Measurement-Based Dependability Evaluation of Multiprocessor Systems.

Abstract

In the previous reporting period, we showed that, when many users run the same software, such as an operating system, the majority of field software failures are rediscoveries of known problems. This clearly shows that the number of faults identified in a software system is not the only important factor; rediscoveries can seriously degrade software dependability. The effects of rediscoveries are: (1) more failures than the number of faults, (2) wasted service resources due to repeated data collection and diagnosis of the same problem, and (3) being unable to provide service to users until diagnosis is complete even if a solution already exists. To address the second and third effects, we developed symptom-based diagnosis of rediscovered software failures and showed that the majority of rediscoveries can be automatically diagnosed based on symptoms, such as procedure call trace and problem detection location.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1996
Accession Number
ADA327988

Entities

People

  • R. I. Iyer

Organizations

  • University of Illinois Urbana–Champaign

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Computer Programming
  • Detection
  • Device Drivers
  • Distributed Computing
  • Electronic Mail
  • Engineering
  • Fault Tolerance
  • Fault Tolerant Computing
  • Machine Languages
  • Measurement
  • Models
  • Operating Systems
  • Reliability
  • Reliability Engineering
  • Software Development
  • Test And Evaluation

Fields of Study

  • Computer science
  • Engineering

Readers

  • Educational Psychology
  • Mental Health of Military Veterans with Posttraumatic Stress Disorder (PTSD): Risk Factors, Prevalence, Symptoms, and Treatment.
  • Software Engineering.