LAUSR

Popularity of cloud computing is being increased drastically by the use of people all over the world in their comfort zone. For the purpose of withholding the performance of cloud, it is significant to design an efficient scheduling methodology. Researches have designed methods like directed acyclic graph, MinES, MinCS, ant colony optimization, cross-entropy stochastic scheduling, interlacing peak scheduling method, etc for enhancing cloud and user experience. The major problem existed in these methods was higher execution time, overload issues and higher power consumption. To overwhelm the problems, we design a novel framework that is comprised with queue manager (QM), scheduler (SH), virtual machine manager (VMMA), VM allocator (VMA) and VM power manager (VMP). Firstly, in QM the incoming tasks are split into two queues based on task’s deadline, they are represented as urgent queue (UQ) and waiting queue (WQ). Secondly, we perform hybrid scheduling which combines grey system and neural network (GSNN) that considers three significant parameters as task length, CPU intensive and memory intensive. This GSNN scheduling is enabled to withstand effectively even for ‘N’ number of tasks and that leads to minimization of execution time. Then thirdly, each task is allocated to corresponding VM with respect to the capacity and workload of VM. Finally VMP keeps updating the VM information for computing the underutilized hosts then it performs VM migration to put up the idle host to OFF state, for reducing the unwanted power consumption. Simulation results of our entire framework shows improvements when compared with state-of-the-art methods.