Facilitating efficient and reliable monitoring through HAMSA

Monitoring is a fundamental building block of any network management system. It is needed to ensure that the network operates within the required parameters, and to account for user activities and resource consumption. In the SNMP paradigm, network management systems have been structured using a two-tier architecture with managers being thick clients, and the target agents being thin servers. This architecture may be unreliable at times since it depends on the management station having an access to the targets. Network distance between the manager and the network elements also imposes high overhead traffic, large processing overheads, and long control loops. To overcome these drawbacks, distributed network management architectures based on a middleware layer were proposed. However, such approaches suffer both from the need to modify network elements, and the high (and sometime hard to predict) overhead and complexity. In this paper we study a solution based on a lightweight middleware architecture that aims primarily at improving availability and efficiency of monitoring applications. We describe the Highly Available Monitoring Services Architecture (HAMSA), present its implementation details, and evaluate its performance. Specifically, we demonstrate how the system can be easily deployed and used for several monitoring applications. HAMSA allows a high level of availability and abstraction, with relatively low overhead.