Research On Automatic Data Processing Method Of Radio Heliograph Based On DOCKER

Today,astronomical software can satisfy various requirements of astronomical data processing.However,there are still many problems in practical application,such as complicated to operate,tough to deploy and so on.With these challenges,a method combinating cloud with container technologies can provide a platformed solution.MingantU SpEctral Radioheliograph(MUSER)has a large amount of data processing requirements every day.Although OpenCluster-a distributed solution-is introduced,there are still many inconveniences in operation and deployment.This paper focus on the practical problems in MUSER data processing,combined with Docker – a kind of lightweight containerization technologies,designed a astronomical private cloud of software service model based Docker.The main works of this paper is as follows:1.Aiming at difficulties of MUSER data processing applications in a singlemachine environment,such as too many sophisticate dependencies,needs of multiple computing mode support and so on,this paper studied the encapsulation method of MUSER software based on Docker container,and further studies of deployment of MUSER container in different computing modes has been done.Finally,the advantages of the MUSER container in the data processing compared with the virtual machine are verified by the experiment.2.Aiming at difficulties of MUSER data processing application in a cluster environment,such as complex to construct,insufficient in scalability and so on.Further more,this paper do more researches in Docker container clustering technology and solve the key technologies based on the astronomical distributed computing framework – OpenCluster.Finally,an automatic processing platform for astronomical data is implemented,which can be deployed and extended automatically,and corresponding tests are carried out to verify its availability and performance;3.Aiming at the requirements of dynamic extension of computing node in astronomical data processing,the automatic data processing platform implemented was improved,and a load predicting extension strategy based on exponential smoothing algorithm was added.It is verified by tests that this predictive expansion strategy has better adaptability to general loads than the responsive expansion strategy.Based on the above research,the work described in this paper will provide a reference for the future construction of Docker-based astronomical private cloud services.