this paper, we outline its design and detail the predictive performance of the forecasts it generates. While the forecasting methods are general, we focus on their ability to predict the TCP/IP end-to-end throughput and latency that is attainable by an application using systems located at different sites. Such network forecasts are needed both to support scheduling [5], and by the metacomputing software infrastructure to develop quality-of-service guarantees [10, 17]. Keywords: scheduling, metacomputing, quality-of-service, statistical forecasting, network performance monitoring

This paper discusses the I-WAY project and provides an overview of the papers in this issue of IJSA. The I-WAY is an experimental environment for building distributed virtual reality applications and for exploring issues of distributed wide area resource management and scheduling. The goal of the I-WAY project is to enable researchers use multiple internetworked supercomputers and advanced visualization systems to conduct very large-scale computations. By connecting a dozen ATM testbeds, seventeen supercomputer centers, five virtual reality research sites, and over sixty applications groups, the I-WAY project has created an extremely diverse wide area environment for exploring advanced applications. This environment has provided a glimpse of the future for advanced scientific and engineering computing. 1 A Model for Distributed Collaborative Computing

Fast networks have made it possible to coordinate distributed heterogeneous CPU, memory, and storage resources to provide a powerful platform for executing high-performance applications. However, the performance of these applications on such systems is highly dependent on the allocation and efficient coordination of application tasks. A key component for a performance-efficient allocation strategy is a predictive model which provides a realistic estimate of application performance under varying resource loads. In this paper, we present a model for predicting the effects of contention on application behavior in heterogeneous systems. In particular, our model calculates the slowdown imposed on communication and computation for non-dedicated twomachine heterogeneous platforms. We describe the model for the Sun/CM2 and Sun/Paragon coupled heterogeneous systems. We present experiments on production systems with emulated contention which show the predicted communication and computation costs...

Software for a growing number of problem domains has complex, time varying behavior and unpredictable resource demands (e.g., WWW servers and parallel input /output systems). While current performance analysis tools provide insights into application dynamics and the causes of poor performance, with a posteriori analysis one cannot adapt to temporally varying application resource demands and system responses. We believe that the solution to this performance optimization conundrum is integration of dynamic performance instrumentation and on-the-fly performance data reduction with real-time adaptive control mechanisms that select and configure resource management algorithms automatically, based on observed application behavior, or interactively, through high-modality virtual environments. We motivate this belief by first describing our experiences with performance analysis tools, input/output characterization, and WWW server analysis, and then sketching the design of interactive and clos...

Fast networks have made it possible to coordinate distributed heterogeneous CPU, memory and storage resources to provide a powerful platform for executing highperformance applications. However, the performance of parallel applications on such systems is highly dependent on the mapping of application tasks to machines. In this paper, we propose a mapping strategy for applications formed by multiple tasks targeted to heterogeneous platforms. We first define a mapping model, the match-tree, which reflects the data movement and conversion costs of distributed algorithms and allows for alternative implementations of individual tasks on different machines. We then define the find-mapping and split-partition algorithms, based on the match-tree model, to determine the best allocation of tasks to resources in heterogeneous systems. We illustrate the use of these algorithms with a sample distributed application. 1 Introduction In the last decade, distributed heterogeneous systems have emerged ...

Heterogeneous networks are increasingly being used as platforms for resource-intensive distributed parallel applications. A critical contributor to the performance of such applications is the scheduling of constituent application tasks on the network. Since often the distributed resources cannot be brought under the control of a single global scheduler, the application must be scheduled by the user. To obtain the best performance, the user must take into account both application-specific and dynamic system information in developing a schedule which meets his or her performance criteria. In this paper, we define a set of principles underlying application-level scheduling and describe our work-in-progress building AppLeS (application-level scheduling) agents. We illustrate the application-level scheduling approach with a detailed description and results for a distributed 2D Jacobi application on two production heterogeneous platforms. 1 Introduction Fast networks have made it possible ...