61 results about "Head-of-line blocking" patented technology

A low latency switch architecture for high performance packet-switched networks which is a combination of input buffers capable of avoiding head-of-line blocking and an internal switch interconnect capable of allowing different input ports to access a single output simultaneously.

Methods, systems, and computer program products for killing prioritized packets in multiple queues using time-to-live values to prevent head-of-line blocking. In one example, a method for scheduling prioritized packets in queuing system includes receiving a plurality of packets having a plurality of different priorities. The method can also include assigning the packets to the queues, wherein at least some of the queues include packets of a plurality of different priorities. In addition, the method can include assigning a first time-to-live (TTL) value to a first packet in a first queue. The method can also include altering the first TTL value of the first packet in response to a second packet of a second queue being scheduled. Further, the method can include discarding the first packet in response to the first TTL value having a predetermined relationship with respect to a predetermined value.

A low latency switch architecture for high performance packet-switched networks which is a combination of input buffers capable of avoiding head-of-line blocking and an internal switch interconnect capable of allowing different input ports to access a single output simultaneously.

Systems, methods and software useful for overcoming network congestion problems including head-of-line blocking issues and other network congestion problems. In certain aspects, flow control mechanisms implemented in a switch device or other network device manage buffer and system level resources using a scheduler to control the amount of data requested from a local SAN fabric. Switches and other network devices configured according to the present invention monitor each individual SCSI task, and are configured to apply flow control measures to each active session when buffering resources become scarce, such as when buffering data for a slower-speed WAN link or TCP / IP based interconnects of any speed.

A method for preventing head of line blocking in an Ethernet system. In one embodiment, a network interface detects whether there is traffic flow congestion between the network interface and a data processing unit such as a CPU or other peripheral. If yes, the network interface communicates the congestion status to its attached Ethernet switch. In another aspect of the invention, the Ethernet switch then stops serving the congested port or queue, and informs a switch from which the traffic flow causing the congestion originates. In a further aspect, the originating switch then reduces bandwidth for the traffic flow causing the congestion. In a still further aspect, the originating switch can take the bandwidth that it acquired because of reducing the congesting traffic flow, and use it to increase bandwidth for other traffic flow.

To avoid packet out-of-sequence problems, while providing good load balancing, each input port of a switch monitors the outstanding number of packets for each flow group. If there is an outstanding packet in the switch fabric, the following packets of the same flow group should follow the same path. If there is no outstanding packet of the same flow group in the switch fabric, the (first, and therefore subsequent) packets of the flow can choose a less congested path to improve load balancing performance without causing an out-of-sequence problem. To avoid HOL blocking without requiring too many queues, an input module may include two stages of buffers. The first buffer stage may be a virtual output queue (VOQ) and second buffer stage may be a virtual path queue (VPQ). At the first stage, the packets may be stored at the VOQs, and the HOL packet of each VOQ may be sent to the VPQ. By allowing each VOQ to send at most one packet to VPQ, HOL blocking can be mitigated dramatically.

A method is disclosed for maintaining a table of recent accesses for each port for use in predicting whether a request for data from a source device is likely to be sent to a high speed or low speed destination device. The table of recent accesses lists every source device attached to that port and the speed of the destination device with the most recent access to each source device. When an OPN primitive is received at the source port, the source device is identified and used with the table of recent accesses to predict whether the destination device is likely to be high speed or low speed, and ultimately whether to send data from the source device or reject the request.

A method, computer program product, and a data processing system for posting and retrieving WQEs to a shared receive queue in a manner that alleviates head-of-line blocking issues is provided. The present invention provides a shared receive queue and a posting routine for allowing non-sequential work request postings in the shared receive queue. Additionally, a fetch routine is provided for non-sequential fetching of work queue elements from the shared receive queue.

A head of line (HOL) blocking count value, which is directly proportional to the committed traffic load for traffic flow through an output port within a multi-stage switch mesh, is computed for a path by adding the values associated with all output ports within the path. In selecting a route, all paths from the source to the destination are identified and sorted by head of line blocking count value. Rather than selecting a path based on traffic load, the path having the lowest head of line blocking count value and sufficient capacity for the requested traffic is selected, with selection between paths having equal head of line blocking count values being made based on traffic load.

Aspects of the present disclosure provide various apparatuses and methods of reordering DRB flow packets using an in-band solution to reduce delay due to data packet buffering in head-of-line blocking scenarios across multiple DRB flows. When a packet of a flow is lost or not received, other flows carried in the same DRB can forward later received packets without waiting for the missing packet to be retransmitted and received.

The method thereof comprises: counting the numbers of pause frames received by the port; when the numbers of pause frames received by the port in a preset closing time is more than maximum allowed numbers of pause frame, the state of port will be modified in order to make the forwarding device not to forward message to the port. The apparatus thereof comprises: a pause frame counting module for counting the pause frames and a port setting module for modifying the state of port. The pause frame counting module is connected to the port setting module.

An Ethernet switch for preventing head of line blocking by passing an Ethernet packet carrying congestion status information among Ethernet switches over the Ethernet. When a port of an Ethernet switch is congested, the switch informs other Ethernet switches by broadcasting to other switches over the Ethernet an Ethernet packet indicating that the port is congested. If another Ethernet switch finds that one of its traffic classes is destined to the congested port, this switch drops packets of that traffic flow until receiving an Ethernet packet indicating that the congestion is alleviated.

The invention relates to the technical field of electronic communication, and specifically to a method for preventing head-of-line blocking (HOL). The SDN-based method for preventing head-of-line blocking is applied to a SDN network comprising a plurality of switches and comprises the steps as follows: step 1, switches sense the state of an output port according to a set sensing method and report an HOL event to a SDN controller when the output port enters a head-of-line blocking state; step 2, the SDN controller calculates a forward path of the HOL event according to the HOL event so as to obtain an entrance switch and a corresponding input port; step 3, the SDN controller issues OF information to the input port; step 4, the input port drops a message causing the HOL event or limits the speed of the message causing the HOL event after receiving the OF information. By adopting an SDN framework, the method provided by the invention directly limits the message at the source so as to prevent the impact and influence to the whole SDN network to achieve the whole network defense for preventing head-of-line blocking.

A method, computer program product, and a data processing system for posting and retrieving WQEs to a shared receive queue in a manner that alleviates head-of-line blocking issues is provided. The present invention provides a shared receive queue and a posting routine for allowing non-sequential work request postings in the shared receive queue. Additionally, a fetch routine is provided for non-sequential fetching of work queue elements from the shared receive queue.