Design And Implementation Of Service Call Middleware For Smart Pipe Gallery System

With the rapid development of the Internet of Things technology,more and more intelligent terminal devices have emerged in large numbers.How to efficiently,conveniently and low-costly manage the intelligent terminal devices connected to the cloud platform has become an urgent problem for the IoT cloud platform.Constructing a high-resource utilization,elastic and scalable distributed cloud platform has become one of the effective methods to solve this problem.This thesis researches the design and implementation of the service call middleware of the smart corridor system.The smart pipe gallery system runs on the IoT cloud platform built on Kubernetes,and the different business capabilities in the system are independent of each other.Within the Kubernetes cluster,you can use the service name and port to establish a TCP connection to make calls between services.Most of the services within the cluster are small data volume and high concurrent requests.The existing smart pipe gallery system uses the HTTP protocol.The heavy protocol header will affect the call efficiency between services.In the smart pipe gallery system,after the client initiates the request,the thread will block for waiting for the server's response to return,resulting in a waste of server resources.The popular service call frameworks,such as Dubbo,are not compatible with the features of Kubernetes.The topology-aware service routing introduced in this thesis cannot be applied to the Dubbo framework.This thesis combines the characteristics of the smart pipe gallery system,designing service call middleware,improving the internal service calling method of the distributed cloud platform cluster,improving the efficiency of the internal service call of the distributed cloud platform,reducing the server resource occupancy rate of the smart pipe gallery system,compatible with the cloud platform upgrades after the platform..According to the characteristics of the smart gallery system,this thesis gives a design and implementation scheme of service call middleware.This solution mainly improves the service call of the smart gallery system from the following points:(1)Custom communication protocols.In order to meet the characteristics of the small amount of call requests between services in the intelligent pipe gallery system,the problem of using HTTP protocol headers is made up by custom communication protocols,and the network transmission time of service call data is reduced,thereby reducing request response time.In designing the communication protocol,it draws on the advantages of some other communication protocols,comprehensively considers the characteristics of the smart pipe gallery system,and is compatible with the implementation of the link tracking function and the cloud platform's flow control upgrade function.(2)Multi-threaded technology.Use the Java thread pool to perform asynchronous tasks,avoid the overhead of time caused by thread creation,and shorten the response time of requests.The native multi-threaded component of Java is transformed to support the tracking of service call chain information to facilitate problem location and performance analysis.(3)Codec technology.Serialize and deserialize request parameter objects to reduce the size of data transmitted over the network.The parameter object encoding supports Protobuf and json.Protobuf s codec performance is better than json,which is suitable for the interface modification with higher call frequency.(4)Service registration and discovery.Introduced a new feature of Kubernetes,topology-aware service routing,which forwards client requests to the nearest server,avoiding forwarding requests to farther servers,shortening the path for requesting network transmission,and further reducing response time.