Research And Implementation Of Node And Network Resource Management Mechanism For Cloud System

With the continuous development of cloud computing,data center carries an increasing number of switches,servers and other hardware equipment,massive resources have brought enormous management pressure,on the same time,massive resources take a threat to the healthy and stable operation of cloud platform.There are two problems in the management of cloud resources,on the one hand,cloud computing system needs efficient node resource monitoring mechanism to coordinate the deployment of tasks and to improve the efficiency and stability of the system.However,the traditional centralized monitoring(CMM)can easily lead to a single point of failure and bottlenecks in terms of performance,these facts are not being suitable for large-scale cloud environment.On the other hand,the complex and heavy business brings heavy network load to cloud data center,at the same time,due to the load imbalance,there are many problems in the cloud data center network,such as low resource utilization,long time-delay,poor flexibility and no-reliability and low throughput,this is not an effective solution to the problem by traditional network architecture or simply adding new network resources.Therefore,its necessary to build network management mechanism for cloud systems,in order to effectively improve the utilization of resources and quality of service,and reduce operating costs in cloud data center,especially to solve network congestion,low utilization and reliability.According to these two problems,this thesis proposes a new distributed collaborative monitoring mechanism from point of node resource monitoring,which includes a distributed collaborative monitoring model(DCMM)and an adaptive threshold control algorithm(ATCA),and also proposes a multi-path traffic scheduling mechanism based on SDN(MTSS)from point of network resource management.The following are a summary of the contents.(1)This thesis proposes a distributed collaborative monitoring model(DCMM)for cloud computing infrastructures,which is based on the mutual perception and monitoring with each node to achieve the data nodes' self-management under the normal conditions and the information is sent to the master node in time under the abnormal conditions.It can balance the monitoring load and avoid the problems of single point of failure and performance bottleneck.(2)This thesis proposes an adaptive threshold control algorithom(ATCA),which is based on historical monitoring data and dynamically sets appropriate threshold to identify unnecessary duplicate monitoring data sent back to monitoring node.It can reduce monitoring data transmission,reduce the impact of monitoring system on the whole system.(3)The thesis proposes a multi-path traffic scheduling mechanism based on SDN(MTSS)for cloud system network,the network load balancing algorithm is used in MTSS to select the lightest path for the newly added data flow,MTSS periodically monitors the bandwidth utilization of each link,makes full use of the idle links in the network,and dynamically selects the appropriate forwarding path to schedule the traffic on the heavy link according to link load conditions,then programmable data forwarding is achieved to balance load.MTSS can effectively improve the network load balancing and resource utilization,reduce the packet loss rate,time-delay and improve the network throughput rate.(4)The thesis designs and builds cloud resource management system on the basis of collaborative monitoring and traffic scheduling mechanism.It introduces the architecture of the system int detail,and introduces each function and specific results respectively according to the system interface.