Dynamic service selection with end-to-end constrained uncertain qos attributes

  • Authors:

  • Rene Ramacher;Lars Mönch

  • Affiliations:

  • Chair of Enterprise-wide Software Systems, Univerity of Hagen, Hagen, Germany;Chair of Enterprise-wide Software Systems, Univerity of Hagen, Hagen, Germany

  • Venue:
  • ICSOC'12 Proceedings of the 10th international conference on Service-Oriented Computing
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Services and service compositions are executed in an uncertain environment with regard to several aspects of quality. Static service selection approaches that determine the entire service selection prior to the execution of a service composition are extensively discussed in the literature. Nevertheless, the uncertainty of quality aspects has been not well addressed in the service selection phase so far, leading to time-consuming and expensive reconfiguration of service compositions at their execution time. Due to the uncertain and dynamic nature of the execution environment, a dynamic service selection approach is highly desirable. In a dynamic service selection, the services are selected during the execution of a service composition taking into account the conditions caused by already executed services. A dynamic service selection contributes to implement robust service compositions to support reliable business processes, where robustness is measured in terms of fulfilling quality constraints of a service composition. In this paper, we examine a dynamic service selection approach based on a Markov decision process. The service selection is considered from a cost minimizing point of view with an end-to-end constrained execution time. A simulation study demonstrates that the dynamic service selection outperforms an optimal static service selection in an uncertain environment with respect to robustness of the service compositions and cost minimizing.