Fault tolerant high performance Information Services for dynamic collections of Grid and Web services

  • Authors:

  • Mehmet S. Aktas;Geoffrey C. Fox;Marlon Pierce

  • Affiliations:

  • Community Grids Laboratory, Indiana University, 501 N. Morton Suite 224, Bloomington, IN 47404, United States and Computer Science Department, School of Informatics, Indiana University, Bloomingto ...;Community Grids Laboratory, Indiana University, 501 N. Morton Suite 224, Bloomington, IN 47404, United States and Computer Science Department, School of Informatics, Indiana University, Bloomingto ...;Community Grids Laboratory, Indiana University, 501 N. Morton Suite 224, Bloomington, IN 47404, United States

  • Venue:
  • Future Generation Computer Systems
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

E-Science Semantic Grids can often be thought of as a dynamic collection of semantic subgrids where each subgrid is a collection of a modest number of services that are assembled for specific tasks. We define a Gaggle as a modest number of managed and actively interacting Grid/Web Services, where services are put together for particular functionality. The information management requirements in Gaggles include both the management of large amounts of relatively static services and associated semantic information as well as the management of multiple dynamic regions (sessions or subgrids) where the semantic information is changing frequently. We design a hybrid, fault tolerant, and high performance Information Service supporting both the scalability of large amounts of relatively slowly varying data and a high performance rapidly updated Information Service for dynamic regions. We use the two Web Service standards: Universal Description, Discovery, and Integration (UDDI) and Web Services Context (WS-Context). We evaluate our approach by applying various tests to investigate the performance and sustainability of the centralized version of our implementation that is applied to sensor and collaboration grids. The experimental study on system responsiveness of the proposed approach shows promising results. This study indicates that communication among services can be achieved with efficient centralized metadata strategies, with metadata coming from more than two services. In contrast point-to-point methodologies provide service conversation with metadata only from the two services that exchange information. In addition, our performance indicates that efficient mediator services also allow us to perform collective operations such as queries on subsets of all available metadata in service conversation.