Priority based Modeling and Comparative study of Google Cloud Resources between 2011 and 2019

The cloud resource allocation for jobs must be further optimized and prioritised due to ever increasing demand for cloud computing resources to handle big data. In this research, we have examined the relationship between resource allocation, usages, and priority of tasks to reveal the influence of priority in resource allocation and resource usages. The analysis and modeling of this paper have used the Google cloud public dataset of 2011 and 2019. After processing and cleaning of one month data of Google cloud, we have revealed, the tasks are classified in 12 priorities in the 2011 cluster model whereas 500 priorities in the 2019 cluster model. However, both models have grouped these priorities into five groups. Therefore, we have modeled resource allocation versus usages based on five main priority groups using XGBoost (Extreme Gradient Boosting) and correlation coefficient. The comparative study on the developed models shows, the priority grouping of 2019 has better evenly distribution of resources for jobs but less efficient in most of the priority groups for resource allocation. Based on the performance parameters of the developed models, the resource allocation works more efficiently for most of the 2011 priority groups except 'other'. These findings are useful for researchers to develop a balanced priority-based resource allocation-usages model to further optimise resources to reduce the management cost of cloud clusters.