The Design And Implementation Of USB Disk-based Server-Cluster Deployment And Monitoring System

With the growing demand for cluster, the development of cluster solution has received increasing attention, especially in the field of small and medium-sized cluster for generic service. USB disk-based Server-cluster solution aims at small and medium-sized cluster project for the research and development of generic server clusters. The solution makes the deployment easier and realizes system's high availability. Based on the modular design concept, it provides a rapid development and feasibility of solution for the clustering software.The USB disk-based solution of server-cluster gives a three-layer logical structure: Customer Service Layer; Node Service Layer; Data storage Layer. The application software system for cluster is divided into seven core modules. They're user interface, load balancing module, service provider module, storage management module, data backup and recovery module, the overall database and monitoring subsystem for cluster.In the solution, monitoring subsystem, a complete system, is an independent group to specific cluster applications. With the help of the data acquisition module, data gathering module and data showing module, monitoring subsystem realizes a Web-based real-time monitoring service for cluster nodes. Additionally, monitoring service runs on some strategies to achieve the reliability and scalability of the system.The most notable characteristics of USB disk-based server-cluster is that it gives a USB disk-based deployment and installation for clusters. The program for deployment uses USB disk operation system integration technology, Image-based cluster deployment technology and Network Computer technology to realized the deploying and configuration all cluster nodes only with a specific USB disk.The solution is used into a Streaming media cluster server prototype system called Bluegrid. On the platform of this prototype system it conducted some performance tests for system deployment and monitoring subsystem. Test results show that the solution has good availability and scalability.