Comprehensive approach for optimization traffic routing and using network resources in a virtual data center.

Abstract Today, telecommunication networks are the basis for deploying different types of cloud applications and services in data centers. The growing use of the cloud computing concept to provide access to applications and services every year increases the amount of converged network traffic. Existing approaches for traffic routing which using in the software defined-network don't provide enough flexible solutions, able to adapt to changes in traffic flows in real time. At peak load, this leads to an overabundance of traffic to specific physical network nodes which are not prepared to handle a large data flow. One of the approaches that make it possible to solve this problem is the organization of adaptive traffic routing. In this paper are describes the developed approaches for effective control of traffic flow in the virtual data center. Our approaches are using the methods of the data mining and machine learning to more accurately classify and identify traffic flows of cloud applications and network services. This information helps for optimization traffic routes in real time. Another element of the proposed complex solution is solving the task of identifying routes in the virtual network of a data center. Within the framework of the research, the problem of route identification has been decomposed into two consecutive subproblems. The first part is aimed at the classification of communication channels in a virtual data center for a number of features that are characteristic of all superimposed networks. At the second part, based on the received data, the routes are clustered in accordance with the QoS requirements imposed by the current traffic flows. The received data are combined with information about the current flows of the circulating network and are fed to the input of the neural network to decide on the choice of a suitable route. Thus, the solution of the task of adaptive routing allows determining the priorities when creating the bandwidth of traffic flows, as well as establishing rules for managing the network of the data center. [ABSTRACT FROM AUTHOR]