Migration Energy Aware Reconfigurations of Virtual Network Function Instances in NFV Architectures

Abstract

Network function virtualization (NFV) is a new network architecture framework that implements network functions in software running on a pool of shared commodity servers. NFV can provide the infrastructure flexibility and agility needed to successfully compete in today's evolving communications landscape. Any service is represented by a service function chain (SFC) that is a set of VNFs to be executed according to a given order. The running of VNFs needs the instantiation of VNF instances (VNFIs) that are software modules executed on virtual machines. This paper deals with the migration problem of the VNFIs needed in the low traffic periods to turn OFF servers and consequently to save energy consumption. Though the consolidation allows for energy saving, it has also negative effects as the quality of service degradation or the energy consumption needed for moving the memories associated to the VNFI to be migrated. We focus on cold migration in which virtual machines are redundant and suspended before performing migration. We propose a migration policy that determines when and where to migrate VNFI in response to changes to SFC request intensity. The objective is to minimize the total energy consumption given by the sum of the consolidation and migration energies. We formulate the energy aware VNFI migration problem and after proving that it is NP-hard, we propose a heuristic based on the Viterbi algorithm able to determine the migration policy with low computational complexity. The results obtained by the proposed heuristic show how the introduced policy allows for a reduction of the migration energy and consequently lower total energy consumption with respect to the traditional policies. The energy saving can be on the order of 40% with respect to a policy in which migration is not performed

Joint Energy Efficient and QoS-Aware Path Allocation and VNF Placement for Service Function Chaining

“…This section is divided into three different categories: VNF/VM placement problems [7]- [17], routing and service function chaining [18]- [28], and green service function chaining [5], [29]- [33]. From now on we refer to service function as VNF (Virtual Network Function).…”

“…On the other hand, Constraint (19) specifies the essential servers for active flows (i.e., those servers that deliver at least one VNF to a flow should be in ON mode). 4) Converting Nonlinear Constraints to Linear Forms: in the above proposed formulations, constraints (12), (17), and (19) are nonlinear (i.e., only the first problem "QoS and SFC Constraints without VNF Ordering" is linear). In order to make the formulation suitable for solving largescale problems (i.e with large number of flows and complex network configurations), these constraints are converted into linear form.…”

Balancing the migration of virtual network functions with replications in data centers

“…A typical example are migrations during the off-peak hours in DCNs to reduce energy consumption. As presented in [1], the goal to migrate the VNFs to a few servers only, such that the remaining servers can be switched off. To address the issue of service interruptions that negatively impact QoS during migrations, [2] derived a trade-off between the power consumption and QoS degradation to determine whether a migration is appropriate.…”

A Robust Optimization Based Energy-Aware Virtual Network Function Placement Proposal for Small Cell 5G Networks with Mobile Edge Computing Capabilities

“…For this reason, the problem of VNF placement has been attracting increasing attention during the last years [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25]. Nonetheless, the optimal placement of virtual elements to physical resources aimed at minimizing power consumption with service performance constraints for the 5G scenario under study is a complex issue.…”

“…Besides, [20] presents a suboptimal placement solution that combines traffic and energy cost optimization derived by means of a Markov approximation with matching theory. Apart from that, research works in [22][23][24] go a step forward and cope with the migration problem of active microservers needed in low traffic conditions, to turn off or suspend microservers so as to achieve energy savings. Moreover, a common assumption of the different optimization models proposed to solve different versions of the VNFs placement problem is that input data is perfectly known, which is difficult in practice.…”