Intelligent agents: a way to reduce the gap between applications and networks

Network management (NM) has been in some way an activity apart from any other carried on the networks. Monitoring networks and systems resources independently from how users and applications are using them, has proven to be an inadequate approach since in most of the situations it´s impossible to trace the relationship between network or systems faults and the applications affected by them. Furthermore, the degradation of network and systems resources can be considered as a fault from an application point of view, and there aren´t any form to be aware of this degradation/fault, since applications requirements are not known to the network management system (NMS), and even when it happens the network can apparently still be viewed as functioning. Currently, the management efforts are used mostly to diagnose the origin of the faults, or to notify the systems administrators of the presence of a fault which in our opinion could be perceived by the network manager at very late stage. In this article we propose a way to bring the management closer to application and user requirements. The approach relies on spreading intelligent agents through network segments or domains, these agents being the management front-ends responsible to: collect QoS requirements concerning application operation over the network monitor required resources, perform when possible corrective actions, or at least provide management staff with powerful data so that they can take the corrective actions faster than before. We expect to benefit from being aware of how network and systems are used to develop a proactive management in real time, without the burden of classical approaches