Juniper Networks, Inc.

Scott Mackie

Dennis C Ferguson

Nitin Kumar

Ross W Callon

David M Katz

Swamy Sadashivaiah Renu Kananda

Surya Chandra Sekhar Nimmagadda

Sandeep Bajaj

James Washburn

Rajiv N Patel

Kaushik Ghosh

Andrzej Szyszko

Timothy L Patrick

Manoj Leelanivas

Nayan S Patel

Hsien-Chung Woo

Yanfang Zhang

Jeffrey Y Lo

Ashok Kumar Padhy

Sreenivas Gadela

A network device may receive an original configuration that includes configuration objects, and may generate, based on the original configuration, a dependency graph that includes nodes representing and entries representing the configuration objects. The network device may receive a configuration update that includes new configuration objects, and may update the dependency graph based on the configuration update and to generate an updated dependency graph that includes new nodes and/or new entries representing the new configuration objects. The network device may test the configuration update, based on the updated dependency graph, to determine whether the configuration update fails or succeeds. The network device may selectively implement the configuration update based on the configuration update succeeding or perform a rollback of the configuration update, based on the configuration update failing, to restore the original configuration.

Rollback-on-error support for forwarding components of a network device

In general, techniques are described for organizing execution of distributed operating systems for network devices. A device comprising hardware computing nodes may be configured to perform the techniques. The hardware computing nodes may execute a protocol by which to discover a topology of the plurality of hardware computing nodes, and determine, based on the topology, a subset of the plurality of hardware computing nodes to manage execution of a distributed operating system. The determined subset of the plurality of hardware computing nodes may execute a communication bus by which to synchronize operating system state information between the subset of the plurality of hardware computing nodes. The hardware computing nodes may further execute, based on the operating system state information, the distributed operating system to provide an execution environment in which one or more applications execute.

Organizing execution of distributed operating systems for network devices

In general, techniques are described for maintaining coherency in distributed operating systems for network devices. A network device comprising hardware computing nodes may be configured to perform the techniques. The hardware computing nodes may execute a distributed operating system. At least one the hardware computing nodes may determine whether one or more of the plurality of hardware computing nodes has failed and is no longer supporting execution of the distributed operating system, and determine whether remaining ones of the plurality of hardware computing nodes exceeds a quorum threshold. The at least one of the hardware computing nodes may further restart, when the remaining ones of the plurality of hardware computing nodes is less than the quorum threshold, the distributed operating system.

Maintaining coherency in distributed operating systems for network devices

In general, techniques are described for communicating state information in distribute operating system. A network device comprises a first hardware node and a second hardware node. The first hardware node may execute a first instance of a distributed operating system, and maintain a first data structure that stores a plurality of objects defining a portion of state information. The second hardware node may execute a second instance of the distributed operating system, and maintain a second data structure that stores synchronized versions of the plurality of objects. The first hardware node may further receive updated state information, update the first data structure to include the updated state information, and synchronize the updated first data structure with the second data structure. The second hardware node may synchronize the second data structure with the updated first data structure.

Communicating state information in distributed operating systems

In one example, a method comprises generating, by a forwarding manager for an internal forwarding path executed by a plurality of packet processors of a forwarding unit of a network device, a dependencies structure that specifies one or more dependencies for a plurality of nodes, wherein the plurality of nodes represent different types of forwarding path elements of the forwarding path, wherein the plurality of nodes is binded to a first set of one or more packet processors of the plurality of packet processors; and rebinding, by the forwarding manager, a second set of one or more packet processors of the plurality of packet processors to the plurality of nodes, wherein rebinding the second set of one or more packet processors to the plurality of nodes is performed in a reverse direction of the dependencies structure.

Graph-based rebinding of packet processors

In one example, a method comprises receiving, by a forwarding manager for an internal forwarding path executed by at least one packet processor of a forwarding unit of a network device, one or more packet processing operations from a control unit of the network device; generating, by the forwarding manager based on the one or more packet processing operations, a plurality of nodes each comprising a unique token, wherein a first node of the plurality of nodes includes a token reference set to a value for the token of a second node of the plurality of nodes; configuring, by the forwarding manager based on the nodes, the forwarding path to include respective forwarding path elements for the plurality of nodes; and processing, by the packet processor, a packet received by the forwarding unit by executing the forwarding path elements.

Node representations of packet forwarding path elements

In one example, a method includes detecting, by a forwarding manager for an internal forwarding path executed by at least one packet processor of a forwarding unit of a network device, that the at least one packet processor of a plurality of packet processors has become available for processing packets for forwarding; in response to the detecting, ceasing, by the forwarding manager, execution of control processing operations received from a control unit of the network device; and programming, by the forwarding manager, a plurality of forwarding path elements of the at least one packet processor based on a dependencies data structure, wherein the dependencies data structure comprises one or more dependencies that each indicates, for a node of a plurality of nodes, one or more nodes that reference the node, and wherein each node of the plurality of nodes corresponds to a single forwarding path element of the plurality of forwarding path elements.

Individual network device forwarding plane reset

In some examples, a network device comprises a first application and a second application; a forwarding unit comprising an interface card to receive a packet; a packet processor; an internal forwarding path of the forwarding unit; a forwarding unit processor; a first interface; and a second interface. The first application is configured to configure, via the first interface, the internal forwarding path to include a sandbox that comprises a container for instructions to be configured inline within the internal forwarding path. The second application is configured to configure, via the second interface, the sandbox with second instructions that determine processing of packets within the sandbox. The packet processor is configured to process, in response to determining a packet received by the forwarding unit is associated with a packet flow controlled at least in part by the second application, the packet by executing the second instructions configured for the sandbox.

Network device data plane sandboxes for third-party controlled packet forwarding paths

In general, this disclosure describes a high-level forwarding path description language (FPDL) for describing internal forwarding paths within a network device. The FPDL enables developers to create a template that describes a section of an internal forwarding path within the forwarding plane of a network device. The FPDL provides syntactical elements for specifying the allocation of forwarding path structures as well as enabling the run-time construction of internal forwarding paths to interconnect the forwarding path structures in a manner specific to packet, packet flow, and/or interface properties, for example. In conjunction with late binding techniques, whereby the control plane of the network device provides arguments to template parameters that drive allocation by the packet forwarding engines of forwarding path structures specified by the FPDL, the techniques provide control plane processes a unified interface with which to manage the operation of the packet forwarding engines.

Growing community of inventors

Santa Cruz, CA, United States of America

Scott Mackie