Hierarchical Scheduling In Parallel And Cluster Systems

Multiple processor systems are an important class of parallel systems. Over the years, several architectures have been proposed to build such systems to satisfy the requirements of high performance computing. These architectures span a wide variety of system types. At the low end of the spectrum, we can build a small, shared-memory parallel system with tens of processors. These systems typically use a bus to interconnect the processors and memory. Such systems, for example, are becoming commonplace in high-performance graph ics workstations. These systems are called uniform memory access (UMA) multiprocessors because they provide uniform access of memory to all pro cessors. These systems provide a single address space, which is preferred by programmers. This architecture, however, cannot be extended even to medium systems with hundreds of processors due to bus bandwidth limitations. To scale systems to medium range i. e. , to hundreds of processors, non-bus interconnection networks have been proposed. These systems, for example, use a multistage dynamic interconnection network. Such systems also provide global, shared memory like the UMA systems. However, they introduce local and remote memories, which lead to non-uniform memory access (NUMA) architecture. Distributed-memory architecture is used for systems with thousands of pro cessors. These systems differ from the shared-memory architectures in that there is no globally accessible shared memory. Instead, they use message pass ing to facilitate communication among the processors. As a result, they do not provide single address space.

Computer system performance evaluation is a key discipline for the understanding of the behavior and limitations of large scale computer systems and networks. This volume provides an overview of the milestones and major developments of the field.The contributions to the book include many of the principal leaders from industry and academia with a truly international coverage, including several IEEE and ACM Fellows, two Fellows of the US National Academy of Engineering and a Fellow of the European Academy, and a former President of the Association of Computing Machinery.

Computer system performance evaluation is a key discipline for the understanding of the behavior and limitations of large scale computer systems and networks. This volume provides an overview of the milestones and major developments of the field. The contributions to the book include many of the principal leaders from industry and academia with a truly international coverage, including several IEEE and ACM Fellows, two Fellows of the US National Academy of Engineering and a Fellow of the European Academy, and a former President of the Association of Computing Machinery. Contents:Ken Sevcik as an Advisor and Mentor (E Lazowska et al.)Shadow Servers and Priority Scheduling (J P Buzen)On the Chronology of Dynamic Allocation Index Policies: The Pioneering Work of K C Sevcik (E Coffman)Operational Analysis (P J Denning)Function Approximation by Random Neural Networks with a Bounded Number of Layers (E Gelenbe et al.)The Achilles' Heel of Computer Performance Modeling and the Model Building Shield (V De Nitto Personè & G Lazeolla)Wireless Network Simulation: Towards a Systematic Approach (S K Tripathi et al.)Location- and Power-Aware Protocols for Wireless Networks with Asymmetric Links (G Wang et al.)Multi-Threaded Servers with High Service Time Variation for Layered Queueing Networks (G Franks et al.)Quantiles of Sojourn Times (P G Harrison & W J Knottenbelt)Asymptotic Solutions for Two Non-Stationary Problems in Internet Reliability (Y Kogan & G Choudhury)Burst Loss Probabilities in an OBS Network with Dynamic Simultaneous Link Possession (T Battestilli & H Perros)Stochastic Analysis of Resource Allocation in Parallel Processing Systems (M S Squillante)Periodic Task Cluster Scheduling in Distributed Systems (H Karatza) Readership: Graduate students, Internet engineers, computer scientists, system engineers, and computer designers. Also suitable for use in professional development seminars in computers and networks. Keywords:Computer Networks;Distributed Systems;Performance Analysis of Computer Systems;Performance Control and Optimization;System Performance Evaluation;Analytic Models;Simulation Methods

This book presents revised versions of tutorial lectures given at the IEEE/CS symposium on modeling, analysis, and simulation of computer and telecommunication systems held in Orlando, FL, USA in October 2003. The lectures are grouped in three parts on performance and QoS of modern wired and wireless networks, current advances in performance modeling and simulation, and other specific applications of these methodologies. This tutorial book is targeted to both practitioners and researchers. The practitioner will benefit from numerous pointers to performance and QoS issues; the pedagogical style and plenty of references can be of great use in solving practical problems. The researcher and advanced student are offered a representative set of topics not only for their research value but also for their novelty and use in identifying areas of active research.

Communication networks and computer systems research is entering a new phase in which many of the established models and techniques of the last twenty years are being challenged. The research community is continuing to free itself from past intellectual constraints so that it may fully exploit the convergence of computing and communications. Evaluating the performance of emerging communications and computer systems constitutes a huge challenge. Thus, current research provides a set of heterogeneous tools and techniques embracing the uncertainties of time and space varying environments when the requests for diverse services are made in real time, and with very different quality of service expectations. These novel techniques will lead to fast and economic service deployment and effective dynamic resource management, and hence to new business strategies and infrastructures that will facilitate the emergence of future services and applications. This volume contains contributions and presentations made by leading international researchers at a workshop which was held in April 2004 to honour Professor Erol Gelenbe on the occasion of his inaugural lecture as the Dennis Gabor Chair at Imperial College London.

Online scheduling is recognized as the crucial decision-making process of production control at a phase of “being in production" according to the released shop floor schedule. Online scheduling can be also considered as one of key enablers to realize prompt capable-to-promise as well as available-to-promise to customers along with reducing production lead times under recent globalized competitive markets. Online Scheduling in Manufacturing introduces new approaches to online scheduling based on a concept of cumulative delay. The cumulative delay is regarded as consolidated information of uncertainties under a dynamic environment in manufacturing and can be collected constantly without much effort at any points in time during a schedule execution. In this approach, the cumulative delay of the schedule has the important role of a criterion for making a decision whether or not a schedule revision is carried out. The cumulative delay approach to trigger schedule revisions has the following capabilities for the practical decision-making: 1. To reduce frequent schedule revisions which do not necessarily improve a current situation with much expense for its operation; 2. To avoid overreacting to disturbances dependent on strongly an individual shop floor circumstance; and 3. To simplify the monitoring process of a schedule status. Online Scheduling in Manufacturing will be of interest to both practitioners and researchers who work in planning and scheduling in manufacturing. Readers will find the importance of when-to-revise policies during a schedule execution and their influences on scheduling results.

An essential task in radar systems is to find an appropriate solution to the problems related to robust signal processing and the definition of signal parameters. Signal Processing in Radar Systems addresses robust signal processing problems in complex radar systems and digital signal processing subsystems. It also tackles the important issue of defining signal parameters. The book presents problems related to traditional methods of synthesis and analysis of the main digital signal processing operations. It also examines problems related to modern methods of robust signal processing in noise, with a focus on the generalized approach to signal processing in noise under coherent filtering. In addition, the book puts forth a new problem statement and new methods to solve problems of adaptation and control by functioning processes. Taking a systems approach to designing complex radar systems, it offers readers guidance in solving optimization problems. Organized into three parts, the book first discusses the main design principles of the modern robust digital signal processing algorithms used in complex radar systems. The second part covers the main principles of computer system design for these algorithms and provides real-world examples of systems. The third part deals with experimental measurements of the main statistical parameters of stochastic processes. It also defines their estimations for robust signal processing in complex radar systems. Written by an internationally recognized professor and expert in signal processing, this book summarizes investigations carried out over the past 30 years. It supplies practitioners, researchers, and students with general principles for designing the robust digital signal processing algorithms employed by complex radar systems.

Hard real-time systems are very predictable, but not sufficiently flexible to adapt to dynamic situations. They are built under pessimistic assumptions to cope with worst-case scenarios, so they often waste resources. Soft real-time systems are built to reduce resource consumption, tolerate overloads and adapt to system changes. They are also more suited to novel applications of real-time technology, such as multimedia systems, monitoring apparatuses, telecommunication networks, mobile robotics, virtual reality, and interactive computer games. This unique monograph provides concrete methods for building flexible, predictable soft real-time systems, in order to optimize resources and reduce costs. It is an invaluable reference for developers, as well as researchers and students in Computer Science.