Distributed Service Level Agreement-Driven Dynamic Cloud Resource Management

Abstract

The size and complexity of Cloud systems are growing more rapidly than expected, and hence, the management of these resources is a major research area. Resource provision with respect to SLA (Service Level Agreement) is directly tied up with customer satisfaction. Failure management is a real-time metric which needs to be addressed by providing continuous availability of resources to the users. This paper contributes to these issues by handling the client without intervention by a human. Continued availability of the client is successfully accomplished by deploying distributed sets of Orchestrators and SLA manager. Ability to deploy nodes remotely and restart nodes is the major area of analysis in this thesis. Monitoring of SLA from a single point for each cloud resource management system is the bottleneck in times of SLA manager failure, which is the brain of the system, and hence we need to overcome the same. Distributing the SLA manager is one of the approaches and which is also being proposed as a solution to the SLA manager failure. SLA Provides a lot of potential in managing cloud resources. In this paper, we build a model to show how WS-Agreement works using JSON. Then, we implement Load Balancing via Grid Gain to make a virtual copy of the same infrastructure as we obtained in SLA. This achieves the jobs to run on virtual machines by eliminating disaster recovery using particle swarm optimization algorithm.