Research On Service Function Chain Resource Allocation Mechanism Based On Service Load

With the continuous expansion of network scale and the explosive growth of various types of mobile devices,the variety and number quantity of network-based services are also increasing.In this case,in order to provide better services to meet the dynamic changes of users,network service operators need to reduce operating costs and shorten the time to market for new network services.Network Function Virtualization?NFV?^[1],[2]is a key technology for network transformation.By decoupling network functions from dedicated hardware,NFV can reduce capital expenditures and operating expenses,and help service providers to provide various network services in a more flexible and efficient way.NFV.Based on general hardware technologies and virtualization technologies,NFV runs the required network functions in the form of virtual network functions?VNFs?on standard commercial servers,storage,and switches.It enables cloud service providers to dynamically instantiate service functions to provide services to tenants,which facilitates unified management and scheduling of network functions.NFV can realize the dynamic deployment of VNF on the Network Function Virtualization Infrastructure?NFVI?,promote infrastructure resource sharing,and effectively reduce the investment cost and operating cost of the network.In the NFV architecture,a chain structure formed by several VNFs in an orderly manner according to service requirements is called a Service Function Chain?SFC?,which is a standard way to implement network services.SFCs can be deployed on the underlying physical nodes and physical links according to different deployment strategies.In this way,network traffic can pass through predefined SFCs through the established logic service requirements to achieve the required services.In order to meet the needs of network services and enhance the user experience,how to allocate resources quickly,efficiently and extensibly is one of the important issues faced by NFV.Based on the dynamic changes of service load,this paper studies the resource allocation mechanism of SFC.From the perspective of normal service load,high service load and service demand overload,the deployment of multi-dimensional resources in NFV,the service chain deployment considering fragment resources and SFC migration reconfiguration are studied to meet the load balancing target and increase the request acceptance rate.The main work of this paper is as follows:1.For the problem of joint optimization of node multidimensional resources and link resources in resource allocation,a multi-dimensional resource allocation strategy based on multi-stage directed graph is proposed.Under the condition of meeting the constraints of network resources,the optimization model is established with the goal of balancing node multi-dimensional resource load and minimizing link resource load.In each phase diagram,the service node decision algorithm is used to limit the physical nodes where VNF need to be deployed to ensure a more balanced VNF deployment.Then the strategy represents the node and link resources as the weight of the directed graph,finds the deployable location of each VNF corresponding phase map,and calculates the optimal deployment path.The simulation results show that the algorithm optimizes the node resource load,balances the node resources and link bandwidth,supports more service requests,and has a lower algorithm complexity.2.For the problem of node and link resource optimization considering resource fragmentation in resource allocation,this paper proposes a multi-path resource allocation strategy for service function chain.The algorithm splits the service function chain into groups of sub-chains,obtains the optimal link between each set of corresponding VNFs based on the weighted maximum matching algorithm for bipartite graph,and then combines the optimal partial paths to deploy the sub-chains in a multi-path way.This kind of deployment can make more efficient use of network resources,which is conducive to improving the"pressure-bearing"capability of heavy-load networks.The results show the proposed algorithm obtains better SFC request acceptance rate and node resource utilization,reduces link resource utilization and increases network revenue.3.In order to solve the SFC migration problem when traffic peaks and business needs exceed the capacity of existing open servers,this paper proposes a service function chain migration reallocation strategy for overload control.First,the upper and lower thresholds of server reconfiguration are defined.When the load does not reach the lower threshold,the SFC is deployed only in the limited servers.When the load exceeds the upper threshold,a new server group is opened.This paper proposes a cost model to estimate the total cost of migration reallocation,and then proposes a service chain migration strategy based on particle swarm optimization.Under the condition of ensuring network load balancing,the algorithm selects the SFC and some of the VNFs therein with the least migration cost for migration.The experimental results show that compared with some existing heuristic algorithms,the algorithm can effectively reduce the SFC migration reconfiguration cost and ensure network load balancing.