Google Inc.

Enfabrica Corporation

Carlo Contavalli

Shrijeet Mukherjee

Shimon Muller

Rochan Sankar

Ariel Hendel

Gurjeet Singh

Shashidhar R Gandham

Nandita Dukkipati

Harold Jason Schiller

Nils Krahnstoever

Thomas DeWeese

Vytautas Valancius

Michael A Glover

Daniel Eugene Eisenbud

Ahmed Said Mohamed Tawfik Issa

Michael Todd Joseph

An adaptive generic receive offload (A-GRO) system and method are disclosed. In some embodiments, the system comprises a host including a host protocol stack and a host memory, and a network interface card that is communicatively connectable to the host. The A-GRO system is configured to: receive a packet from a network, parse the packet to a header and a payload, classify and map the packet into a particular flow based on contexts associated with a plurality of flows and the header, and move the header and the payload to separate queues associated with the particular flow in the host memory, without holding and stalling the packet in hardware of the NIC. By maintain packet coherence information including header chains, the A-GRO allows the host to skip processing the packets between the first and last headers in a GRO aggregation. The A-GRO system also improves mis-ordering packet handling.

System and method for adaptive generic receive offload

A system for congestion control using a flow level transmit mechanism is disclosed. In some embodiments, the system comprises a source SFA and a receive SFA. The source SFA is configured to detect and classify a congestion notification packet (CNP) generated based on congestion in a network; select a receive block from a plurality of receive blocks based on the CNP; forward the CNP to a dedicated congestion notification queue of the receive block; identify a transmit queue from a plurality of transmit blocks based on processing the congestion notification queue, wherein the transmit queue originated a particular transmit flow causing the congestion; and stop the transmit queue.

System and method for congestion control using a flow level transmit mechanism

A server fabric adapter (SFA) communication system is disclosed. In some embodiments, the SFA communication system comprises an SFA communicatively coupled to a plurality of controlling hosts, a plurality of endpoints, and a plurality of network ports. The SFA is configured to receive a network packet from a network port of the plurality of network ports; separate the network packet into different portions, each portion including a header or a payload; map each portion of the network packet to: (i) a controlling host of the plurality controlling hosts, the controlling host being designated as a destination controlling host, or (ii) an endpoint of the plurality of endpoints, the endpoint being designated as a destination endpoint; and forward a respective portion of the network packet to the destination controlling host or the destination endpoint.

Server fabric adapter for I/O scaling of heterogeneous and accelerated compute systems

Systems and methods of performing rate limiting traffic shaping with a time-indexed data structure in a network device are provided. A network interface driver of the network device can received packets at the packet layer of a network host from a plurality of applications. The network interface driver can prevent one of the applications from sending additional packets for transmission until the application received a transmission completion notification indicating a packet previously forwarded to the packet layer of the network host has been transmitted. The network interface driver can process the received packets to determine a transmission time for each packet based on at least on rate limit policy. The network interface driver can store an identifier associated with the respective packet in a time-indexed data structure at a position associated with the transmission time determined for the packet. The network interface driver can determine that a time indexed in the time-indexed data structure has been reached and in response transmit a packet associated with the identifier stored in the time-indexed data structure at a position associated with the reached time. The network interface driver can communicate a transmission completion notification subsequent to the network interface driver transmitting the packet.

Device and method for scalable traffic shaping with a time-indexed data structure

Systems and methods of performing traffic shaping in a network device are provided. A network interface driver of the network device can store descriptors associated with packets received from multiple streams in a transmission queue in a first order. The network interface driver can transfer the descriptors to a traffic shaping module. In response to determining that a packet from a first stream, among the received packets, has been successfully transmitted by a network card, the network interface driver can communicate a packet transmission completion message corresponding to the packet to a software application that has awaited receipt of a packet transmission completion message before forwarding additional data packets from the first stream to the network interface driver. The network interface driver can communicate packet transmission completion messages corresponding to the packets received from the multiple streams to the software application in a second order, different from the first order.

Systems and method for single queue multi-stream traffic shaping with delayed completions to avoid head of line blocking

Network load balancing may be performed in a cluster by deploying multicasting or replication. In particular, one or more multicast trees may be deployed within the cluster. Packets may be multicast towards the hosts over the multicast tree(s) to reach the hosts in a pool of hosts associated with a given Virtual Internet Protocol ('VIP') address. Packets may be filtered before the packets reach the hosts to ensure that each host only receives packets for the session the host is handling. The filtering may be performed at various levels of the multicast tree. Replication may be deployed instead of multicasting to realize the same network load balancing.

Network load balancing using multicast or packet replication

A forwarding device, such as a router, in a network may communicate with a service device according to a networking protocol. According to exemplary embodiments, logic for communicating according to the networking protocol may be relocated from the forwarding device to one or more remote controllers. The remote controllers may encapsulate networking messages and forward the networking messages to the forwarding device using a tunnel, thereby allowing the messages to follow the same path through the network as the messages would have followed if the messages had originated at the forwarding device. Accordingly, the forwarding device can be made simpler, updates to the networking protocol or remote controllers may be made without ceasing operations at the forwarding device, errors in the network may be quickly diagnosed, and a backup remote controller may take responsibility for communications in the network if a primary controller is rendered inoperable.

GRE tunnels to resiliently move complex control logic off of hardware devices

A system and method for processing multicast streams is disclosed. A router receives the multicast streams and communicates the streams via communication tunnels to a set of ingestion pipelines. Each communication tunnel can be terminated upon the ingestion pipeline receiving a multicast stream. The set of ingestion pipelines process and deliver the multicast streams across existing IP infrastructure for real time distribution.

Growing community of inventors

Mountain View, CA, United States of America

Carlo Contavalli