Some embodiments provide a novel method for sharing data between user-space processes and kernel-space processes without copying the data. The method dedicates, by a driver of a network interface controller (NIC), a memory address space for a user-space process. The method allocates a virtual region of the memory address space for zero-copy operations. The method maps the virtual region to a memory address space of the kernel. The method allows access to the virtual region by both the user-space process and a kernel-space process.

A method and apparatus of a network element that processes a packet in the network element is described. In an exemplary embodiment, the network element receives a data packet that includes a destination address. The network element receives a packet, with a packet switch unit, wherein the packet was received by the network element on an ingress interface. The network element further determines if the packet is to be stored in an external queue. In addition, the network element identifies the external queue for the packet based on one or more characteristics of the packet. The network element additionally forwards the packet to a packet storage unit, wherein the packet storage unit includes storage for the external queue. Furthermore, the network element receives the packet from the packet storage unit and forwards the packet to an egress interface corresponding to the external queue.

A memory management method includes: determining that available storage space of a first memory in a network device is less than a first threshold, where the first threshold is greater than 0 and the first memory stores a first packet queue; and deleting at least one packet at the tail of the first packet queue from the first memory based on the available storage space of the first memory being less than the first threshold. When the available storage space of the first memory is less than the first threshold, a packet queue, namely, the first packet queue, is selected and a packet at the tail of the packet queue is deleted from the first memory.

Provided are methods and apparatuses for packet scheduling for software-defined networking in an edge computing environment. A packet scheduling method according to an exemplary embodiment of the present disclosure comprises: receiving packets arriving at a queue connected to a switch in a software-defined network in an edge computing environment; moving the packets in the queue forward one position based on the order of arrival each time a packet is served by the switch; and if a new packet enters the switch while the buffer in the queue is full, pushing out the packet at the front and putting the new packet at the end of the queue.

A priority queue sorting system including a priority queue and a message storage. The priority queue includes multiple priority blocks that are cascaded in order from a lowest priority block to a highest priority block. Each priority block includes a register block storing an address and an identifier, compare circuitry that compares a new identifier with the stored identifier for determining relative priority, and select circuitry that determines whether to keep or shift and replace the stored address and identifier within the priority queue based on the relative priority. The message storage stores message payloads, each pointed to by a corresponding stored address of a corresponding priority block. Each priority block contains its own compare and select circuitry and determines a keep, shift, or store operation. Thus, sorting is independent of the length of the priority queue thereby achieving deterministic sorting latency that is independent of the queue length.

When a measure of buffer space queued for garbage collection in a network device grows beyond a certain threshold, one or more actions are taken to decreasing an enqueue rate of certain classes of traffic, such as of multicast traffic, whose reception may have caused and/or be likely to exacerbate garbage-collection-related performance issues. When the amount of buffer space queued for garbage collection shrinks to an acceptable level, these one or more actions may be reversed. In an embodiment, to more optimally handle multi-destination traffic, queue admission control logic for high-priority multi-destination data units, such as mirrored traffic, may be performed for each destination of the data units prior to linking the data units to a replication queue. If a high-priority multi-destination data unit is admitted to any queue, the high-priority multi-destination data unit can no longer be dropped, and is linked to a replication queue for replication.

Technologies for packet forwarding under ingress queue overflow conditions includes a computing device configured to receive a network packet from another computing device, determine whether a global packet buffer of the NIC is full, and determine, in response to a determination that the global packet buffer is full, whether to forward all the global packet buffer entries. The computing device is additionally configured to compare, in response to a determination not to forward all the global packet buffer entries, a selection filter to one or more characteristics of the received network packet and forward, in response to a determination that the selection filter matches the one or more characteristics of the received network packet, the received network packet to a predefined output. Other embodiments are described herein.

A method and apparatus of a network element that processes a packet in the network element is described. In an exemplary embodiment, the network element receives a data packet that includes a destination address. The network element receives a packet, with a packet switch unit, wherein the packet was received by the network element on an ingress interface. The network element further determines if the packet is to be stored in an external queue. In addition, the network element identifies the external queue for the packet based on one or more characteristics of the packet. The network element additionally forwards the packet to a packet storage unit, wherein the packet storage unit includes storage for the external queue. Furthermore, the network element receives the packet from the packet storage unit and forwards the packet to an egress interface corresponding to the external queue.

A method of managing a queue and a communication node that may maintain state information for each flow of a corresponding node, may estimate a time of arrival of each packet of each flow based on flow information that is received from other communication nodes within a collision range and that includes the number of flows and the state information, and may determine dropping and queue scheduling associated with the packets based on the estimated time of arrival (ETA).

The disclosed technology concerns methods, apparatus, and systems for designing and generating networks-on-chip (“NoCs”), as well as to hardware architectures for implementing such NoCs. The disclosed NoCs can be used, for instance, to interconnect cores of a chip multiprocessor (aka a “multi-core processor”). In one example implementation, a wire-based routerless NoC design is disclosed that uses deterministically specified wire loops to connect the cores of the chip multiprocessor. The disclosed technology also comprises network interface architectures for use in an NoC. For example, a core can be equipped with a low-area-cost interface that is deadlock-free, uses buffering sharing, and provides low latency.