Capacity enhancement of a direct communication link using a variable redundancy delivery network. An estimated information rate between a source node and a terminal node may be partitioned into a first information rate provided via the direct communication link and a second information rate to be provided via the variable redundancy delivery network. One or more parameters of the variable redundancy delivery network may be calculated to provide the second information rate based on a non-uniform probability density of messages requested by the terminal node. Capacity and reliability of storage media devices in the variable redundancy delivery network may be traded off to provide the second information rate. The variable redundancy delivery network may implement various coding schemes and per-message coding rates that may be determined based on the non-uniform probability distribution of the source message library.

Packets received by a network switch device from upstream network devices, coupled to respective ones of a plurality of ports of the network switch device, are temporarily stored in an internal memory of the network switch device. In response to detecting congestion in the internal memory of the network switch device, a flow control engine triggers, during respective timeslots of a timing schedule and while the flow control engine continues to monitor congestion in the internal memory of the network switch device, transmission of respective flow control messages via different subsets of ports, among the plurality of ports, to control flow of packets from different subsets of upstream network device, among the plurality of upstream network devices, to the network switch device so that flow control is distributed over time among upstream network devices of the plurality of upstream network devices.

A flow control method includes: when congestion is detected, determining, by a first switching device, a key flow from a plurality of data flows; generating a back pressure message including a flow attribute value of the key flow; sending the back pressure message to an upstream device of the key flow; and pausing, by the upstream device of the key flow, sending of the key flow, where the back pressure message has no impact on sending of another data flow other than the key flow by the upstream device of the key flow. The present disclosure further provides a switching device that can implement the flow control method.

A method for universal scaling of software network functions involves receiving, at a switch of a network, a batch of data units during a first period. The network further includes one or more network function (NF) instances of an NF service, and a scaling controller. The switch transmits one or more units of data during the first period to an NF instance of the NF service. An estimated maximum safe data unit rate is determined for the NF instance, and a representative safe data unit rate is determined for the NF service. A total number of data units designated to be received by the NF service during the first period is determined, and a total number of NF instances of the NF service to be provisioned in the network is determined at the scaling controller using the estimated total number of data units and the representative safe data unit rate.

A network adapter includes a receive (RX) pipeline, a transmit (TX) pipeline, hardware-implemented congestion-control circuitry, and a congestion-control processor. The RX pipeline is configured to receive packets from a network and process the received packets. The TX pipeline is configured to transmit packets to the network. The hardware-implemented congestion-control circuitry is configured to receive, from the TX pipeline and from the RX pipeline, Congestion-Control (CC) events derived from at least some of the packets transmitted to the network and from at least some of the packets received from the network, and to pre-process the CC events. The congestion-control processor is configured to receive the pre-processed CC events from the congestion-control circuitry, and to throttle a transmission rate of the packets transmitted to the network by the TX pipeline responsively to the pre-processed CC events.

Systems and methods for using multiple CE (congestion experienced) flags in both FECN (forward explicit congestion notification) and BECN (backward explicit congestion notification) in a high performance computing environment. An exemplary method can provide a first subnet comprising a plurality of switches, a plurality of host channel adapters, and a plurality of end nodes. The method can receive, at an end node attached to a host channel adapter, an ingress packet from a remote end node, wherein the ingress packet traversed at least a portion of the first subnet prior to being received at the end node. The method can, on receiving the ingress packet, send a response message from the end node attached to the host channel adapter to the remote end node, the response message indicating that the ingress packet experienced congestion during the traversal of the at least a portion of the first subnet.

A network adapter includes a receive (Rx) pipeline, a transmit (Tx) pipeline and congestion management circuitry. The Rx pipeline is configured to receive packets sent over a network by a peer network adapter, and to process the received packets. The Tx pipeline is configured to transmit packets to the peer network adapter over the network. The congestion management circuitry is configured to receive, from the Tx pipeline and from the Rx pipeline, Congestion-Control (CC) events derived from at least some of the packets exchanged with the peer network adapter, to exchange user-programmable congestion control packets with the peer network adapter, and to mitigate a congestion affecting one or more of the packets responsively to the CC events and the user-programmable congestion control packets.

A memory device includes a memory component that stores data and a processor. The processor may receive requests from a requesting component to perform a plurality of data operations, generate a plurality of packets associated with the plurality of data operations, and continuously transmit each of the plurality of packets until each of the plurality of packets are transmitted. Each of the plurality of packets after the first packet of the plurality of packets is transmitted on a subsequent clock cycle immediately after a previous packet is transmitted.

Technology is disclosed herein for monitoring a network path. In an implementation, a device on a network path obtains a burst capacity of the network path, determines a round trip time associated with a burst of traffic sent over the network path, and determines a predicted throughput of the network path based at least in part on the burst capacity of the network path and the round trip time of the burst of traffic.

A device for determining oversubscription of a first virtualized network function (70) in order to enable flow control in the virtualization of at least one node in a communication network comprises a first virtualized network function (70) having packet handling resources comprising at least one queue (108, 112, 114, 120) and being a downstream network function located downstream from at least one second upstream network function in a packet flow between the network functions. The device comprises flow control functionality (79) set to monitor at least one queue (108, 112, 114, 120), determine if the first virtualized network function (70) is oversubscribed based on the monitoring, and generate, in case the first virtualized network function is determined to be oversubscribed, an instruction for at least one upstream network function to change its transmission rate to the first virtualized network function (70) in order to reduce the oversubscription.