Design And Implementation Of Computing Platform For Real-time Monitoring System

The information age is developing rapidly,the deployment mode of a single server has been difficult to meet the business needs of large and medium-sized enterprises,and the deployment mode is replaced to cluster mode.As the server cluster grows in size,the operation and maintenance of the cluster faces challenges.Operation and maintenance relies on monitoring,and computing is the most important part of the monitoring life cycle.Today's operation and maintenance environment directly puts forward new requirements for the functions,cost,performance,real-time and reliability of the monitoring computing platform.After in-depth analysis of the computing requirements of server cluster monitoring,it is found that big data and other related technologies can be applied to the field of monitoring computing,that is,multiple computers are used to perform parallel computing on massive monitoring raw data.Start with the popular open source parallel big data computing framework Hadoop MapReduce and Spark,analyze their design ideas and basic implementation,as well as implementation and optimization of IO,parallelization and fault tolerance.Then introduce the design and implementation of a high performance computing platform suitable for use in a monitoring environment.Driven by the computing requirements of the monitoring field,linked with the open source parallel computing framework above,the architecture design and basic working principle of the computing platform are derived,and then introduce how the computing platform builds series of function plug-ins for solving different monitoring problems based on its plug-in design,so as to realize the corresponding monitoring calculation function.After the basic function is implemented,starting from performance optimization,design and implement the thread and scheduling,input sharing and fault tolerance of the computing platform.After researching and testing the computing platform,it is found that the platform has basically met the computing needs in the monitoring field-whether it is the basic computing function of monitoring,or features based on performance considerations such as thread scheduling,task management,plug-in design,input sharing,have all basically met expectations.