Research On Traffic Sensing And Control Technology Of Multi-tenant Based On Openflow

With the rapid development of information technology, especially the emergence of big data, companies need to take full advantage of network virtualization technology to build their own virtualization platform or hire other companies’ virtual data centers. Multi-tenant shared network is an emerging technology, not only to meet the urgent needs of enterprise data center network, reducing enterprises’ deployment and operational costs, but also to greatly improve the efficiency of the physical network resources. In the case of multiple tenants, due to the upper application’s diversities and vary demands over time and space, the usage of network resources varies widely.In order to meet different users’ needs and not produce interference with each other, we need to study virtualization technology to make logical partitions for limited physical resources. In order to better allocate resources while ensure tenant services to be effectively guaranteed, we need to study how to allocate limited and resources, and by adjustment to meet the needs of the tenants in time to improve utilization of resources.In this paper, we study multi-tenant network virtualization technology, do design based Openflow technology-oriented multi- tenant virtual data center network and network resources perception and control technology. Including:Establishments of different users’ service type. Multi-tenant virtualized network has to meet the isolation requirements between networks, to ensure that each tenant different quality requirements of service, this paper, combines multi- tenant network requirements with the characteristics of software-defined networking Openflow to establish service model based on OpenvSwitch to ensure the identity of the user traffic, the service type separation of the user.Allocation and scheduling algorithm of path and bandwidth for multi-tenants. Based on OpenvSwitch virtual switches, to form abstract available physical network resources, forming a virtual network layer. By introducing VM contains a number of tenant requests N, the number of the group containing the VM K, VM and external communication bandwidth requirements Inter, VM and VM communication between tenants Intra and a description of the parameters ?, the interface abstraction is:(N, K, Inter, Intra,?), then through the interface to do network resource allocation to improve the al ocation of bandwidth efficiency.Network resources perception and control technology. In order to ensure user quality of service features for multi-tenant network, this paper presents the flow of perception based on neural network algorithm. Openflow controllers distr ibuted access of each user’s status by sensing algorithms predict user traffic, and according to the forecast results to control OpenvSwitch.Under limited bandwidth, ensure users’ services quality of different types of services.Experimental results show that through the application of our perception, control, and distribution loop control can effectively improve network allocated efficiency while meet the different needs of users.By our proposed algorithm bandwidth resource allocation method, we provides a viable approach for multi-tenant data center network bandwidth allocation in the data center network. Network-based multi-tenant data centers Openflow can effectively isolate traffic between tenants, to ensure secure communications tenants. Bandwidth allocation algorithm proposed in this paper can improve the utilization of data center network bandwidth; promote a reasonable allocation of bandwidth for multiple tenants, the introduction of flexibility to adapt to the dynamic factor of the request of tenants. Flow control algorithm perceived trends BP neural network algorithm to predict traffic, can make a single tenant in the total network resources without changing the applicatio n to the data center, a high priority to ensure maximum quality of service, improve tenant services satisfaction...