Methods related to communication between and within nodes in a high performance computing system are presented. Processing time for message exchange between a processing unit and a network controller interface in a node is reduced. Resources required to manage application state in the network interface controller are minimized. In the network interface controller, multiple contexts are multiplexed into one physical Direct Memory Access engine. Virtual to physical address translation in the network interface controller is accelerated by using a plurality of independent caches, with each level of the page table hierarchy cached in an independent cache. A memory management scheme for data structures distributed between the processing unit and the network controller interface is provided. The state required to implement end-to-end reliability is reduced by limiting the transmit sequence number space to the currently in-flight messages.

Methods and systems for providing IP services in an integrated fashion are provided. According to one embodiment, a flow cache is established having multiple entries each identifying one of multiple VR flows through a VR-based network device and corresponding forwarding state information. A packet is received at an input port of a line interface module of the network device and forwarded to a VRE. Flow-based packet classification is performed by the VRE. An attempt is made to retrieve an entry of the flow cache based on a result of the flow-based packet classification. On a flow cache hit, one or more appropriate packet transformations are identified for application to the packet and it is determined whether to process the packet with a VSE based on the corresponding forwarding state information. On a flow cache miss, the new VR flow is added to the flow cache by performing flow learning.

Example implementations relate to router advertisement caching. A controller may comprise a processing resource and a memory resource storing machine-readable instructions to cause the processing resource to perform a number of actions. For instance, the controller may cache a router advertisement (RA) from a router, select, in response to a client device associating with the controller, the cached RA from the router, and unicast the RA from the router to the client device.

In a method for forwarding a packet, a physical forwarding unit receives a data packet, may find a first subnet route based on a destination IP address of the data packet, and forwards the data packet to a virtual forwarding unit via an output interface of the first subnet route. The virtual forwarding unit receives the data packet may find a first host route based on the destination IP address of the data packet, and forwards the data packet based on the first host route.

A distributed multilayered network routing architecture comprises multiple layers including a controller layer comprising a controller, a control plane layer comprising one or more control plane subsystems, and a data plane layer comprising one or more data plane subsystems. A controller may be coupled to one or more control plane subsystems. A control plane subsystem may in turn be coupled to one or more data plane subsystems, which may include one or more software data plane subsystems and/or hardware data plane subsystems. In certain embodiments, the locations of the various subsystems of a distributed router can be distributed among various devices in the network.

The embodiments described herein provide a data transmission system comprising a plurality of video routers, a supervisory system for transmitting one or more router configuration signals to one or more video routers, and a control communication network for coupling the plurality of video routers and the supervisory system. Each router in the system comprises a backplane including a plurality of backplane connections, at least one line card and at least one fabric card. Each line card comprises a plurality of input ports and output ports where each input and output port is coupled to a respective external signal through the backplane. Each line card further comprises a line card cross-point switch having a plurality of input switch terminals and a plurality of output switch terminals. Each fabric card comprises a fabric card cross-point switch having a plurality of input switch terminal and a plurality of output switch terminals. Furthermore, each line card and each fabric card comprises a card controller where the card controller selectively couples one or more input switch terminals of a cross-point switch to the output switch terminals of that cross-point switch. The cross-point switches being manipulated by the card controller may belong to one or more different cards within the same video router.

In general, techniques are described for identifying a result set of multiple paths through a network for one or more label switched paths between a source node and a destination node. In some examples, the identified paths are computed to be broadly separate to avoid overlapping network elements. A device that routes the label switched paths to the network may select from the result set of multiple paths to route each of the label switched paths. In response to detecting a failure of a network element along the routed path for a label switched path, the device may select a new path from the result set that includes already-identified paths.

A RF router for routing n input signals to m destinations, where the router comprises a backplane coupled to a plurality of RF input terminals, a plurality of RF output terminals, a plurality of splitters and a plurality of connectors. The backplane is also coupled to a controller and a plurality of connectors for receiving a plurality of switching matrices. The RF router comprises a plurality of uv input switch matrices, a plurality of pq intermediate switch matrices and a plurality of rs output switch matrices, where at least one of the plurality of uv input switch matrices, the plurality of pq intermediate switch matrices and the plurality of rs output switch matrices are redundant.

A data center security system and method are provided that leverage server systems on a chip (SOCs) and/or server fabrics. In more detail, server interconnect fabrics may be leveraged and extended to dramatically improve security within a data center.

A system provisions global logical entities that facilitate the operation of logical networks that span two or more datacenters. These global logical entities include global logical switches that provide L2 switching as well as global routers that provide L3 routing among network nodes in multiple datacenters. The global logical entities operate along side local logical entities that are for operating logical networks that are local within a datacenter.