Research On Resource Allocation Mechanism Of Dynamic Service Function Chain

In recent years,software technology has flourished,hardware performance has continued to increase,and novel network services have emerged.The rigid underlying network infrastructure makes it difficult for ISPs to cope with the challenges brought by current network users and new traffic surges.Technologies such as Service Function Chain came into being.By concatenating the modular virtual network service functions,SFC builds a complete end-to-end network service in the network environment,which brings a new evolution direction to the traditional underlying network.The resource allocation of the service function chain also referred to as the mapping of the SFC is to instantiate the service functions included in the SFC in the network,and the traffic need to pass through the set of network functions.In the dynamic scenario,when allocating resources for SFC,not only the resource constraints need to be considered,but also the survivability and dynamics of the SFC.When the service request arrives dynamically,the selection of the mapping scheme of the current service request may have an unpredictable impact on subsequent service requests.In a dynamic scenario,how to implement dynamic SFC survivability mapping decision-making is an urgent problem to be solved.First,this thesis studies the survivability mapping problem of dynamic SFC.This thesis proposes a matching-based heuristic algorithm for finding a mapping solution for efficient resource utilization rapidly.In order to solve the survivability of SFC,this thesis proposes two protection schemes,such as path protection and segment protection.Both protection schemes can guarantee survivability for SFCs when dealing with single node or single link failures.The algorithm finds the network function node mapping scheme by matching between SFC and entwork.The network function is then connected by the shortest path and the original SFC link is not intersected with the backup link.Simulation results show that,compared with the comparative greedy algorithm,the algorithm can achieve a good balance between running time and throughput in dynamic scenarios.Then,this thesis studies the decision mapping problem of dynamic SFC.This thesis proposes a decision algorithm based on the dual theory of linear programming,which is used to evaluate the selection of different mapping schemes to improve the throughput of the network.The algorithm decompose the network according to the amount of resources,and preferentially deploys the dynamically arrived SFC in a sub-network with more resources.The algorithm evaluates each arriving SFC based on the dual theory and prioritizes the mapping deployment scheme with a smaller evaluation factor.Subsequent theoretical analysis and simulation experiments show that the algorithm can quickly give the decision of the mapping scheme and can accommodate higher throughput.