Large number of users are shifting to the cloud system for their different kind of needs. Hence the number of applications on public cloud is increasing day by day. Public clouds considered and is the most convenient platform for common cloud users with generic needs and lesser security concerns. Public cloud can cater to the needs of a large group of users and provide a variety of services. Lower cost and timely availability are the other advantages one expects from public clouds. These features make it very much convenient and attractive choice. But on the other hand, handling public cloud become unmanageable in comparison to other counterparts. Monitoring so many users, tasks and resources are difficult task. Sometimes public clouds are divided on geographically. Geographic partitioning of public cloud can resolve these issues by adding manageability and efficiency in this situation. But, partitioned clouds introduce different ends for submission and operations of cloudlets and virtual machines. This ends for task submission and resource allocation adds complexities also. A concrete mechanism is to be designed for handling the load allocation and processing of the nodes. The proposed work is addressing the same issue by advising a combination of centralized and decentralized load balancing. The main objective of this work is to fix a VM for a cloudlet, which can process it in minimum time and without overloading or underloading the datacenters. Another objective under consideration is to reduce the number of jobs left unhandled due to threshold constraints.
A novel Agent Based Load Balancing Model for maximizing resource utilization in Grid Computing
