Towards Policy-Supported Adaptable Service Systems

Abstract

Abstract. This paper presents a policy-supported architecture for adaptable service systems based on the combination of Reasoning Machines and Extended Finite State Machines. Policies are introduced to obtain flexibility with respect to specification and execution of adaptation mechanisms. The presented architecture covers two aspects: service system framework and adaptation mechanisms. The service system framework is a general framework for capability management. Adaptation mechanisms are needed for autonomous adaptation. The adaptation mechanisms can be based on static or dynamic policy systems. Capability management for of a simple music video-on demand service system with runtime simulation results based on the proposed architecture is presented.

On Capability-Related Adaptation in Networked Service Systems

“…There are several works that use policies to specify the adaptation, such as [3,4], [24], [25][26][27][28]. In [24] a framework that defines autonomic applications as dynamic composition of autonomic elements is described.…”

“…In [24] a framework that defines autonomic applications as dynamic composition of autonomic elements is described. Our approach as well as the approaches described in [3,4,[25][26][27][28] go beyond by adding mechanisms to adapt policies or the way of using policies. Such policy adaptation can be grouped into three categories: 1) changing the policy parameters, considered in [3,4,25,26]; 2) enabling/disabling a policy, found in [3,4,25]; 3) using techniques to select the most suitable policy and action; for instance, rewarding policies and their actions, presented in [27,28].…”

“…It is expressed with Conditions: CPU utilization > 95%, Constraints: system time between 18:00-24:00 and service time request > 2 minutes, and Actions: ignore new service requests of users of ordinary SLA classes. [3,4] is now defined as:…”

“…The streaming system is illustrated in Figure 10. The streaming case is the same as applied in a previous work [4]. The goal-based policy and the Learning Mechanism, however, were not applied in [4].…”

“…A goal-based policy ontology for capability-related adaptation is presented. The experience background for this publication is work with the TAPAS architecture and platform [3][4][5].…”