Arista Networks, Inc.

Padmanabh Ratnakar

Vijay Rangarajan

Sahil Midha

Punit Kumar Jaiswal

Gagandeep Singh Arneja

Rajshekhar Biradar

Petr Budnik

Naveen Kumar Devaraj

T B Binoshmon

Abdul Haseeb Jehangir

Ashutosh Shukla

Piyush Lodhi

Rohit Shukla

Systems and methods in which a network switch detects a dropped packet and automatically generates a report with information on the dropped packet. Additionally, a visibility packet that includes the dropped packet and one or more additional headers is generated. The headers are used to route the visibility packet to the pipeline of the network switch that originally processed the dropped packet. The headers also include a visibility indicator that causes the pipeline to store visibility data. As the visibility packet is processed by the pipeline, information (visibility data) generated during the processing of the packet is stored in a visibility memory. The visibility data is retrieved from the visibility memory and added to the generated report prior to transmitting the report to an external collector for analysis of the packet drop.

Utilization of visibility feature to obtain and incorporate drop reasons into mirror on drop reports

Systems, methods, and computer-readable media are disclosed for identifying a source of congestion in a network using postcard-based telemetry. Sampled packets from a packet flow are received by a network management collector The packet flow emanates from a host into a network ingress point of an ingress networking device. The packet flow flows through a network made up of intermediate networking devices and terminates at a network egress point. At least a portion of the one or more sampled packets is analyzed to determine whether a congestion indication has been set by a networking device. In response to determining that the congestion indication has been set a congestion notification is provided to a network administrator. The congestion notification provides an indication that the networking device has detected network congestion.

Telemetry-based congestion source detection

In general, one aspect, the disclosure relates to a method for sampling packets in a network. The method includes receiving, by a first network device, a packet, making a first determination, by the first network device, that the packet is to be sampled, in response to the first determination: sampling the packet to obtain sampling data, storing sampling metadata associated with the packet, encapsulating, after the sampling, the packet to obtain an encapsulated packet, where the encapsulated packet comprises a bit that is set in an encapsulation header, wherein the bit is set based on the presence of the sampling metadata, and transmitting the encapsulated packet to a second network device.

Using fields in an encapsulation header to track a sampled packet as it traverses a network

Systems, methods, and computer-readable media are disclosed for characterizing data flows within an in-band telemetry (INT) domain. First aggregated statistical data regarding a flow of INT packets is periodically received from a first aggregating network device. The flow of INT packets flows from an initiator network device via one or more transit network devices to a terminator network device. Second aggregated statistical data regarding the flow of INT packets is periodically received from a second aggregating network device. In response to comparing the first aggregated statistical data and the second aggregated statistical data, one or more dropped INT packets may be implicitly identified. Based on the one or more dropped INT packets in the flow of INT packets, a dropped INT packet notification may be provided.

Detecting in-transit inband telemetry packet drops

This disclosure presents several embodiments of a device configured to perform inband telemetry tracking. The device is configured to store a flow table that tracks metrics of a detected packet flow that passes through the device. More specifically, the flow table tracks several telemetry metrics, each telemetry metric being specific to a tracking time period, and tracking statics based only on packets of the packet flow that were received during that specific time period. At an end of an export time period (which is longer than a tracking time period), the device transmits all stored telemetry metrics from the flow table to a collector.

Interval flow-based inband telemetry

A facility to determine if performance issues between two host computers in a data network includes a central controller identifying endpoints of a flow path between the two hosts. The central controller communicates with endpoint network devices to initiate telemetry tagging traffic on the flow path. A collector receives telemetry communicated in a packet from network devices on the flow path. A network operator can view the collected telemetry to assess whether the performance issue is in the network or not.

Distinguishing network vs server issues when diagnosing application performance problems

Methods and systems are described for inband telemetry. The system receives a plurality of packets, each packet comprising a portion with INT data. The system identifies a packet flow from a source device to a destination device in the plurality of packets and calculates a telemetry metric based on INT data of a plurality of packets of the packet flow. Then the system forwards the calculated telemetry metric to an INT collector.

Growing community of inventors

Karnataka, India

Padmanabh Ratnakar