The project proposes and investigates reconfiguration policies of cloud and bandwidth resources in Network Function Virtualization environments in which the Network Function Virtualization Infrastructure-Point of Presences are interconnected by an elastic optical network. The proposed policies are applied in dynamic traffic scenarios, they are based on Virtual Network Function Instances migrations and reconfiguration of optical circuits. The policy aims at minimizing the sum of three cost components: the cloud resource cost, the bandwidth cost and the reconfiguration cost. The reconfiguration costs are characterized by the revenue loss of a Network Operator due to the bit loss occurring when the optical circuits are reconfigured. We will formulate the optimization problem and because of its complexity we propose and investigate effective heuristics. The project aims at proving that the proposed heuristics allow, in typical traffic and network scenario, for a consistent cost reduction with respect to the solutions proposed in literature in which only the sum of cloud and bandwidth resource costs are considered and reconfiguration costs are not take into account.
The project is organized in four WPs. In the WP1 an extension of the ETSI NFV architecture will be proposed to support algorithms for the cloud and bandwidth resource reconfiguration. WP2 is devoted to propose reconfiguration algorithms based on the minimization of the sum of cloud, bandwidth and reconfiguration costs and taking into account optical signal quality constraints. traffic prediction algorithms will be studied and integrated to the ones studied in the WP2. The proposed algorithms will be tested in the WP4 by providing numerical results to evaluate the effectiveness of the algorithms.
The introduction of algorithms for the cloud and bandwidth reconfigurations in NFV networks architectures and their performance evaluation will produce an important knowledge advances within the scientific community. In fact though reconfiguration algorithms were already studied in the literature, the innovative contributions of the project are:
i)to propose algorithms for the case in which the cloud infrastructures are interconnected by an elastic optical network;
ii)to take into account the reconfiguration costs characterized by the revenue loss of a network operator for the information lost during the optical circuit reconfiguration.
iii)the introduction of sophisticated traffic prediction algorithms based on deep learning and allowing for an optimal dimensioning of the cloud and bandwidth resources avoiding underprovisioning and overprovisioning problems.
The project will provide the networking equipment manufacturers with advanced techniques for the cost reduction in NFV architectures. We will provide operator-proof solutions because we will also make clear how the proposed reconfiguration algorithms can be inserted in the ETSI NFV architecture recognized by the operators as the NFV architecture of the future networks.
The effectiveness of the proposed solutions will be investigated in the real network and traffic scenarios.
The expected impact of the project will be an increased economic efficiency of NFV architecture through a cost reduction and a consequently improved scalability. The proposed techniques provide for a cloud and bandwidth consolidation in low traffic scenario. That allows for a big energy saving due to the switching off of cloud infrastructures during the low traffic periods. The solutions proposed in the project, if successful, will have the impact to reduce the operation cost of the data center networking operator. They are paying astronomical energy bills whose reduction can provide a non-marginal reduction of OPEX, which can offer relief for operators but can also be, at least partially, transferred to reduce the cost for the final user.
Finally the solutions studied in the project will produce a remarkable social impact in reducing the pollution, in fact due to the rapid development of cloud infrastructure, the energy consumption of network equipment has increased significantly and that has a considerable impact on the environment, responsible for high CO2 emissions.