The

When users' tasks in a distributed heterogeneous computing environment (e.g., cluster of heterogeneous computers) are allocated resources, the total demand placed on some system resources by the tasks, for a given interval of time, may exceed the availability of those resources. In such a case, some tasks may receive degraded service or be dropped from the system. One part of a measure to quantify the success of a resource management system (RMS) in such a distributed environment is the collective value of the tasks completed during an interval of time, as perceived by the user, application, or policy maker. The Flexible Integrated System Capability (FISC) measure presented here is a measure for quantifying this collective value. The FISC measure is a flexible multidimensional measure, and may include priorities, versions of a task or data, deadlines, situational mode, security, application-and domain-specific QoS, and task dependencies. For an environment where it is important to investigate how well data communication requests are satisfied, the data communication request satisfied can be the basis of the FISC measure instead of tasks completed.

No abstract

Abstract

This paper focuses on the problem of monitoring the end-to-end performance of message passing to support adaptive applications to be executed using the MSHN system (Management System for Heterogeneous Networks). Eight commercial and research tools and application components that attempt to measure perceived end-to-end message passing performance were identified. Two were dismissed; one because of recently published findings and the other because it is typically used in too many inconsistent configurations. The remaining six are carefully described in the paper. We were able to characterize each as either passive or active, determine whether they require domain-specific knowledge of an application, identify sources of inaccuracies, and enumerate their limitations. Based upon this survey, and previous analytical experiments, we conclude that the optimal monitoring mechanism: (1) should be passive; (2) should not require domain-specific knowledge of an application; (3) should minimize sources of error; and (4) should have few limitations. No single tool or application component surveyed has all of these characteristics. Based upon the surveyed work and other recent research in distributed systems, we have synthesized a new tool whose mechanisms have all of the desired characteristics. This paper describes our mechanism, and how we implemented it, in detail.