Research And Implementation Of Cloud Platform Computing And Storage Performance Evaluation Oriented To NFV

Traditional telecom operators have proposed NFV(Network Function Virtualization),claiming to replace traditional network elements by deploying software on the cloud platform.The telecommunication services in the NFV scenario have high throughput and real-time requirements,and require high performance on the cloud platform.In order to verify whether the cloud platform can support the upper layer services,this paper evaluates the computing and storage performance of the cloud platform.This paper analyzes and studies the industry's existing cloud platform evaluation indicators and systems,summarizes the characteristics of the combination of telecom services and cloud platforms,and proposes to measure the computing and storage performance of the cloud platform from four aspects,namely,the capacity of the cloud platform,Basic performance,business performance and agility,proposed refinement indicators and testing tools for each part.According to the service performance,PDCP(Packet Data Convergence Protocol)in the mobile communication network is chosen as the benchmark of service performance.Analysis of its encryption and decryption,integrity protection and header compression functions,from the open source OpenGL implementation of the OpenAirInterface PDCP layer code decoupling,as an independent test load.Researching Yardstick,an open source performance testing framework,re-developing it,integrating testing tools into it,writing test cases,designing test schemes,and implementing an automated cloud computing evaluation and storage performance evaluation system for NFV.Build OpenStack cloud platform as experimental environment,use influxdb,grafana to build test result display platform.Run automated evaluation system for experiments and tests.