Multiplex communications – Pathfinding or routing – Switching a message which includes an address header
Reexamination Certificate
1999-09-30
2003-03-04
Chin, Wellington (Department: 2731)
Multiplex communications
Pathfinding or routing
Switching a message which includes an address header
C370S396000
Reexamination Certificate
active
06529515
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to computer networks and more particularly to network management techniques for use in computer networks.
BACKGROUND OF THE INVENTION
The emerging next generation of routers for computer networks exhibit not only high performance, but also enhanced functionality, such as support for virtual private networks and quality of service (QoS) guarantees. In order to achieve this functionality, features such as per flow queuing and fast Internet Protocol (IP) filtering are generally incorporated into the router hardware. However, the management of a network comprised of such devices and efficient use of the enhanced functionality introduces many new challenges which are not adequately addressed by existing network management techniques.
The majority of conventional network management systems are typically centralized around some type of management station. In this type of centralized architecture, a manager queries the managed objects, builds a view of the network, and sends alerts if a problem is detected. The manager can also try and take corrective actions by sending configuration commands to network entities.
There are many drawbacks to the above-noted centralized architecture, and these drawbacks generally become more evident as a network grows in size and complexity. For example, as the number of controlled elements increases, the requirements for computational power from the management system and bandwidth from the network that connects it also increase. In addition, in a large network, some of the controlled entities are distant from the management station, such that control loops exhibit long delays, and control traffic wastes bigger portions of the network bandwidth.
In order to alleviate the above-noted scalability problems, various types of distributed control architectures have been proposed in recent years. See, e.g., G. Goldszmidt et al., “Distributed Management by Delegation,” 15th International Conference on Distributed Computing Systems,” IEEE Computer Society, Vancouver, Canada, June 1995; Y. Yemini et al., “Towards programmable networks,” IFIP/IEEE Intl. Workshop on Distributed Systems Operations and Management, October 1996; A. A. Lazar, “Programming telecommunication networks,” IEEE Network, 11(5):8-18, September/October 1997; and M. Zapf et al., “Decentralised snmp management with mobile agents,” Sixth IFIP/IEEE International Symposium on Integrated Network Management, IM'99, May 1999, Boston, Mass.
Most of these conventional approaches delegate some of the central management tasks to distributed software agents. This is very useful to alleviate the load from the manager, reduce network traffic, and shorten control loops. However, this type of delegation does not support a truly distributed agent system where agents can communicate with their neighbors to efficiently carry out distributed tasks. Such distributed tasks are important in implementing functions such as scalable self-healing in network operations, e.g., local rerouting around congested spots, or cooperated caching.
Many of the above-noted conventional approaches also use a distributed object paradigm that abstracts the implementation details. Abstractions, such as Common Object Request Broker Architecture (CORBA), Distributed Component Object Modeland (DCOM), Java Remote Method Invocation (RMI), are helpful in designing and building distributed agent systems, but they hide the true cost of the implementation details. As a result, conventional agent-based systems tend to be, in many cases, inefficient in their use of network resources, primarily in their use of bandwidth. These and other conventional network management approaches thus fail to provide efficient distribution of the management task in the network.
Another important issue in distributed agent systems for management and control is the location at which these agents reside. Clearly, this has a major impact on the performance since it effects the delay in the control loop. In general, existing software agent solutions assume the existence of available hosts to run their application-level programs. An optimal location for an agent would be in the router kernel where all the necessary local information is available, and action can be taken locally. However, such a solution has been deemed impractical due to the inability to interfere with router real-time operation constraints. Clearly, the closer the agent is to the controlled system, the better it can perform. Nonetheless, conventional distributed network management approaches have been unable to provide optimal agent location.
It is therefore apparent that a need exists for an improved distributed network management approach which overcomes the problems associated with the above-described conventional approaches.
SUMMARY OF THE INVENTION
The invention provides an active network approach to network management that simplifies the deployment of distributed network management applications in IP networks and other types of computer networks. In an illustrative embodiment of the invention, a distributed network management function is implemented in a computer network using a set of active nodes. Each of the active nodes comprises a router and a logically-separate active engine. In addition to performing conventional forwarding functions, the router in a given one of the active nodes diverts active packets associated with the network management function to the corresponding active engine for processing. The active engine supports one or more sessions, based at least in part on the active packets, for implementing at least a portion of the network management function. Each of the sessions supported by the active engine corresponds to a particular distributed task to be performed in the network, and has associated therewith a unique network identifier, such that different programs on different network nodes can belong to the same session. The router and active engine at a given one of the nodes may reside on the same machine, or on physically-separate machines.
In accordance with another aspect of the invention, the given active node may be configured to support a blind addressing mode. The blind addressing mode may be used to send designated packets from the active node via unspecified network nodes in a particular direction toward a destination node, such that a router in a first unspecified active node along a route to the destination node intercepts the packet and sends it to an associated processing engine of that node. In this manner, the active node can communicate with other active nodes of the network without knowing their specific addresses or locations.
Examples of network management functions that may be implemented using the active network approach of the invention include collect-en-route and report-en-route information gathering programs, and message dissemination applications with ad-hoc definition of a particular group of message receivers.
By utilizing active network techniques to distribute and execute network management applications, the invention provides efficient use of network resources, without increasing the complexity of application development. Advantageously, a computer network configured in accordance with the invention enables the safe execution and rapid deployment of new distributed management applications in a network layer. This active network approach can be gradually integrated into, e.g., an otherwise conventional IP network, and allows smooth migration from conventional IP to programmable networks.
REFERENCES:
patent: 5509123 (1996-04-01), Dobbins et al.
patent: 5951649 (1999-09-01), Dobbins et al.
patent: 6148411 (2000-11-01), Ichinohe et al.
G. Goldszmidt et al., “Distributed Management by Delegation,” 15th International Conference on Distributed Computing Systems, IEEE Computer Society, Vancouver, Canada, pp. 333-340, Jun. 1995.
Y. Yemini et al., “Towards programmable networks,” IFIP/IEEE Intl. Workshop on Distributed Systems Operations and Management, pp. 1-11, Apr. 1996.
A.A. Lazar, “Pr
Raz Danny
Shavitt Yuval
Chin Wellington
Lucent Technologies - Inc.
Pham Brenda
LandOfFree
Method and apparatus for efficient network management using... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for efficient network management using..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for efficient network management using... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3081134