Design And Realization Of Mass TCP Connection Management Cluster System For IoT Terminal

With the increasing scale of the Internet of Things(IoT)market,IoT terminals have developed rapidly and the number has increased exponentially.Connection management of massive terminal TCP network access through single server has been difficult,and multi-service nodes have become the inevitable choice for massive TCP connection management and data transmission services.However,in the case of multi-service node architecture and massive TCP connection,how to maintain the consistency of the connection information of massive terminals on different service nodes and ensure the reliability of upstream and downstream data transmission is an urgent problem to be solved.In response to this problem,the cluster coordination mechanism and the upstream and downstream message processing mechanism are designed and implemented in this thesis.At the same time,based on the above mechanism,a new type of TCP connection management cluster system is designed,which can meet the requirements of high concurrency,high load and real-time data transmission for massive terminal communication.This thesis,the key issues and performance improvement strategies is firstly studied in combination with the system business requirements,and the overall design scheme and functional module division of the system are designed.Secondly,the system cluster coordination mechanism,upstream message processing mechanism,downstream message processing mechanism and asynchronous network connection management module are designed and implemented in detail.Based on the simple data model of Zookeeper and the variety of node types,the system metadata storage model is designed to realize the election of cluster nodes and the offline monitoring function,which ensures the consistency of data between cluster nodes.Based on the distributed publish-subscribe messaging middleware Kafka,the system provides a unified cluster upstream message service interface,and the service layer invokes the interface to obtain service processing data.Based on the high availability of the distributed call middleware GRPC and the cluster coordination mechanism,the data of different terminals can be quickly sent between different service nodes,and the downstream data transmission efficiency is improved.Implement system session sharing management mechanism,efficiently manage terminal connection and timing detection timeoutconnection,and maximize system resources.To ensure the correctness and security of data transmission and reception,the reasonable custom communication protocol was used to solve the problem of sticky packets and unpacking during data transmission and complete data encoding and decoding,encryption and decryption.Finally,the system is tested for function and performance,the function modules of the system work normally,the performance test results with more than 6000 show that the average response time of the system design scheme is 60% shorter than the Java NIO solution,and the throughput is increased by 53%,and a single server can handle 20,000 terminal high concurrent connection requests,achieving the system's intended target.