Search

Cloud computing is advancement of IT sector where enduser are provided with the services of infrastructure and applications on the basis of pay-per use model. Some of the cloud based application services are social networking, web hosting and content delivery. Several elements are present in cloud that is clients, datacenter and distributed servers. With cloud computing there is high availability, flexibility, less overhead for users and reduced cost. The major problem associated with cloud computing is balancing the load among cloud by choosing effective load balancing algorithm. Load in cloud can be any form of CPU load, memory capacity, delay and network load. Load balancing is the technique to distribute the load among various nodes of distributed system for better resource utilization and response time. One more concern is to balance the load among thousands of virtual machines. In this paper, we proposed the hybrid of two methodologies, a decentralized load balancing architecture called tldlb which provide load balancing and high availability. It uses the the algorithm nn-dwrr, for dispatching large number of client requests to different virtual machine for providing services by reducing the SLA violation. But in our proposed methodology we used JIQ in place of nn-dwrr according to which initially cloudlet will be assigned to idle virtual machine and if virtual machine in virtual machine list is not idle then cloudlet will be assigned on the basis of virtual machine having minimum response time and minimum queue length.

As seen in many studies the relationship of object oriented matrices of the software and the calculated maintenance effort metric is very complicated, complex and nonlinear in nature. So with this kind of behavior, we can have got a research area where we can work upon to minimize the maintenance effort which can be used to develop and deploy models and systems for the forecasting of software maintenance effort of the software. In our work we are using Support Vector Machine for the regression for forecasting of software maintenance factor with the Univariate and Multivariate approach. For better performance of Support Vector Machine we are using one of the popular kernel function, i.e. Radial Basis Kernel Function. We are using one commercial software products QUES Dataset (Quality Evaluation System) for the work. Dependent variable in our work is maintenance effort and independent variables are OO metrics. We will verify our work on Univariate and Multivariate premise.