Filtering and Transformation for High-Volume 
XML Message Brokering


Research Papers

Related Work




Network Monitoring


Distributed monitoring systems monitor large numbers of nodes connected by wide-area networks. Based on the requirements of front end users (e.g. system administrators or computer users), they collect statistics on different aspects of the state of the nodes, e.g. CPU, memory, disk, or bandwidth utilization. Such a monitoring system takes the form of a hierarchical structure: A monitor may send information to subsequent monitors where the statistics on a larger network is aggregated. Each monitor performs several tasks: It pulls information from the nodes it is monitoring and its upstream monitors; It passes the merged information to downstream monitors, when requested; It may also compose new messages based on the requirements of the front end users.

Front end users may have varying requirements. For example, some require detailed information on CPU utilization, while some others focus on network statistics. Furthermore, they may require the same information to be updated on a different frequency basis. Therefore, a monitor needs to be implemented by a sophisticated message broker that takes the specifications of front end users, transforms an XML message containing a tremendous amount of information on many nodes to customized messages, and delivers these messages to the front end users based on the required frequency.