Abstract: A system for deadlock recovery of distributed devices may include a processor and memory. The processor may transmit packets to a device, receive a pause message indicating that the packet transmission should be paused, and initiate a timer and pause the packet transmission in response to receiving the pause message. The processor may enter a deadlock recovery state if the timer reaches a timeout before a resume message is received that indicates that the packet transmission can resume. The processor may, while in the deadlock recovery state, drop packets that have a packet age that is greater than a threshold, and may exit the deadlock recovery state upon dropping a packet that has a packet age less than the threshold, or upon receiving the resume message. The processor may re-initiate the timer if the resume message has not been received, otherwise the processor may resume the packet transmission.

Abstract: A network device for processing data includes at least one ingress module for performing switching functions on incoming data, a memory management unit for storing the incoming data and at least one egress module for transmitting the incoming data to at least one egress port. The at least one egress module includes an egress scheduling module and multiple queues per each of the at least one egress port. Each of the multiple queues serve data attributable to a class of service, and the egress scheduling module is configured to service a minimum bandwidth requirement for each of the multiple queues and then to service the multiple queues to allow for transmission of a maximum allowable bandwidth through a weighting of each of the multiple queues.

Abstract: A system and method for adjusting an energy efficient Ethernet (EEE) control policy using measured power savings. An EEE-enabled device can be designed to report EEE event data. This reported EEE event data can be used to quantify the actual EEE benefits of the EEE-enabled device, debug the EEE-enabled device, and adjust the EEE control policy.

Abstract: A distributed switch architecture using permutation switching. In one embodiment, the distributed switch architecture facilitates connections between a plurality of ingress nodes and a plurality of egress nodes, wherein each of the plurality of ingress nodes and plurality of egress nodes are coupled to a plurality of ports (e.g., 40 gigabit Ethernet (GbE), 100 GbE, etc.). A plurality of crossbar switch modules are provided that are configured for coupling to a single output from each of the plurality of ingress nodes, and for coupling to a single input from each of the plurality of egress nodes. Permutations of connections for a crossbar switch module are defined by a permutation connection set that is stored in a permutation engine.

Abstract: Various methods and systems are provided for oversubscription buffer management. In one embodiment, among others, a method for oversubscription control determines a utilization level of an oversubscription buffer that is common to a plurality of ingress ports and initiates adjustment of an ingress packet rate of the oversubscription buffer in response to the utilization level. In another embodiment, a method determines an occupancy level of a virtual oversubscription buffer associated with an oversubscription buffer and initiates adjustment of an ingress packet rate in response to the occupancy level. In another embodiment, a rack switch includes an oversubscription buffer configured to receive packets from a plurality of ingress ports and provide the received packets for processing by the rack switch and a packet flow control configured to monitor an occupancy level of the oversubscription buffer and to initiate adjustment of an ingress packet rate in response to the occupancy level.

Abstract: A system for multicast switching for distributed devices may include an ingress node including an ingress memory and an egress node including an egress memory, where the ingress node is communicatively coupled to the egress node. The ingress node may be operable to receive a portion of a multicast frame over an ingress port, bypass the ingress memory and provide the portion to the egress node when the portion satisfies an ingress criteria, otherwise receive and store the entire frame in the ingress memory before providing the frame to the egress node. The egress node may be operable to receive the portion from the ingress node, bypass the egress memory for the portion and provide the portion to the first egress port when an egress criteria is satisfied, otherwise receive and store the entire multicast frame in the egress memory before providing the multicast frame to an egress port.

Abstract: The systems and methods disclosed herein allow for a switch (in a packet-switching network) to track buffer statistics, and trigger an event, such as a hardware interrupt or a system snapshot, in response to the buffer statistics reaching a threshold that may indicate an impending problem. Since the switch itself triggers the event to alert the network administrator, the network administrator no longer needs to sift through mountains of data to identify potential problems. Also, since the switch triggers the event prior to a problem arising, the network administrator can provide remedial action prior to a problem occurring. This type of event-triggering mechanism makes the administration of packet-switching networks more manageable.

Abstract: Disclosed are various embodiments that provide an architecture of memory buffers for a network component configured to process packets. A network component may receive a packet, the packet being associated with a control structure and packet data, an input port set and an output port set. The network component determines one of a plurality of control structure memory partitions for writing the control structure, the one of the plurality of control structure memory partitions being determined based at least upon the input port set and the output port set; and determines one of a plurality of packet data memory partitions for writing the packet data, the one of the plurality of packet data memory partitions being determined independently of the input port set.

Abstract: A computer-implemented method for implementing a congestion management policy, the method including, determining a minimum congestion state for a first queue, based on a minimum guarantee use count of the first queue, determining a shared congestion state for the first queue, based on a shared buffer use count and a shared buffer congestion threshold, wherein the shared buffer congestion threshold is further based on an amount of remaining buffer memory and determining a global congestion state based on a global shared buffer use count. In certain aspects, the method further includes implementing a congestion management policy based on the minimum congestion state, the shared congestion state and the global congestion state. Systems and computer-readable media are also provided.

Abstract: A system and method for hierarchical adaptive dynamic egress port and queue buffer management. Efficient utilization of buffering resources in a commodity shared memory buffer switch is key to minimizing packet loss. Efficient utilization of buffering resources is enabled through adaptive queue limits that are derived from an adaptive port limit.

Abstract: Disclosed are systems and methods for cut-through switching in port-speed-mismatched networks. Specifically, systems and methods are described in which data packets from an ingress device are paced, thereby matching the data rate of the ingress device with the data rate of the egress device.

Abstract: Processing techniques in a network switch help reduce latency in the delivery of data packets to a recipient. The processing techniques include speculative flow status messaging, for example. The speculative flow status messaging may alert an egress tile or output port of an incoming packet before the incoming packet is fully received. The processing techniques may also include implementing a separate accelerated credit pool which provides controlled push capability for the ingress tile or input port to send packets to the egress tile or output port without waiting for a bandwidth credit from the egress tile or output port.

Abstract: The systems and methods described herein allow for the scaling of output-buffered switches by decoupling the data path from the control path. Some embodiment of the invention include a switch with a memory management unit (MMU), in which the MMU enqueues data packets to an egress queue at a rate that is less than the maximum ingress rate of the switch. Other embodiments include switches that employ pre-enqueue work queues, with an arbiter that selects a data packet for forwarding from one of the pre-enqueue work queues to an egress queue.

Abstract: The various embodiments of the invention provide mechanisms to reduce headroom size while minimizing dropped packets. In general, this is done by using a shared headroom space between all ports, and providing a randomized delay in transmitting a flow-control message.

Abstract: Various example embodiments are disclosed. According to an example embodiment, a method may include determining, by a port processor, a buffer length based on an amount of data stored in a port controlled by the port processor, comparing the buffer length to a low-power buffer threshold, determining a link utilization based on a number of packets transmitted by the port, comparing the link utilization to a link utilization threshold, and placing the port into a low-power state based on the comparison of the buffer length to the low-power buffer threshold and the comparison of the link utilization to the link utilization threshold.

Abstract: An apparatus comprising a plurality of physical ingress ports configured to receive data, each data having a data type; a plurality of physical egress ports configured to transmit data; a memory configured to buffer data that has been received; a plurality of virtual routing devices, wherein each of the virtual routing devices is associated with a particular data type and each of the virtual routing devices is configured to: virtually buffer data associated with the respective data type, and regulate the quality of service provided to the respective data type; and a data manager configured to manage the receipt and transmission of data.

Abstract: A system and method for hierarchical adaptive dynamic egress port and queue buffer management. Efficient utilization of buffering resources in a commodity shared memory buffer switch is key to minimizing packet loss. Efficient utilization of buffering resources is enabled through adaptive queue limits that are derived from an adaptive port limit.

Abstract: Various example embodiments are disclosed. According to an example embodiment, a method may include determining, by a port processor, a buffer length based on an amount of data stored in a port controlled by the port processor, comparing the buffer length to a low-power buffer threshold, determining a link utilization based on a number of packets transmitted by the port, comparing the link utilization to a link utilization threshold, and placing the port into a low-power state based on the comparison of the buffer length to the low-power buffer threshold and the comparison of the link utilization to the link utilization threshold.

Abstract: Example methods and apparatus for hierarchical bandwidth management are disclosed. An example method includes, receiving a data packet included in a first data traffic flow having a first rate of traffic. The example method further includes marking the data packet with a first marker type if the first rate of traffic is less than or equal to a first threshold, otherwise marking the data packet with a second marker type. The example method also includes combining the first data traffic flow with a second data traffic flow having a second rate of traffic to produce a third data traffic flow having a third rate of traffic. The example method still further includes, if the data packet is marked with the first marker type, forwarding the data packet in the third data flow.

Abstract: Apparatus and methods for intelligent congestion feedback are disclosed. An example apparatus includes a data interface configured to receive data packets from a source endpoint via an intermediate node. The data packets include a field indicating whether data congestion for data being sent to the destination endpoint is occurring. The example apparatus also includes a timer. The example apparatus further includes a feedback loop interface configured to selectively enable a feedback loop to the source endpoint and to transmit congestion notification (CN) messages to the source endpoint over the feedback loop. Upon receiving a data packet indicating that congestion has occurred due to the data packets from the source endpoint to the destination endpoint, the destination endpoint is configured to set the timer to a preset time value; start the timer reverse counting from the preset time value to zero, enable the feedback loop and transmit the CN messages.