Approaches for adaptive optimization of routes in a distributed software-defined infrastructure of a virtual data center

In this investigation, we presented a description of the intelligent system for optimization traffic of the cloud applications which processing Big data in network environment. A feature of the developed system is the use of modern virtualization technologies and machine learning methods for managing traffic flows when organizing access to cloud applications and services. The developed solution is based on the hybrid approach. We combining computing powers of CPUs and graphics cards (GPUs) in the analysis and processing of Big data in real time. It allows solving task of placement for network elements and dataset of cloud application in virtual data center. Also we formalize the optimization problem to determine the minimum time for the find minimal cost path for data transfer between storages nodes. This problem is formalized in the form of graph model and it is NP complete. In order to find a suboptimal solution for polynomial time, it is proposed to use the genetic algorithms and neural network approaches.