Cloud Sensor Ontology and Linked Data to Support Autonomicity in Cloud Application Platforms

Abstract

Cloud application platforms with their numerous deployed applications, platform and third-party services are becoming increasingly complex, dynamic and data-intensive, and require novel intelligent approaches to be applied in order to maintain them at an operational level. By treating cloud application platforms as distributed networks of software sensors and utilising techniques from the Semantic Sensor Web area, we have developed a monitoring framework which allows us to detect, diagnose and react to emerging critical situations in complex environments of cloud application platforms in a dynamic manner. In this paper, we focus on our use of a Sensor Cloud Ontology to: (i) represent cloud-based logical software sensors; (ii) homogenise monitored sensor data in the form of RDF streams; and (iii) apply stream and static reasoning to these monitored values in order to detect critical situations. We also explain how utilisation of Linked Data principles can help achieve a more flexible and extensible architecture to define diagnosis and adaptation policies. We discuss benefits associated with our approach, as well as potential shortcomings and challenges.