Abstract: An apparatus and method for performing congestion management in a switch or router is disclosed. The apparatus acquires and or generates performance metrics for a plurality of switching module coupled via a switching fabric, generates a performance index based on the attributes from the switching modules, and allocates switching fabric bandwidth based on a weighted combination of the performance metrics. The performance metrics may include past and presents values of some attributes as well as predicted values.

Abstract: A high speed switching device including an aggregator, a plurality of intermediate ports, and ingress packet processing resources, is disclosed. The aggregator includes an external switch port operatively coupled to a transport medium, and a demultiplexer that divides an ingress data stream into a plurality of intermediate streams. The plurality of intermediate ports transmit the intermediate streams to ingress packet processing resources. The ingress packet processing resource may include one or more classification devices for parsing, switching or routing, and queueing packets of the ingress stream, prior to transmission to a switching fabric. The switching device optionally includes an egress processor for multiplexing egress data from the switching fabric. Intermediate port traffic may be dynamically modified to minimize load imbalance present in an ingress switching device and or egress switching device.

Abstract: Apparatus, and an associated method, for facilitating communication of data in a multi-stage communication network. A hash value generator generates hash values. A seed value is used in the generation of the hash values. As traffic flow conditions change, feedback is provided, and the seed value is adjusted or changed. And, the manner by which the hash values are generated is changed, responsive to the change of the seed value. Redistribution of traffic flow is provided, thereby to reduce, or compensate for, unbalanced traffic flow in the network.

Abstract: An apparatus and method for performing congestion management in a switch or router is disclosed. The apparatus acquires and or generates performance metrics for a plurality of switching module coupled via a switching fabric, generates a performance index based on the attributes from the switching modules, and allocates switching fabric bandwidth based on a weighted combination of the performance metrics. The performance metrics may include past and presents values of some attributes as well as predicted values.

Abstract: Lookup scheme in which a tuple representing a plurality of flow properties is parsed into multiple subtuples for application in recursive lookups. A first subtuple including a first subset of bits from the tuple is applied to the flow information database and returns a result including a nickname having a smaller bit count than the first subtuple. A second subtuple including a second subset of bits from the tuple and the nickname are combined and applied to the flow information database. The lookups continue until a result indicates that no recursion is required. The final lookup result includes flow information applicable to one or more of modifying, enqueuing or forwarding the packet. A truncated lookup capability enables common processing across a group of distinct flows having common flow properties.

Abstract: A system and method of performing distributed traffic engineering is provided. A network of nodes coupled to a central management module is created. Traffic engineering functions are distributed between the central management module and at least one of the nodes. A feedback regarding an offending source is sent from the at least one of the nodes to the central management module or another one of the nodes.

Abstract: A method and apparatus for accomplishing source learning in a data switch of the type having a plurality of switching modules where each supports one or more external network devices and a backplane interconnecting the switching modules. Each switching module has logic resident thereon for performing distributed source learning, including configuring unknown source addresses “seen” in inbound packets and for notifying the other switching modules that such source addresses were “seen” on a port thereof. Address-port associations are thereby configured on the switch using distributed logic, i.e. without intervention by a centralized management entity. In regard to configuring destination addresses—when a destination address is unknown, packets are delivered over a multicast queue until the destination address is found. Once the destination address is found, a method of flow integrity is used to avoid out of order packet delivery when the device transitions from using a multicast flood queue to a unicast queue.

Abstract: A high speed switching device including an aggregator, a plurality of intermediate ports, and ingress packet processing resources, is disclosed. The aggregator includes an external switch port operatively coupled to a transport medium, and a demultiplexer that divides an ingress data stream into a plurality of intermediate streams. The plurality of intermediate ports transmit the intermediate streams to ingress packet processing resources. The ingress packet processing resource may include one or more classification devices for parsing, switching or routing, and queueing packets of the ingress stream, prior to transmission to a switching fabric. The switching device optionally includes an egress processor for multiplexing egress data from the switching fabric. Intermediate port traffic may be dynamically modified to minimize load imbalance present in an ingress switching device and or egress switching device.