Research And Implementation Of CDCloud Desktop Cloud Automatic Expansion Method

With the continuous development of cloud computing technology,desktop cloud has attracted more and more attention due to its many advantages compared with traditional desktop infrastructure construction and management,and has become an important direction of cloud computing research and application.The automatic expansion capability of the desktop cloud is directly related to the performance and user experience of the entire desktop cloud system,and is an important research content in the key technologies of the desktop cloud.At present,most of the research on automatic expansion of desktop cloud only focuses on the scalability of back-end computing resources,but from the perspective of the whole system,there are few comprehensive considerations in terms of system architecture and other aspects combining both front-end and back-end.This paper studies the automatic expansion method of desktop cloud from the aspects of scalable system architecture,forecast-based resource expansion,and desktop access protocols supporting expansion,and the researched automatic expansion method is implemented and applied on the CDCloud desktop cloud platform independently developed by the laboratory team.First,this paper proposes and designs a desktop cloud architecture that supports an automatic expansion process.Aiming at the expansion difficulties caused by the high coupling of hardware and software resources in the traditional desktop cloud access method,the frontend of the architecture realizes non-fixed client access independent of the endpoint device and its operating system by customizing the network startup and client system,which improves the flexibility of front-end extensions.Aiming at the expansion difficulties of traditional desktop cloud in desktop application distribution and management,based on the idea of container and virtual machine integration,a back-end architecture based on container application distribution is designed,which enhances the automatic expansion capability of the back-end.Secondly,based on the proposed and designed desktop cloud architecture,this paper researches and gives a prediction-based resource expansion method,designs an automatic expansion strategy,and realizes the automatic expansion of resources.In the automatic expansion strategy,the load forecast is based on the exponential smoothing in the time series,and a dynamic exponential smoothing load forecasting model is established to make the forecasting accuracy higher;when determining the expansion amount,the expansion amount is established by analyzing the relationship between resource allocation and load The model uses the feedback control method to calculate the performance attenuation coefficient caused by resource competition,and realizes the estimation of the amount of virtual machine resources for pre-expansion judgment during application distribution.Thirdly,aiming at the problem that the SPICE protocol architecture depends on the QEMU/KVM virtualization platform,which limits automatic expansion,this paper proposes a method to improve the native SPICE protocol architecture with the X11 SPICE protocol.By integrating the QXL drive device and SPICE Server on the server side into the virtual machine and using the network sound service solution to fill the gap in X11 SPICE protocol sound transmission,the direct interaction with the client is realized,and then eliminates automatic expansion limitations caused by the SPICE protocol’s dependence on virtualization platforms.Finally,the testing is carried out by combining the actual physical environment and simulation.Simulation testing is applied to the load prediction model,expansion amount model,and automatic expansion strategy;the access method,automatic expansion,application distribution pre-expansion,and the improved SPICE audio and video transmission performance are tested under the CDCloud desktop cloud platform.The experimental results show that the automatic expansion method proposed in this paper meets the automatic expansion requirements of the CDCloud desktop cloud platform,and has the value of further promotion,application and research.