QoS-Driven Proactive Adaptation of Service Composition

Abstract

Proactive adaptation of service composition has been recognized as a major research challenge for service-based systems. In this paper we describe an approach for proactive adaptation of service composition due to changes in service operation response time; or unavailability of operations, services, and providers. The approach is based on exponentially weighted moving average (EWMA) for modelling service operation response time. The prediction of problems and the need for adaptation consider a group of services in a composition flow, instead of isolated services. The decision of the service operations to be used to replace existing operations in a composition takes into account response time and cost values. A prototype tool has been implemented to illustrate and evaluate the approach. The paper also describes the results of a set of experiments that we have conducted to evaluate the work.

Risk-Based Proactive Process Adaptation

“…In particular, we use multilayer perceptrons as a specific form of ANNs. We use the implementation of multilayer perceptrons (with their standard parameters) of the WEKA open source machine learning toolkit 1 . As attributes for the prediction models, we take the expected and actual times for all services of the process until the point of prediction.…”

“…Thereby, proactive process adaptation addresses shortcomings of reactive adaptation, such as loss of money (e.g., due to contractual penalties) or time-consuming roll-back and compensation activities [1,20].…”

“…Proactive adaptation relies on predictive process monitoring to forecast potential problems [1]. Predictive process monitoring predicts how an ongoing process instance will unfold up to its completion [19,22].…”

Adaptive composition in dynamic service environments

“…to prevent future failure or improve performance), these are still very limited and mainly focus on the change detection part, giving little or no consideration to the actual adaptation process. Proposed proactive change detection methods include applying performance prediction techniques [8,23,9], testing the behaviour of services 160 using generated test cases [7,24], and subscribing to change requests with the registry [25]. Such detection efforts can be considered complementary to our work (which focuses instead on the latter change handling step).…”

“…Like us, a few approaches also consider subsequent proactive adaptation actions (e.g. [8,9,26]), but these actions are mostly instantiated for corrective purposes, to prevent an anticipated problem, ignoring 165 optimisation opportunities (G2). Furthermore, no proper modeling and management of the adaptation process, to avoid its interference with the application's execution (G5), is provided.…”