Online Testing for Proactive Adaptation

Contact person

The goal is to enable industry building adaptive systems with proactive capabilities. The contribution is about using online testing to actively collect quality data for predicting failures of services in a service-oriented system. Online testing means testing services in parallel to their normal use and operation. The aim is to achieve accurate failure predictions by increasing the amount of available monitoring data (e.g. QoS data). The idea is to exploit synergies with existing monitoring data. On the one hand, to use monitoring data to build usage models which can be used for test case selection. Thus, services are only tested when there is not sufficient monitoring data. On the other hand, monitoring data is used together with the results of Online testing to predict failures of services in a service-oriented system.  As part of the life-cylce of adaptive systems, this prediction can be employed for triggering adaptations and thus improved proactive adaptation decisions.

Technical Information

The contribution describes a framework for proactive adaptation using online testing. The framework is composed of components for monitoring services, updating usage models,  selecting online tests based on usage models, testing services and finally predicting failure of the services.

Demo

N/A

Publications 

• Osama Sammodi, Andreas Metzger, Xavier Franch, Marc Oriol, Jordi Marco, Klaus Pohl: Usage-Based Online Testing for Proactive Adaptation of Service-Based Applications. COMPSAC 2011: 582-587

End-to-End Quality Provision and SLA Conformance, Adaption and Monitoring, and Composition and Coordination

Prototype (single scenario)

Relationship with Future Internet and Internet of Services

Relationship with Cloud

None

Web site

N/A