Program State Monitoring And Error Recovery In Private Cloud Environment

In the Internet era,with the development of computer and network technology,numerous massive data processing tasks have been generated by many application scenarios.Cloud computing connects and integrates large-scale computing resources,thus provides elastic service.This kind of service provides computing resources(computation,storage,network)as public utilities(water,electricity)to users.Cloud computing is a sound solution to massive data processing problems.Meanwhile,rapid development of software technology results in more extensive usage and more complex functionality of computer programs,and also causes the probability of program faults to increase.It is of great significance that program faults should be identified and well-handled in practical work.Social network information is very valuable in the fields of user behavior research and emergent events analyzing.However,using computer program to collect social network information usually faces the problems such as enormous amount of data and extremely long execution time.This thesis thoroughly studies cloud computing technology along with program state monitoring and error recovery theory,analyzes the characteristics of data-collecting programs in detail,then designs and implements a social network information collecting system in private cloud environment.This thesis contains the following main aspects of work:1.Designs and implements a social network information collecting system in private cloud environment.The system has the following main functionality: communication service,job execution and monitoring,job dispatching and scheduling,and user interaction.Using the strong computing power of private cloud environment,the system is able to process jobs in a distributed and parallel manner,thus increases the efficiency of job execution.2.Implements state monitoring,anomaly detection,and error recovery mechanism for information collecting programs.This thesis snapshots information collecting program during execution to acquire its state data(system resource usage data and system call sequence),then inputs state data into a Na?ve Bayesian Classifier to implement state monitoring and anomaly detection.Also,this thesis analyzes the characteristics of information collecting programs,then uses application level checkpointing technique to checkpoint key application data,thus implements error recovery mechanism.3.Runs a series of tests on the communication service and job execution control functionality of the information collecting system,and also runs a series of tests on the anomaly detection and error recovery mechanism of information collecting programs.The results of the above tests verify that the information collecting system has sufficient capability of communication service,verify the correctness of the information collecting system's job execution control functionality,and also verify the correctness of anomaly detection and error recovery mechanism of information collecting programs.