Computer networks (such as the Internet and the Global Information Infrastructure) have become critical for education, research, business, and military operations. Recent advances in network infrastructure technology (such as gigabit Ethernet, ATM and Ipv6) have enabled the development of high performance local area and wide area networks. Efficient management of these networks is essential. However, existing algorithmic methods for managing networks have not matured to the point where performance bottleneck and fault detection, isolation, correlation, and correction can be automated scalably. Thus, it is crucial to build efficient and user-friendly network monitoring, visualization, and control (NMVC) systems. The proposed effort is aimed at the design, prototype implementation, and demonstration of a highly scalable NMVC system with advanced algorithmic and human-in-the-loop capability. This capability allows network administrators to calibrate and fine-tune network and application parameters in real-time according to observed traffic patterns. The goal of the NMVC system is to ensure adequate quality of service to network users, while maintaining high network resource utilization. The proposed system leverages the following innovations to meet its objectives: * An inexpensive, yet high performance network monitoring and observation probe that does not interfere with high speed network traffic. This probe can be programmed to collect pertinent information at any layer in the protocol hierarchy (including application-specific information). * An efficient probing mechanism for endsystems (i.e., workstations and servers) that can collect relevant performance satistics from both applications and kernel resident protocols (such as TCP/IP in Unix). The endsystem probes introduce minimum interference with protocol operation and can be dynamically and remotely invoked and configured for the information of interest. * A highly flexible and highly efficient software network man agement agent design that extends the functionality and scalability of the network probe by filtering data based on programmable criteria (such as protocol header attributes or application-level frames). This agent can be configured, installed, and controlled remotely at network nodes by administrators or automated management applications. * Efficient online event ordering algorithms that can help synthesize and display a consistent view of network health, status, and performance using events collected throughout a large-scale high speed network. * A View Choreographer that allows management applications and administrators to specify the mapping of network events to higher-level events and to visualization objects and updates. The Choreographer uses semantically-guided filtering algorithms to enable low latency updates of displays that present an accurate real-time snapshot of current network state. * A visualization system that is highly configurable and can support multiple simultaneous real-time display views of a large-scale network. * Algorithmic and interactive steering mechanisms that perform critical network management functions such as protocol parameter tuning for better end-to-end performance, bandwidth enforcement, fault detection and recovery, and quality of service monitoring and control. * A prototype system to be demonstrated on a high speed IP/ATM testbed running a variety of applications (including multimedia teleconferencing systems and high speed digital imaging).
