Research And Verification Of Resource Scheduling Algorithm For OVirt Platform

The oVirt platform manages a large number of resources,whether the allocation of resources is reasonable will affect the task processing capabilities of overall system and user experience.Load balancing as an important component of the cloud platform,still has some shortcomings in oVirt,which will lead to the overall performance degradation and unreasonable allocation of resources.In order to improve the oVirt cluster performance and reduce the energy consumption of the system,the PBORLB and ECPTLB algorithms are proposed respectively,and the verification system of the oVirt platform resource scheduling algorithmis is designed to verify the above algorithms.In the design process of load balancing algorithm,the advanced achievements of load balancing theory are studied.The architecture of oVirt platform and its load balancing implementation mechanism are analyzed.Finally,the oVirt load balancing algorithm is optimized to improve load balancing performance and reduce migration cost.The main work of this thesis is as follows:(1)Based on load balancing technology and oVirt cloud platform architecture and its source code,the load balancing algorithm in oVirt platform is analyzed.The advantages and disadvantages of load balancing design in oVirt cloud platform are summarized.(2)In order to overcome the shortcomings of incomplete resource management,inappropriate load balancing granularity and ignoring virtual machine migration performance loss in vm_evenly_distributed and evenly_distributed algorithm,the PBORLB algorithm is designed.The PBORLB algorithm considers the different resource requirements of each virtual machine,and uses load forecasting to improve the hysteresis of load balancing algorithm in oVirt platform.Then the genetic algorithm is designed to solve the multi-objective optimization problem,which includes population coding,genetic operator design and fitness function design and so on.(3)For the power-saving algorithm in oVirt platform failed to consider the cost of virtual machine migration and the situation that system performance degradation caused by virtual machine resource requirements change frequently,the ECPTLB algorithm is designed.The ECPTLB algorithm focuses on reducing energy consumption while considering the system performance by guaranteeing redundant computing resources.(4)The verification system of oVirt platform resource scheduling algorithm(ORSAVS)based on the Master/Slave architecture is designed to verify the above algorithm.The overall structure and detailed design of the verification system are described and the detailed design scheme of each module is given.Finally,the PBORLB and ECPTLB algorithms are analyzed and validated by the verification system.The experimental results show that the PBORLB algorithm has its advantages in improving load balancing and reducing virtual machine performance loss and the ECPTLB algorithm can reduce the energy consumption while reducing the migration costs effectively.