Towards a Multi-Model Cloud Workflow Resource Monitoring, Adaptation, and Prediction

Workflow configuration, re-configuration execution, monitoring and adaptation over a cloud environment are considered very challenging activities. This is due to the fact that such activities are resource-aware, require intensive processing, and should adapt to dynamic cloud changes. In this research, we propose a multi-model for workflow resource monitoring, resource prediction, and resource adaptations. Three adaptation strategies are proposed to capture changes in environment resources, categorize various violations and take the necessary actions to adapt resources according to workflow needs. Workflow resource prediction uses ARIMA to predict resource shortage and support adequate adaptation. However, extreme adaptation is supported by continuously monitoring various workflow environment entities. We also evaluate workflow trust based on QoS to support the different adaptations strategies. We implemented our model on a cloud environment and we experimented different adaptation scenarios. The results validated the effectiveness of our monitoring, prediction and adaptation schemes in detecting violations and hence, predicting accurately cloud resource shortages and takes the appropriate actions to deal with these violations.