Arista Networks, Inc.

Sriram Sellappa

Imtiyaz Mohammad

Mayukh Saubhasik

Hugh Weber Holbrook

Vipul P Shah

Venkitraman Kasiviswanathan

Neal Thomas Nuckolls

Asang Kamalakar Dani

Rajnish Gupta

Dhanashree Nagre

Ramakrishna Shivaramaiah Paduvalli

Eswaran Baskaran

Vishal Bandekar

Suhas Raghunath Joshi

Ramakrishnan Ganapathy Iyer

Simon Liang

Song Yuan

Anil Joshi

Ashwini Kumar

Ramya Edara

Niranjan Nagaraju

Systems and methods for handling resource requests in programming of network device tables. A feature agent of a network device enters a resource request for a resource in a resource table and a resource server returns a response to the request, where the response includes a resource value and a dependency constraint encapsulating validity criteria for utilizing the resource value. When the resource value and dependency constraint are returned, the feature agent writes an entry to a software table, where the software table entry includes the resource and the corresponding dependency constraint. A hardware examines the software table entry and determines whether the dependency constraint has been satisfied. When the dependency constraint has been satisfied, the hardware agent writes an entry corresponding to the software table entry in a hardware table. The network device then processes packets according to the hardware table with the new entry.

Non-blocking distributed resource allocation and resolution across forwarding agents using dependency propagation

Systems and methods for maintaining order consistency for asynchronous writes in a network device. In one embodiment, a method includes issuing, by a feature agent of a network device, a plurality of writes to software tables which may be contained in the feature agent. For each write, a corresponding entry is added to a log associated with the software tables, the log entry identifying an order of the write among the plurality of writes. A hardware agent accesses the log to select a next one of the entries in the identified write order and then accesses the software table to retrieve content identified by the selected log entry. The hardware agent then translates the information from the software table to a different form if necessary and writes the information to one or more hardware tables that are indicated by the write associated with the selected log entry.

Write order consistency for asynchronous hardware writes among distributed forwarding and hardware agents with versioned change log

Prefix compression routes provided via exact match using redirection and mirroring Forwarding Equivalence Class entries in hardware. In a network device, a first table is stored having a first entry, a second table is stored having a second entry, and a third table is stored having a third entry including routing information for routing data packets. The first entry references a first memory location in the second table, the second memory location stores the second entry, and the second entry referencing a second memory location in the third table. A data packet is received, and the first entry is accessed based on a destination address of the data packet. Routing information is obtained as a result of accessing the first entry. The data packet is sent by the network device according to the routing information.

Prefix compression routes in exact match using redirection and mirroring FECs in hardware

Some embodiments provide techniques for optimized programming of forwarding data in network device hardware. An operating system executing on the network device receives information associated with a network topology of a network(s) to which the network device belongs. Based on this information, the operating system can generate various data structures that facilitate the routing and forwarding of data through the network device. Based on the generated data structures, the operating system may then program hardware resources in the network device in order to implement routing and forwarding operations stored in the data structures. During generation of the data structures, the operating system may perform some operations to optimize the programing of the hardware resources in a manner that reduces the amount of hardware resources that would otherwise be used without such optimizations.

Optimized programming of forwarding data in network device hardware

In some embodiments, a method sets a threshold for utilization of a first table, wherein the utilization is based on layer 3 addresses and layer 2 addresses being stored in the first table. When a utilization of the first table does not meet the threshold, the method stores a layer 3 address in the first table. The first table uses a first type of lookup to determine a next hop address for the layer 3 addresses or the layer 2 addresses, and the first table also stores one or more layer 2 addresses. When the utilization of the first table meets the threshold, the method stores the layer 3 address in a second table where the second table uses a second type of lookup to determine the next hop address for layer 3 addresses.

Routing table selection based on utilization

In general, embodiments described herein relate to methods and systems for reorganizing processing information hierarchies to remove duplicative and/or redundant portions of a processing information hierarchy such that they, for example, require fewer resources of the network devices on which they are stored.

System and method for processing information hierarchy management

A method for network traffic forwarding using multi-virtual routing and forwarding instances forwarding lookups is provided. The method includes: receiving network traffic on an interface of a network device associated with a first virtual routing and forwarding instance (VRF) fallback policy specifying a first set of rules; making a first determination that the network traffic includes network traffic information matching the first set of rules; initiating execution of VRF lookups in a plurality of VRFs programmed in the network device to obtain a set of VRF lookup results; making a second determination, using the set of VRF lookup results and the first VRF fallback policy, that the network traffic is to be forwarded using a first VRF among the plurality of VRFs; and initiating, based on the second determination, forwarding of the network traffic using the first VRF.

Network traffic engineering with multi-virtual routing and forwarding lookup

In one embodiment, a method is provided. The method includes receiving a data packet via an ingress interface of the network device. The method also includes determining whether the data packet comprises an Internet Protocol version 4 (IPV4) or an Internet Protocol version 6 (IPV6) packet. The method further includes in response to determining that the packet comprises an IPV4 packet, identifying a first entry in an adjacency table. The first entry is associated with an address prefix. The address prefix is associated with first Internet Protocol (IP) address of the data packet. The first entry indicates a next hop for the data packet. The adjacency table comprises a second entry associated with the address prefix. The method further includes forwarding the packet to the next hop indicated by the first entry in the adjacency table, via an egress interface of the network device.

Avoiding recirculation of data packets in a network device

A method and apparatus of a network element that processes changes to forwarding information is described. In an exemplary embodiment, the network element receives an indication of a change to a network topology of the network. The network element forwarding information includes a first plurality of prefixes and a plurality of adjacencies and each of the first plurality of prefixes references one of the plurality of adjacencies. In addition, in response to receiving the indication of the change to the network topology, the network element identifies a change to the plurality of adjacencies. Furthermore, the network element updates, in-place, the plurality of adjacencies in-place with the change to the plurality adjacencies.

System and method of processing in-place adjacency updates

A network device includes a first network processor that forwards packets based on a first forwarding information table; a second network processor that forwards packets based on a second forwarding information table; a first group of ports operably connected to the first network processor; and a second group of ports operably connected to the second network processor. The first forwarding information table specifies that packets, received by the first network processor, that specify a destination device reachable by the first group of ports and the second group of ports are forwarded by a port of the first group of ports. The second forwarding information table specifies that packets, received by the second network processor, that specify the destination device reachable by the first group of ports and the second group of ports are forwarded by a port of the second group of ports.

Load balancing on multi-chip network switch without full bi-section bandwidth

A method for managing traffic in a network. The method includes receiving an overlay frame comprising a header portion and obtaining, from the header portion, a first label and a second label. The method further includes generating an expanded label comprising the first label and the second label, making a first determination that an overlay network table comprises an entry for the expanded label, and based on the first determination, processing the overlay frame using the entry.

Growing community of inventors

Santa Clara, CA, United States of America

Sriram Sellappa