Design And Implementation Of Video Conference Syetem Based On Dynamic Scaling Of Resources

Nowadays,with the continuous improvement of network bandwidth and network speed,the continuous popularization of 4G and 5G networks,and the continuous optimization of the performance of video and audio coding technologies,it is possible to transmit high-quality multi-party audio and video streams in real-time.Besides,people's demand for video information is constantly increasing.No matter colleges or enterprises,there are many application scenarios,such as online video conferences,online academic discussions,and online education.There is considerable demand in the market.In such scenarios,WebRTC technology makes it easier to obtain video streams.Therefore,a video conference system based on WebRTC has great application scenarios.In this paper,the WebRTC framework,container technology,and video conferencing system were investigated in detail,and a video conferencing system capable of dynamically scaling resources was designed and implemented.At present,most video conferencing systems on the market use traditional business architectures and do not use microservice architectures to provide services,resulting in poor system fault tolerance and single functions.The media service architecture in the market also mostly uses the stand-alone mode,which cannot effectively solve the situation of sudden traffic and waste of resources,and the deployment cost is relatively high.This article uses Kubernetes and container technology to apply to the entire video conferencing system to make the system containerized and serviced.The HPA module of Kubernetes is used according to the indicators collected by the Prometheus monitoring platform to the microservice module container of the conference management system and the container in the media service architecture.Carry out dynamic shrinkage and expansion,and dynamically scale the interface of microservice modules and media service resources.The paper describes the design and implementation of the entire system mainly divided into two parts: video conference management background and media service architecture.The video management background is mainly divided into the following modules to elaborate detailed design and implementation: user management module,address book module,conference management module,and a notification module.The user management module mainly includes functions such as login and logout and is used for subsequent interaction with other functional modules.The address book module is mainly for group management,friend management,and group management of friends.The conference management module includes conference creation,cancellation,personnel addition and deletion,and conference status management.The notification module mainly includes functions such as internal message notification,a short message,and email notification.The detailed design and implementation of the media service architecture mainly include Docker container production,dynamic load balancing implementation,Kubernetes container scaling up and down,Prometheus container monitoring system integration,and container automation continuous integration.Through the dynamic scaling function of media resources and the dynamic load balancing service based on the secondary development of Nginx,a highly concurrent media service architecture was proposed and implemented,and the architecture was verified.The system tested the conference management system according to the requirements of software engineering and measured the system performance of the media service architecture by collecting Qo E indicators.The media service architecture supports SFU and MCU network modes,container dynamic scaling,and dynamic load balancing.When the resource demand fluctuates greatly,it has good service support.By monitoring the CPU and bandwidth indicators of the entire cluster,compared with the media service architecture in stand-alone mode,when resources are doubled,performance is also doubled,and the overall performance of the system linearly stacks with the number of containers.In this test environment,3 nodes can accommodate about 100 people to conduct a video conference at the same time.