Research And Implementation Of Stream Computing Framework For Application Performance Management

As an important branch of IT technology, the application performance management (APM) system helps enterprises to ensure the availability,reliability and performance of key applications of their business. The data collected by the application performance management system enters the processing system in a "flow" manner, which requires complicated parallel computation in real time. Therefore, how to build low-latency, high-throughput and continuous and reliable operation of data "flow" is the a hot area of research for application performance, but also a challenging subject.However, the current market does not have a unified framework for the application of performance management. Existing stream computing frameworks don't support hot plug for data processing module, and the features such as computing transmission overhead, low utilization of system resources are not friendly to application performance management.is not friendly. In this paper, a flow-based computing framework named sFrame for application performance management is studied and designed based on the deep analysis of application requirements of stream data processing in application performance management system. At the same time, the task scheduling algorithm in the framework is studied and optimized, which ensures the high availability and load balance of the system while guaranteeing the rationality of task scheduling.This paper first introduces the background and significance of the flow-based computing framework for application performance management, and this paper proposes the functional and non-functional requirements of the sFrame framework according to the requirements of the application performance management system for streaming computing.Then this paper gives the design of core functions, interface designs and detailed interactive flows of sFrame. Finally, according to the requirements analysis and summary design, this paper designs the flow calculation framework for application performance management, gives the definition of the main classes, and completes the code implementation and tests. At the end of the paper, the related work of this paper is summarized and the future research direction is put forward.