The Research And Design Of Resources Monitoring Framework In HA System

With more and more applications having been transformed onto the Internet ,the HA service in computer clusters plays a very important role. Among pluralities of advantages of clusters .redundancy of hardware and software is the most prominent, through which the computer systems are able to provide consistent service to their clients. By detecting status of nodes as well as status of service programs (daemons) in the cluster ,the HA can redistribute tasks and progresses among the nodes to balance the workload in the case of services or nodes being too busy or failed.The HA in a computer cluster is defined that computers cooperate with each other , ie., if one node is failed, other computers can take over all tasks which are running on it originally. Therefore, the entire system may still available and users are not aware of switching of node. The components in a HA system should support RAS features (Reliability,Availability and Supportability), including resources monitor, logic management and system health monitor etc.In fact, HA is a wide domain and a well-designed HA system should embrace reliable communication sub-system, nodes management, quoram sub-system and simultaneous controlling sub-system etc.To achieve the goals above, in this dissertation, it focuses on the universal presentation of resources, resources monitoring, resources failover, logic system and HA in database systems.