Research On SLA-based Distributed Mutual Exclusion Algorithm In Local Cloud Computing

Cloud computing is a model aimed to providing ubiquitous, convenient,on-demand network access to shared pool of configurable computing resources. It’sget more and more attention and application because of its good scalability andconvenience. Since applications in the cloud computing environments need to shareresources, it is need to provide a mutually exclusive access control services under acloud computing environments in order to prevent the concurrent use of criticalresources by multiple applications and control the use of critical resources.In previous work, distributed mutual exclusion algorithm is divided into twotypes: token-based mutual exclusion algorithm and permission-based mutualexclusion algorithm. In this article we has been briefly introduced the distributedmutual exclusion algorithms.Because of the token-based mutual exclusionalgorithms present an average message traffic which is lower than that ofpermission-based ones. Thus they are more suitable for controlling concurrentaccesses to shared resources in clouds whose number of nodes is often very large.However, the existing distributed algorithms can not meet the requirements of mutualexclusion algorithms in the cloud computing environment. Because they do not takeinto account guarantee SLA(Service Level Agreement), and applied in cloudenvironment directly will bring the increase of the SLA violation.In this paper, we studied on two types of SLA violation:(1) request priorityinversion: when QoS requirements can be associated to priority, i.e., differentSLA-levels can be mapped to different priorities, SLA violation refers to low priorityrequests get response before the high-priority requests;(2) request response timeoverdue: when QoS requirements are expressed in terms of request response time,SLA violation refers to requests were not satisfied before a given response time.For the two cases mentioned above, this paper has present two SLA-orienteddistributed mutual exclusion algorithms.(1) SLA-oriented priority-based mutualexclusion algorithm: this algorithm has modified the Kanrar-Chaki algorithm, throughthe introduction of a ‘level’ heuristic growth strategy, slowing the growth rate of low priority to reduce the number of priority inversion, and through the introduction ofthe ‘distance’ token scheduling policy and traffic optimization strategies reduce theaverage message.(2) SLA-oriented response time based mutual exclusion algorithm:the algorithm through the access control to the requests to greatly increase theprobability that the system accepted requests get response before the given responsetime, and improve the utilization of the critical resources by presenting the tokenscheduling with preemption strategy.Finally, the simulation results of these two algorithms show that two algorithmsreduce the number of SLA violations, ensure the users’ QoS requirements, while alsoincreasing efficiency in the use of critical resources to reduce the number of messagetransmission in the network.