Service Function Chaining Placement And Routing In Data Centers

Nowadays,data centers deploy a large number of servers and network devices to provide various service functions(e.g.firewalls,content caches,WAN optimizers,etc.)and sufficient network connection.Most traffic in data centers is subject to treat-ment by multiple service functions(SFs)which form an ordered traditional Network Service Chain.Traditional Network Service Chain exists some disadvantages such as tightly coupling between network equipments,inefficiently packet forwarding,network equipment unable to be pooled,low expansibility and so on.Network Function Virtu-alization(NFV)makes that traditional network function existing in hardware can be deployed in the standard servers in the form of software.Service Function(SF)can be dynamically created and flexibly extended.Software Defined Networking(SDN)makes the controller to generate flow tables which automatically form service chains under the central control.Service Chain based on SDN and NFV is called Service Func-tion Chain(SFC).The main research work of this article as follows:(1)Based on the energy-aware service function placement problem for SFC in data centers,a new service function placement algorithm Merge-RD is proposed to save the energy consumption in data centers.First,we describe the system model,and give the power models of server and switch.Then,we formulate the energy-aware SF placement problem as a binary integer programming.Furthermore,we define a new metric called Relation Degree(RD)to quantize traffic intensity and dependency across SFs.Then we describe each step of the energy-aware algorithm Merge-RD in detail,including SF mapping and link mapping.Finally,simulation results show that Merge-RD outper-forms some existing placement algorithms in terms of the energy consumption and the transmission delay of the SFCs.The packet delivery ratio is guaranteed to a certain extent.(2)Based on the scheduling of Service Function and the routing of Service Function Chain,we use heuristic to solve this problem.First,we describe the system model,and give SF Instance Scheduling Model and SFC Routing Model.We adopt Quantum Genetic Algorithm(QGA)to implement simultaneously the SF Scheduling and SFC Routing.Finally,simulations are performed to show how the heuristic works correctly by guaranteeing a good packet delivery ratio,a nice delay distribution and a reasonable load distribution of SF instances.