Map Reduce Autoscaling over the Cloud with Process Mining Monitoring

Over the last years, the traditional pressing need for fast and reliable processing solutions has been further exacerbated by the increase of data volumes - produced by mobile devices, sensors and almost ubiquitous internet availability. These big data must be analyzed to extract further knowledge. Distributed programming models, such as Map Reduce, are providing a technical answer to this challenge. Furthermore, when relaying on cloud infrastructures, Map Reduce platforms can easily be runtime provided with additional computing nodes (e.g., the system administrator can scale the infrastructure to face temporal deadlines). Nevertheless, the execution of distributed programming models on the cloud still lacks automated mechanisms to guarantee the Quality of Service (i.e., autonomous scale-up/-down behavior). In this paper, we focus on the steps of monitoringMap Reduce applications (to detect situations where the temporal deadline will be exceeded) and performing recovery actions on the cluster (by automatically providing additional resources to boost the computation). To this end, we exploit some techniques and tools developed in the research field of Business Process Management: in particular, we focus on declarative languages and tools for monitoring the execution of business process. We introduce a distributed architecture where a logic-based monitor is able to detect possible delays, and trigger recovery actions such as the dynamic provisioning of a congruent number of resources.