Arista Networks, Inc.

Qlogic Corporation

Min H Teng

Hugh Weber Holbrook

Govind Kizhepat

Anshul Sadana

Hitesh Ahuja

François Labonté

Muhammad Khalid Yousuf

Sambath Kumar Balasubramanian

Sonny N Tran

Kevin Martin Amiraux

Kenneth Y Y Choy

Fai Li

Stefan J Rebaud

Alexander S Rose

A system that provides communication services may include two switches for redundancy. The switches may indicate that they are independent devices to upstream devices and indicate that they are the same device to a predetermined device set. If one of the switches enters an undesired state, then the switch in a standby state may modify the upstream devices to preferentially forward packets directed toward the predetermined device set to the standby switch rather than an active switch, and after transitioning the active switch to the standby state, transition the standby switch to the active state.

Network element with dual independent sub network elements with seamless failover

A method and system of accelerating monitoring of network traffic. The method may include receiving, at a network chip of a network device, a network traffic data unit; capturing, by the network chip, the network traffic data unit based on a traffic sampling rate; adding, by the network chip, a sampling header to the network traffic data unit to obtain a sampled network traffic data unit; sending the sampled network traffic data unit from the network chip to a sampling engine; receiving, from the sampling engine, a flow datagram that includes a network traffic data unit portion and a flow datagram header; generating a flow network data traffic unit that includes the flow datagram; and transmitting the flow network data traffic unit towards a collector.

Accelerated network traffic sampling using an accelerated line card

A system and method for calculating latency including a latency calculation device configured to: receive an enqueue notification relating to a packet enqueue operation and including a queue identifier, increment an enqueue counter, and determine that a latency calculation flag is not set. Based on the determination that the latency calculation flag is not set, the latency calculation device is configured to: determine a first time corresponding to the enqueue notification, store the first time, store a latency start count, and set the latency calculation flag. The latency calculation device is also configured to: receive a dequeue notification relating to the packet dequeue operation and including the queue identifier, increment a dequeue counter, determine that the latency start count and the dequeue counter values match, determine a second time corresponding to the dequeue notification, and calculate latency as the difference between the first time and the second time.

Method and system for non-tagged based latency calculation

In some embodiments, a hardware linked-list manager includes a wildcard search controller for generating corresponding queue-specific read requests from wildcard read requests. The linked-list manager may be part of an on-chip interagent switch for allowing a plurality of agents to communicate with each other. The interagent switch may include a crossbar switch and a plurality of hardware linked-list managers integrated on the chip, connected to a random access memory, and connected to the crossbar switch such that the crossbar switch is capable of connecting each of the linked-list managers to each of the agents. Each linked-list manager sends agent-generated data to the memory for storage in the memory as a linked-list element, and retrieves linked-list elements from memory in response to agent read requests. A shared free-memory linked-list manager may maintain a linked list of free memory locations, and provide free memory address locations to a linked list manager upon request.

Switching systems and methods using wildcard searching

Growing community of inventors

Los Altos, CA, United States of America

Min H Teng