Analysis of service diagnosis improvement through increased monitoring granularity

Cuiting Chen, Hans Gerhard Gross, Andy Zaidman

Research output: Contribution to journalArticleScientificpeer-review

29 Downloads (Pure)

Abstract

Due to their loosely coupled and highly dynamic nature, service-oriented systems offer many benefits for realizing fault tolerance and supporting trustworthy computing. They enable automatic system reconfiguration when a faulty service is detected. Spectrum-based fault localization (SFL) is a statistics-based diagnosis technique that can be effectively applied to pinpoint problematic services. However, SFL exhibits poor performance in diagnosing services which are tightly interacted. Previous research suggests that an increase in the number of monitoring locations may improve the diagnosability for tight interaction. In this paper, we analyze the trade-offs between the diagnosis improvement through increased monitoring granularity and the overhead caused by the introduction of more monitors, when diagnosing tightly interacted faulty services. We apply SFL in a service-based system, for which we show that 100 % correct identification of faulty services can be achieved through the increased monitoring granularity. We assess the overhead with increased monitoring granularity and compare this with the original monitoring setup. Our experimental results show that the monitoring at the service communication level causes relatively high overhead, whereas the monitoring overhead at a finer level of granularity, i.e., at the service implementation level, is much lower, but highly dependent on the number of monitors deployed.
Original languageEnglish
Pages (from-to)437-471
Number of pages35
JournalSoftware Quality Journal
Volume25
Issue number2
DOIs
Publication statusPublished - Jun 2017

Keywords

  • Residual defect
  • Fault localization
  • Online monitoring
  • Simulator
  • Service framework

Fingerprint Dive into the research topics of 'Analysis of service diagnosis improvement through increased monitoring granularity'. Together they form a unique fingerprint.

  • Cite this