Management Intelligence in Service-Level Reconfiguration of Distributed Network Applications

  • Authors:
  • K. Ravindran

  • Affiliations:
  • Department of Computer Science, City College of CUNY and Graduate Center, New York, USA 10031

  • Venue:
  • SOCASE '09 Proceedings of the AAMAS 2009 International Workshop on Service-Oriented Computing: Agents, Semantics, and Engineering
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

The paper is on generic service-level management tools that enable the reconfiguration of a distributed network application whenever there are resource-level changes or failures in the underlying network sub-systems. A network service is provided to applications through a protocol module, with the latter exercising network infrastructure resources in a manner to meet the application-level QoS specs. Application requests for a network service instantiate the protocol module with parameters specified at the service interface level, along with a prescription of critical properties to be enforced. At run-time, a management module monitors the service compliance to application-prescribed requirements, and notifies the application if a QoS degradation is detected. In our model, the management entity uses policy scripts and rules to coordinate the application-level reconfigurations and adaptations in response to the changes/outages in underlying network services. Our management model is independent of the specifics of problem-domain, which lowers the software development costs of distributed applications through `reuse' of the management module. The paper presents a case study of rate-adaptive video multicast over IP networks, to demonstrate the usefulness of our model.