# US 2012/0236723 A1 · Latency Analysis of Traffic Passing Through an Ethernet Switch

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2012/0236723 A1
- Document ID: US20120236723A1
- Status: Published application
- Kind: A1
- Application number: 13/420,527
- Filed: 2012-03-14
- Published: 2012-09-20
- Earliest claimed: 2011-03-14
- Assignees: Arista Networks, Inc.
- Inventors: Kenneth J. Duda; Anshul Sadana; Hugh W. Holbrook
- Web page: https://patentgrep.com/patent/US20120236723A1

## CPC

- H04L 43/0852: Delays
- H04L 43/106: using time related information in packets, e.g. by adding timestamps
- H04L 47/11: Identifying congestion
- H04L 49/555: {Error detection}
- H04L 47/28: in relation to timing considerations
- H04L 47/30: in combination with information about buffer occupancy at either end or at transit nodes
- H04L 49/351: for local area network [LAN], e.g. Ethernet switches
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

A method for accurately measuring, recording and reporting latency of an Ethernet switch is disclosed. Various packet queues in the switch are monitored at any given time and forwarding latency is calculated. Latency data from multiple switching elements in a network is collected to provide end-to-end forwarding latency of a system.

## Claims

### Claim 1

A method implemented by a network switch comprising a control processor, a queue monitor, network interface ports, corresponding packet queues, and a network management port, the method comprising:
reading depths of the packet queues;
recording the depths together with corresponding timestamps; and
transmitting the depths and corresponding timestamps from the network management port in response to a request received from a network management system.

### Claim 2

A method implemented by a network switch management system connected to a network comprising multiple network switches and multiple servers, the method comprising:
transmitting to the network switches requests for depths of packet queues together with corresponding timestamps;
receiving from the network switches the requested depths of the packet queues together with the corresponding timestamps;
calculating latencies of interface ports of the network switches from the received depths of the packet queues;
calculating an end-to-end latency of the network between two servers by adding the calculated latencies of the interface ports.
