Research On Task Management Strategy And Performance Of Cloud Architecture For Multi-priority Users

By constructing cloud architecture efficiently and flexibly,cloud service system can provide dynamically expanded supply of virtual resource.In order to meet the service needs of users with different priorities,improve the resource utilization of the cloud service system,and maximize the revenue of users and cloud service providers,we investigate the cloud architecture with diverse users,and study task management and performance optimization.Firstly,considering the case that the load of the diverse tasks are lower,we set a group of high-speed server groups in the cloud service system,and propose a low-load cloud architecture with diverse users.The tasks from membership users are regarded as first-class customers,while tasks from normal users are regarded as second-class customers.Then we construct a multi-user multi-server queueing model with preemptive priority.Secondly,for the case that the load of tasks from normal users is higher,we set two server groups of different service ability in the cloud service system,which contains a server group at low speed which only provide services for tasks from normal users,and we propose a high-load cloud architecture for normal users with diverse users.The tasks from membership users are regarded as first-class customers,while tasks from normal users are regarded as second-class customers.Then we construct a multi-user multi-server queueing model with preemptive priority.Then,for the case that the load of tasks from membership users is higher,we set two server groups of same service ability in the cloud service system,which contains a server group at high speed which only provide services for tasks from membership users.And we propose a high-load cloud architecture for membership users with diverse users.The tasks from membership users are regarded as first-class customers,while tasks from normal users are regarded as second-class customers.Then we construct a multi-user multi-server queueing model with preemptive priority.Finally,the steady-state analysis of the queuing models are carried out by using the matrix-geometric solution method,and the system performance is evaluated from the aspect of the average response time of tasks from normal users.We construct revenue functions for the three cloud architectures respectively,and study Nash equilibrium and socially optimal behavior of normal users.We propose a pricing policy for normal users to adjust the arrival behavior of normal users and achieve the unification of Nash equilibrium and social optimization.Without affecting the service quality of membership users,it increases the utilization rate of cloud service resources,so as to maximize the benefits of cloud service providers and cloud users.