摘要:
In one embodiment, an edge device of a core network may receive a plurality of packets from a peripheral network having a plurality of active connections to the core network, where each packet has a destination address and a source address. The edge device may compute a hash on the destination address or the source address of each packet, and determine whether the computed hash corresponds to the edge device. In response to the computed hash not corresponding to the edge device, the edge device may drop the packet, and in response to the computed hash corresponding to the edge device, the edge device may process the packet to forward the packet, where the dropping and processing load balances the plurality of packets over the active connections and prevents formation of loops in the core network.
摘要:
Diversity constraints with respect to connections or links in a client layer are conveyed to a server layer where those links or connections are served by paths in the server layer. A network device in the server layer stores data associated paths in the server layer with identifiers for connections in the client layer. The network device in the server layer receives from a network device in the client layer a request to set up a path in the server layer for a connection in the client layer. The network device in the server layer receives information describing the diversity requirements associated with connections in the client layer. The server layer network device computes a route in the server layer for the connection specified in the request based on the diversity requirements.
摘要:
In one embodiment, a connectivity verification protocol (CVP) session for a particular virtual interface (VI) may operate on a particular group of two or more line cards (LCs) on a network device. The group of LCs may then transmit CVP session packets, at a reduced rate that is sufficient to maintain the CVP session based on a negotiated CVP full rate, onto the particular VI through ingress path processing on the network device. Ingress path processing, in particular, takes transmitted CVP session packets and egresses them onto an appropriate LC of the network device currently responsible for the VI egress. Also, in response to receiving CVP session packets for the VI on an LC of the network device currently responsible for the VI ingress, the receiving LC may forward the received CVP session packets to the particular corresponding group of LCs, which may then process the received CVP session packets.
摘要:
In one embodiment, a method includes receiving at a router, a multicast label distribution protocol message comprising local node information for a protected node and one or more leaf nodes downstream of the protected node in a primary label switched path, creating one or more backup label switched paths to the one or more leaf nodes, detecting a failure at the protected node, and forwarding at the router, traffic for the one or more leaf nodes to the one or more backup label switched paths. An apparatus for multicast label distribution protocol node protection is also disclosed.
摘要:
Disclosed are, inter alia, methods, apparatus, computer-storage media, mechanisms, and means associated with the coordinated updating of forwarding information bases (FIBs) in a multistage packet switching device, which performs at least lookup operations on multiple different FIBs in determining how to forward a packet. One embodiment uses lookup operations on two different FIBs, with these being an ingress FIB on an ingress line card and an egress FIB on an egress line card. In response to a change in the forwarding information for a stream of packets, the egress FIBs are first updated to include both the old and new forwarding information. After all egress FIBs have been updated, the ingress FIBs are updated to use the new forwarding information. This update procedure is designed to eliminate loss or duplication of packets induced during the updating of these FIBs to use the new forwarding information.
摘要:
A method of identifying an egress point to a network location in a data communications network comprising a plurality of nodes and links there between comprises the step, performed at the identifying node, of receiving a notification through the network advertising an adjacency to a network location. The method further comprises the steps, performed at the identifying node, of deriving from the notification adjacency information and identifying, from the adjacency information, the egress point.
摘要:
A method is provided for generating a backup route. Here, a route and a route distinguisher type associated with the route are received and a backup route is generated based on attributes of the route. A particular backup route distinguisher type that is associated with the route distinguisher type is assigned to the backup route. The backup route with the backup route distinguisher type are then advertised. Another method is provided that identifies the backup route. When the route and its route distinguisher type are received from the advertisement, an identification is made as to whether the route distinguisher type is assigned to a backup route. The route may then be designated as a backup route based on the identification.
摘要:
A system and method for measuring convergence performance in a network are disclosed. In one embodiment, the method includes receiving a plurality of probes at a network device, temporarily storing data from at least a portion of the received probes and deleting at least a portion of the temporarily stored data at regular intervals, and receiving a packet indicating a convergence event and storing data from probes received over a predetermined period.
摘要:
A method is disclosed for computing routing information for a data communications network comprising a plurality of nodes. The method comprises the step performed at a computing node of receiving from a remote node in the network a routing information update representing a change of state of the network. The method further comprises processing the update to identify a change of state type and setting, dependent on the change of state type, a delay interval after which routing information is to be computed from the update.
摘要:
A method is disclosed for constructing a backup route from a source node around an adjacent component. The source node derives a first set of nodes reachable from it without traversing the adjacent component and a second set of nodes from which a neighbor node of adjacent components is reachable without traversing the adjacent component. The source node then constructs a backup route via an intermediate node in the intersection of the first and second sets.