Intent-driven Dynamic Service Function Chain Deployment And Scheduling

In conventional networks,service functions are tightly coupled with the underlying equipment,which has disadvantages such as low flexibility,poor scalability,and high operation and maintenance costs.At the same time,the diverse service demands generated by the growing number of network users are forcing the evolution of the network paradigm step by step.The continuous development of Software Defined Network(SDN)and Network Function Virtualization(NFV)technologies makes it possible to dynamically realize complex composite network services by connecting one or more physical or virtual components.Such an orderly combination of network service functions is called a network Function Service Chain(SFC).However,the life cycle of the end-to-end dynamic service function chain and the resources required for its deployment must be controlled by a single or multiple coordinated management and orchestration systems,which requires the network to abstract from specific SFC implementation technologies and provide advanced services Describe the function.Intent-Driven Network(IDN)is considered a cutting-edge method to achieve the required level of abstraction.It allows service providers and or customers to express service requirements and manage service lifecycles through the use of high-level specifications,regardless of the specific solutions and technologies used,because it allows to achieve based on "what you want to achieve" rather than "how" To declare the requested service.However,the current service function chain lacks an integrated deployment method under the background of the cloud network.In the face of complex business requirements,the necessary abstract mechanism is urgently needed to bridge the matching gap between SLA and SFC.This article focuses on two aspects of intent-driven intelligent translation and SFC deployment.The specific research content is as follows:First,this article studies an intent-driven intelligent translation method,which converts irregular and vague natural language into structured,machine-recognizable language.Through the Bi LSTM-CRF-based named entity recognition,the effective information in natural language expression is accurately obtained,the intention language is described based on the intention quintuple and the extended Bacchus paradigm,and the translation framework is realized through the finite state machine.Secondly,this paper studies an intentbased dynamic service function chain deployment framework.The intent-driven translationdeployment integrated framework consists of an intent engine,an SFC leader based on OpenStack-Heat,and a Neutron API extension for the service chain.The intent engine relies on the designed intent-driven intelligent translation framework for users to specify service requirements that have nothing to do with network topology and network technology.The intent-based service function chain leader realizes the modification and conversion of intent JSON expression to HOT expression through the intent refiner,so that Heat can carry out SFC deployment.Extend the existing Neutron API as a receiver to support the service link function.Finally,an intent-driven translation-deployment integrated platform was built,and the deployment efficiency was tested and analyzed.The experimental results show that the intent-driven dynamic service chain deployment solution studied in this paper solves the key technical problems of intent recognition and intent characterization in the process of intent input and translation,and obtains second-level Qo S service list response and SFC instantiation,which effectively reduces manpower costs and error rate.