Research And Application Of Fabric Blockchain Performance Monitoring Technology

In order to apply the fabric blockchain technology to more practical scenarios,it is essential to study the performance testing and monitoring technology of Fabric.By using the performance of a Fabric SDK Node accessing the Fabric v1.3 network as the main monitoring and testing object,this thesis develops a more concrete performance monitoring and testing process which is based on the PTGM(Performance Testing General Model).The precise process includes determining performance indicators,selecting monitoring and testing tools,and analyzing performance influencing factors,etc.As for the selection of monitoring and testing tools,because of the inapplicability of the existing tools to the Fabric v1.3,this thesis proposes a load testing method and a monitoring method which are based on the characteristics of Node.js non-blocking I/O,event-driven,and asynchronous programming.And on the basic of these methods,a set of performance testing and monitoring tools are designed and implemented.The tools mainly have the following characteristics:(1)Versatility: All chaincode functions can be tested.(2)Fine-grained: Analyzing and counting the validity of undelayed and delayed transactions.(3)Flexibility: Realizing free combination of single/multiple commands,loads,and rounds of testing ways.(4)Real-time: The performance monitoring tool can monitor the change of the transaction generation rate of the blockchain network in real time.This thesis also analyzes multiple factors that may have an impact on the performance,including block size,endorsement policies,single-key and multi-key transaction writing methods,transaction size,TLS(Transport Layer Security)and non-TLS networks.In addition,the thesis uses the throughput and average response time as indicators to test these factors by using performance testing tools,and makes a comparison analysis of the specific impacts and causes.Meanwhile,the abnormal problems occurred during the experiment are analyzed and verified.Through these studies,developers can be helped to optimize local performance for the above influencing factors when developing and deploying Fabric applications.