Task Scheduling in the Cloud Using Machine Learning Classification

Cloud computing is a distributed computing model which enables developers to automatically deploy their applications onto the cloud. There are many applications running on a cloud which requires parallel processing capabilities. Applications of such nature require an efficient scheduling algorithm to manage heavy traffic. The drawbacks of existing scheduling algorithms are low resource utilization and more response time. The aim of the proposed system is to improve resource utilization & response time in the cloud using machine learning classification. Rather than implementing single scheduling algorithms, multiple scheduling algorithms are implemented. Selection of the efficient scheduling algorithm is done using machine learning classification. Initially attributes of the tasks and the Virtual Machine’s (VM’s) are extracted and used as training data. Training data are given as input to the machine learning algorithm which then produces classification rules. Based on classification rules efficient scheduling algorithm is selected and tasks are executed. The proposed scheme is implemented and tested in the CloudSim simulation toolkit. WEKA tool is used for testing datasets and for the selection of the classification algorithm. KeywordsCloud computing; Task scheduling; Parallel Workloads; Resource utilization.