Environments for active networks

Active networks have the potential to solve three fundamental problems in traditional (and current) networks: (a) service extensibility, (b) rapid reaction to network events, and (c) obtaining a disaggregated view of network state. Traditional networks expose a fixed set of network services, which are difficult to modify or customize. In contrast, active networks allow service extensibility by providing an environment for execution of nearly arbitrary agents (or applets) within the network. Such a framework can be used to build application level gateways for multimedia processing, such as transcoding and fusion/merging, within the network. A second advantage of active networks is their ability to rapidly respond to asynchronous network events because control algorithms can be co-located with network elements. Finally, traditional networks achieve scaling by aggressively aggregating network state. Applets embedded in the network see the disaggregated state, allowing finer-grained control. The authors describe some preliminary ideas in exploiting an active network framework for next generation networks. They support ideas by describing their experience in using an active network framework for network management.