Resource Management Framework for Multi-tier Service Using Case-Based Reasoning and Optimization Algorithm

The emergence of cloud computing has made elasticity of virtual resources one of the most critical features of cloud service. Such elasticity reflects the fluctuation of resource provisioning due to the variety of service demands. Most high-demanding services adopt a multi-tier architecture. However, offering quality-of-service ($${\texttt {QoS}}$$QoS) guarantee for these services with least resource usage costs under dynamic and unpredictable workloads and different resource demands is a significantly complex problem. Therefore, cloud providers ($${\texttt {CP}}\hbox {s}$$CPs) need to adopt a dynamic resource optimization and provisioning framework. Numerous rule-based and model-based approaches have been designed for dynamic resource provisioning in virtualized data centers. However, these approaches mainly focus on providing service-level $${\texttt {QoS}}$$QoS guarantees for running services and most of them do not address mainly the problem of minimizing the number of running virtual machines in order to increase $${\texttt {CP}}$$CP profit. This research proposes a new resource optimization and provisioning ($${\texttt {ROP}}$$ROP) framework to detect, solve the bottlenecks, and satisfy the service-level $${\texttt {QoS}}$$QoS requirements of running services and to increase the $${\texttt {CP}}$$CP profits. To demonstrate the effectiveness of the proposed $${\texttt {ROP}}$$ROP against other approaches, a prototype running on a cloud platform is developed, and a workload generator and multi-tier service model are adopted. Results show that the $${\texttt {ROP}}$$ROP framework outperforms other existing approaches by 75% in terms of on-demand service configurations while providing service-level $${\texttt {QoS}}$$QoS guarantee for running services.