Arista Networks, Inc.

Eudean Michael Sun

Charles Melvin Aden

Dipankar Bhatt Acharya

Robert Calvin Cyphers

Eric Jahfei Won Yam

Ketan Rasikbhai Kotak

Ethan Barnett Rahn

Gil Adrian Torres

Sriram Chidambaram

Baptiste Elie Franck Covolato

Julien André Alexis Gomes

James Samuel Brinkley

For dynamic installation of extensible network operating system (ENOS) extensions, a network device downloads a full software image (SWI) having a signature and a directory containing software packages from an install source. During downloading, the network device modifies the full SWI on the fly, including reading and removing the software packages (SWIX files) from the directory, as each one is read, determining whether to keep it, for each 'keep' SWIX file, extracting an ENOS extension and a corresponding signature, and storing them on a nonvolatile memory, resulting in a modified SWI with a signature but without the SWIX files. After download is complete, the signed ENOS extensions stored on the nonvolatile memory are verified to determine valid ENOS extensions for inclusion in an overlay filesystem which becomes a root filesystem for the network device. The signed modified SWI is also verified before booting the network device into the modified SWI.

Signed SWI containing signed extensions for dynamic installation

A method, device, and computer readable medium for managing overheat behavior in a network device. The method includes determining that a first temperature exceeds a threshold by at least one temperature sensor disposed in a network device. The method includes logging a temperature sensor name of the at least one temperature sensor, the first temperature, and actions taken prior to determining that the first temperature exceeds the threshold. The method further includes disabling network device capabilities and enabling a subset of network device capabilities after powering cycling the network device.

Overheat restart behavior in network devices

Power management includes receiving power supply unit (PSU) power capacity data and PSU power consumption data from each PSU in a set of PSUs. A priority table is updated with controlled device power consumption data received from a set of controlled devices. If the PSU power consumption data of any PSU exceeds an associated threshold calculated based on the PSU power capacity data for the PSU, then one or more controlled devices are selected based on the controlled device power consumption data in the priority table. When power consumption data of any PSU exceeds the associated threshold, a power response procedure powers off at least one of the selected one or more controlled devices, reduces power supplied to at least one of the selected one or more controlled devices, or reduces an output voltage of any of the PSU for which the PSU power consumption data exceeds the associated threshold.

Centralized adaptive power management

Methods and systems for managing power for a Power over Ethernet (PoE) device are disclosed herein. The method may include obtaining, by a supervisor, power information from a plurality of power supply units (PSUs) to obtain total consumed power; obtaining, by the supervisor, a total system power capacity value associated with the plurality of PSUs; and calculating, by the supervisor, a total available PoE power value using the total consumed power and the total system power capacity value; and making a first determination, by the supervisor, using the total available PoE power value and a PoE power table, whether a powered device should stop receiving power.

Power over Ethernet (PoE) power management

Methods and systems for managing power for a Power over Ethernet (PoE) device are disclosed herein. The method may include: configuring a power sourcing equipment (PSE) controller component with an interface keep-on order that indicates a relative priority for providing power to interfaces of the PSE controller component; storing, by a supervisor, keep-on order values and interface identifiers in entries of a PoE power table; transmitting to the supervisor, by the PSE controller component, a first interface power draw value associated with a first interface and a second interface power draw value associated with a second interface; storing, by the supervisor, the interface power draw values entries associated with the respective interfaces; and performing, by the supervisor and in response to a power event, a power management action.

Power over ethernet (POE) power management

Methods and systems for managing power for a Power over Ethernet (PoE) device are disclosed herein. The method may include obtaining, by a supervisor, power information from a plurality of power supply units (PSUs) to obtain total power consumption information; obtaining, by the supervisor, a total system power capacity value for the plurality of PSUs; and making a determination, by the supervisor, using the total power consumption information, and the total system power capacity value, whether a powered device should stop receiving power.

Power over ethernet (PoE) power management

Methods, computer readable mediums, and systems for power management. The method of power management may include receiving first power supply unit (PSU) capacity data and first PSU consumption data from a first PSU to a supervisor; storing, via the supervisor, the first PSU capacity data and the first PSU consumption data in storage; receiving first controlled device power consumption data from a first controlled device to the supervisor; updating a priority table, stored in the storage, with the first controlled device power consumption data; making a first determination that the first PSU consumption data exceeds a first threshold, where, the first threshold is calculated based on at least the first PSU capacity data; and based on the first determination, initiating a power response procedure.

Methods, systems, and computer readable mediums for hitless repair. Hitless repair may include making a first determination, by a system control device (SCD) agent of a network device, that a SCD of the network device has experienced an error and/or is to be updated; making a second determination, by the SCD agent, that the SCD and the network device support the hitless repair; performing, by the SCD agent, a pre-hitless repair action set; and performing, by the SCD agent and after completing the pre-hitless repair action set, a post-hitless repair action set, including a hitless reset of the SCD.

Growing community of inventors

Walnut Creek, CA, United States of America

Eudean Michael Sun