Design And Implementation Of Function-group Orchestration Approaches In Smart Identifier Network

The combination of hardware and function makes the traditional network static and rigid,which is difficult to meet the needs of today's social development.In recent years,academia and industry have proposed new network architectures based on hardware virtualization technology to solve this problem.Smart Identifier Network is one of them,proposed by Beijing Jiaotong University.In Smart Identifier Network,nodes are defined as network components that perform functions such as data forwarding and collection,and they are divided into physical nodes and virtual nodes;The function-group is defined as a collection of nodes that provide complete network functionality;The function-group orchestration is defined as an approach for creating,deleting,and managing function-groups,and it is a key technology in Smart Identifier Network.This paper designs and implements a function-group orchestration approach with the following three characteristics:(1)Automatic and flexible orchestration;(2)Improves resource utilization;(3)Effectively guarantees the quality of operation.The main work of this paper include the following aspects:First,the paper designs and implements a frame for automatic function-group orchestration.According to function-group requirements,the frame for automatic function-group orchestration perceives and calculates information such as the function-group topology,virtual node specifications(CPU cores,memory and disk)and host physical nodes when creating a function-group.It is also responsible for the deletion and management of function-groups.The frame includes an API module,a matching module,a scheduling module,an issuing module,and a feedback module.Among them,the matching module and the scheduling module are core module.The matching module resolves the functional requirements into specific function-group topologies and node specifications by using the template library and matching rules.The scheduling module coordinates the matching result and the physical resource,and selects a relatively suitable function-group topology and virtual node specifications.It then completes the deployment of the function-group by combining the dynamic mapping approach based on the actual load for virtual node.Secondly,the paper designs and implements a dynamic mapping approach based on actual load for virtual node.It is used to solve the problem of selecting the host physical node for the virtual node when the function-group is deployed,including the over-deployment algorithm and the remapping algorithm.The over-deployment algorithm takes advantage of the characteristic that resources actually used by virtual nodes are mostly less than the resources requested and deploys the virtual node on the physical node based on running load.This can effectively improve resource utilization.The remapping algorithm moves the virtual nodes in the overloaded physical nodes out and reselects the host physical nodes for migrated virtual nodes,thereby reducing the load of the original host physical nodes and ensuring the operating quality of the nodes.Thirdly,the paper designs and implements an energy-saving scheduling approach for virtual node,which is referred to as the energy-saving scheduling approach.It is used to save energy while the virtual nodes are running.The energy-saving scheduling approach cooperatively distributes the load of each physical node by automatic migration of virtual nodes between physical nodes.When the physical node is unloaded,it can be time-phased shut down,thereby effectively reducing the actual number of running physical nodes and saving energy.In addition,in order to ensure that the quality of the operation is not affected while conserving resources,the energy-saving scheduling approach controls the load of the physical nodes of the migration destination when migrating virtual nodes.Finally,in order to verify the characteristics of the function-group orchestration approach,this paper deploys it into the actual test system built on OpenStack,and a lot of verification and test ware performed.The final result shows that the function-group orchestration approach can automatically and flexibly orchestrate function-groups,and can improve resource utilization and ensure operation quality.