A model-driven monitoring approach for Internetware on platform-as-a-service (PaaS)

With the on-demand service provision ability and elastic nature, cloud computing provides a suitable development and running environment for internetware, to support its cooperation and evolvement. Among the three service models of cloud computing, Platform-as-a-Service (PaaS) provides internetware application developers with sufficient and convenient computing resources to develop and run their applications. As with other cloud computing models, monitoring is the basis of on-demand service provision and elastic scalability in PaaS. Apart from this, monitoring is also a required service that should be provided to PaaS consumers. However, due to the diversity and heterogeneity of the objects to be monitored in PaaS, monitoring is not easy to implement. Additionally, it is difficult to perform manual or automatic administration on PaaS based on raw monitoring data in the form of metric data set. In this paper, we propose a model-driven monitoring approach for PaaS. By transforming low-level monitoring data to high-level monitoring model based on meta-model definitions and transforming rules, this approach not only simplifies the implementation of PaaS monitoring system, but also presents a high-level abstraction of the whole PaaS, which reduces the difficulty of PaaS management. We built a graphic modeling tool to assist the definition of the meta-model and transforming rules, and applied this approach in the implementation of a PaaS monitoring system in practice.