Research And Implemention Of The Intelligent Scheduling And Execution Monitoring For Web Service

With the rapid development of the Internet application technology, the Educationinformatization, which based on network and information technology, becoming moreand more popular. Education informatization greatly benefit teachers and students ontheir daily work and learning, as well as teaching quality, facilitate unifiedmanagement and resource sharing.China Education Knowledge Service Cloud came into being in this rapiddevelopment. It deploy services on a private cloud computing platform based onOpenStack, with the rapid growth of users and services, the system faced massivetasks concurrent request, leading to a poor user experience on slow response to taskrequest. The task of this paper is to design a reasonable and transparent servicescheduling strategy, to optimize the over all system performance. Module-basedservice scheduling techniques has advantages in integrating heterogeneous systems,transparent access and scalability, by applying flexible service scheduling algorithms,module can meet different application demands of service scheduling, and the systemcan become more scalable, configurable, and robust.This paper improved the scheduling module of the original system on the privatecloud platform, includes the following:1. In order to overcome single point scheduling failure of the original schedulingsystem, this paper designed a hierarchical scheduling policy. The top layer of thesystem manages and stores all user requests, the scheduling layer schedule tasksaccording to settings. The scheduling system adopts a double backup strategy, backupscheduling will continue the scheduling service when the master scheduling wentwrong.2. For each internal scheduling module, constructed a scheduling policy based onboth sides of the service——double-sides scheduling algorithm. Establishing a taskrequest priority queue to distinguish the task request strength, building serviceattribute hierarchical model base on dynamic hierarchical QoS ontology, taking intoaccount Cpu utilization, memory utilization, network Error_packet, the number ofprocesses, system response time, cost, reliability as parameters to measure serviceperformance. A mathematics model for service selection and dynamic scheduling issetup based on the hierarchy analysis.3. The initial value of the weight for service comprehensive ability of thecalculation model may not appropriate, it need adjust according to the actual operation.This process divided into two sides: On the one hand, the service executionmonitoring, including monitoring service quality attributes and virtual machinerunning condition; On the other hand, designing the self-learning algorithm of this model, improve and update the model base on the running data continuously, to getthe optimal weight distribution.