CAREER: Managing the Complexity of Modern Enterprise Networks
University Of Wisconsin-Madison, Madison WI
Investigators
Abstract
Modern enterprise networks run a variety of services and impose a variety of constraints on network communications, which require pain-staking configuration within the network. Today, there is no way to systematically quantify how complex a network's configuration is. This is important because the more complex the configuration is, the more error-prone and fragile the network becomes. In particular, extreme complexity makes it difficult to modify or upgrade the network in a consistent and accurate fashion. Intellectual Merit: The goal of the Hydra project is to develop techniques for estimating the complexity of enterprise networks and for automating the process of making sophisticated changes to networks, while keeping the network configuration simple, accurate and in tune with the enterprise-wide policies. The primary component of this project is a suite of complexity metrics which describe various aspects of a network's configuration and functioning, such as the "conservativeness" of policies and dependencies in the configuration state, in a simple and abstract fashion. In turn, these metrics will be used to build centralized software tools for automating the tasks of configuring, upgrading and planning enterprise networks. Another key component of the Hydra project is an empirical study of the complexity of existing networks based on real configuration data and an analysis of changes in network complexity over time. Broader Impact: The Hydra management framework will help minimize manual involvement in important network management tasks, and make network management and configuration more automated than it is today. Thus, Hydra can enhance the operational efficiency of enterprises and improve the availability of key services. The results from this project apply both to today's complex enterprises and to future clean-slate architectures. The results from this research will be incorporated into undergraduate and graduate curricula.
View original record on NSF Award Search →