Browse Prior Art Database

UNIFIED PERFORMANCE-RELIABILITY EVALUATION

IP.com Disclosure Number: IPCOM000128482D
Original Publication Date: 1984-Apr-01
Included in the Prior Art Database: 2005-Sep-16
Document File: 18 page(s) / 86K

Publishing Venue

Software Patent Institute

Related People

Meyer, J.F.: AUTHOR [+3]

Abstract

Unified performance-reliability (performability) evaluation is reviewed relative to needs imposed by distributed real-time systems. Such systems typically exhibit properties of concurrency, timeliness, fault tolerance, and degradable performance, calling for unification and generalization of modeling techniques employed in strict evaluations of performance and reliability. An overview of our work in this area is presented, including a discussion of directions currently being pursued.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 5% of the total text.

Page 1 of 18

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

UNIFIED PERFORMANCE-RELIABILITY EVALUATION

J. F. Meyer

CRL-TR-27-84

THE UNIVERSITY OF MICHIGAN COMPUTING RESEARCH LABORATORY1

April 1984

Room 1079, East Engineering Building
Ann Arbor, Michigan 48109
USA
Tel: (313) 703-8000

To appear in Proceedings of 1984 American Control Conference, San Diego, CA, June 1984. UNIFIED PERFORMANCE-RELEABILITY EVALUATION
J. F. Meyer

ABSTRACT

Unified performance-reliability (performability) evaluation is reviewed relative to needs imposed by distributed real-time systems. Such systems typically exhibit properties of concurrency, timeliness, fault tolerance, and degradable performance, calling for unification and generalization of modeling techniques employed in strict evaluations of performance and reliability. An overview of our work in this area is presented, including a discussion of directions currently being pursued.

[ Chapter ] 1. INTRODUCTION

This paper concerns the development of modeling methods and tools for unified performance- reliability evaluation of distributed real-time systems. The types of models considered are principally stochastic models of an analytic nature. However, nonprobabilistic models can play a useful supporting role, as described in Section 3. The intent of this presentation is to provide an overview of work in this area that has been underway at The University of Michigan since 1970. Also discussed are directions we are taking in our current research, both at The University of Michigan and at the Industrial Technology Institute.

Generally, an evaluation of a system seeks to relate and quantify aspects of what the system is and does with respect to what the system is required to be and do. Accordingly, performance- reliability evaluation methods for distributed real-time systems need to be responsive to the kinds of performance-reliability requirements that accompany the use of such systems. Although these requirements differ in detail from application to application, they have some typical properties which, in turn, impose special conditions on system structure and behavior.

In particular, such requirements typically call for high performance when the system is fault-free. This, coupled with the fact that resources are distributed, results in systems which exploit

1 This work was supported in part by the Communications and Network Laboratory of the Industrial Technology Institute. Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the author and do not necessarily reflect the views of the funding agencies.

University of Michigan Computing Research Laboratory Page 1 Apr 01, 1984

Page 2 of 18

UNIFIED PERFORMANCE-RELIABILITY EVALUATION

concurrent (parallel) processing. Secondly, real-time aspects of such requirements often impose conditions on the "timeliness" of processing activities, e.g., requirements for maximum allowable (input-output) response times wh...