Research On IoT Service Generation And Runtime Fault Tolerance Based On Flink

The business system can improve the responsiveness and flexibility of the system by integrating the Internet of Things(IoT)capability to perceive the real-time conditions of the physical environment.By extracting valuable business events from IoT data and encapsulating them into software services with various functions,IoT services can effectively reduce the complexity associated with utilizing IoT data.In the research on the practical application of IoT services,two key problems are encountered:IoT data has the characteristics of multi-source,heterogeneous,and massive.How to implement IoT services with low code and maintenance of massive services? IoT services and their operating environments have dynamic characteristics.How to improve their reliability and ensure the quality of IoT services? To address these challenges,this paper employs Apache Flink as the running platform for IoT services and proposes an automatic generation engine for IoT services.Additionally,a faulttolerant optimization method for IoT service runtime based on Flink is introduced.The main research content of this paper is as follows:1)IoT service automatic generation engine.Aiming at the problems of low code generation,running and maintenance of IoT services,this paper designs the implementation method of the engine.The approach involves using code generation techniques based on templates and models to generate the corresponding service code,which is then automatically submitted to a Flink cluster server.Finally,the correctness of the engine is validated through experiments.2)An on-demand dynamic checkpoint fault-tolerant method.Aiming at the problems of IoT service quality and operation stability,an on-demand dynamic checkpoint fault-tolerant method is proposed based on Flink.This method calculates the recovery delay in real time according to the data fluctuation rate;when the delay exceeds the threshold,the checkpoint operation is actively triggered to avoid high endto-end delay and recovery delay.In order to implement this method,the source code of Flink has been modified,and an interface for actively triggering checkpoints has been added.The experiment found that compared with the traditional static checkpoint mechanism,the operator efficiency of the system has increased by up to 11.9%.3)The prototype of IoT service platform system was designed and implemented.The platform takes the service automatic generation engine as the core,realizes a series of functions such as automatic service generation,automatic running and operation,and is connected to the workflow system and service library management system,using specific examples of temperature alarm service and fire pre-warning service for verification system effectiveness and performance.