The Design And Implementation Of Video Streaming Service Based On Hybrid Cloud

With the rapid development of Internet of Things(IoT)technologies,there's huge amount of data generated by devices on the edge of the network,in which video data occupies a considerable proportion.So far,the video data has not yet been fully exploited,limited by network bandwidth and insufficient video data storage and analysis conditions.Also,the centralized processing mode of traditional public cloud computing cannot fully support the real-time calculation of massive video data.However,it's more feasible to collect and analyze data in real time on the edge computing side,with the low-latency,high-security edge cloud environment combined with the IoT device.In addition,the results of the edge-side analysis can be integrated into structured data and transmitted to the public cloud,using the more powerful computing power of the public cloud for further data analysis and storage.Therefore,this thesis proposes a solution of "hybrid cloud combined with edge computing" for the scenario of real-time video stream analysis and processing.This thesis introduces the design and implementation of video streaming cloud service on hybrid cloud platform.This thesis firstly summarizes the requirements of the project and describes it in a standardized way using unified modeling language.Secondly,this thesis explains the overall architecture design of the system,combined with the software design and business logic.Finally,this thesis explains the implementation of the key parts of this system in detail.This thesis mainly focuses on the real-time video analysis and processing system on private cloud,including the framework of video stream processing application and cloud application management service.The application building module provides users with a framework for building video processing applications and some generic components.Users can quickly build applications for processing video streaming based on the framework and components.The framework adopts the Pipe-and-Filter architecture.The functional modules in the framework are implemented as components.Different components can be combined flexibly on pipelines to construct a video stream processing applications.The cloud application management services provide cloud computing environment and management services for video streaming applications on hybrid cloud platform,including application deployment management and resource management.The deployment and startup of the application in the system are based on the Docker container.The functions of application image downloading,deploying,monitoring,and managing are based on the Kubernetes system.