基于并行计算的系统运行日志采集方法研究

Research on collection method of system running log based on parallel computing

为满足日志服务的可扩展性需求,建立更加稳定的系统运行环境,提出基于并行计算的系统运行日志采集方法。利用已选定的性能指标,确定线程任务的并行量数值,通过分析异构体可移植性能力的方式,完成通用并行计算技术研究。在此基础上,建立日志采集框架,借助AQM分类器,得到准确的日志采集量计算结果,实现基于并行计算的系统运行日志采集方法研究。对比实验结果表明,与分布式采集策略相比,基于并行计算采集方法的QTR指标与WEP指标数值水平更高,可在满足日志服务可扩展性需求的同时,实现对稳定性系统运行环境的构建。

In order to meet the scalability requirements of log services and establish a more stable system operating environment,a method for collecting system operating logs based on parallel computing is proposed.Use the selected performance indicators to determine the value of the amount of parallelism of thread tasks,and complete the research on general parallel computing technology by analyzing the portability of isomers.On this basis,a log collection framework is established,with the aid of the AQM classifier,accurate log collection calculation results are obtained,and the study of system operation log collection methods based on parallel computing is realized.The comparative experiment results show that compared with the distributed collection strategy,the QTR index and WEP index based on the parallel computing collection method have a higher numerical level,which can meet the scalability requirements of the log service and realize the construction of a stable system operating environment.