Cloud Workflow Resource Shortage Prediction and Fulfillment Using Multiple Adaptation Strategies

Extending workflow orchestration to embrace monitoring and adaptation within cloud environment is perceived to be a challenging activity. It has to consider different resources, require heavy processing to adapt to the dynamic nature of cloud environment. In this paper, we propose a multi-model framework for workflow resource monitoring, prediction, and adaptation. The framework supports continuous monitoring of several workflow runtime cloud entities and detect diverse types of violations (e.g. resource saturation). Moreover, collected logs resulted from monitoring are used as a training dataset for predicting resource shortage. Furthermore, two adaptation strategies are proposed to cope with environment resources changes and avoid violations: 1) monitoring-based adaptation and 2) prediction-based adaptation. Both adaptation schemes perform the necessary actions to adapt resources according to workflow required quality levels. To evaluate our monitoring and adaptation approaches we used a real cloud environment where we perform a couple of experimental scenarios. Experiments results showed that our framework and proposed monitoring, prediction and adaptation schemes are efficient in detecting violations, accurately predicting cloud resource shortages and accordingly issuing the proper adapting decisions.