young Investigator Research (YIP) Award. She also received the 2008 Best Paper Award of ACM/IFIP/USENIX International Middleware Conference, the 2009 Computer Management Group (CMG) Graduate Fellowship, and the 2010 Best Student Paper Award of International Teletraffic Congress (ITC). Her expertise and research interests lie in performance evaluation, resource management, capacity planning, and simulation in enterprise systems, data centers and cloud computing.