摘要:
An apparatus and method for encapsulating and forwarding packets on a network are disclosed. The network can include a first subnetwork such as a virtual private network connected to a larger public network such as the Internet. An encapsulating header is attached to a packet to be transferred across the public network from a source node on the private network to a destination node on the private network, such that the packet can be transferred across the public network. The encapsulating header includes a value which is derived from the private header on the packet used to transfer the packet along the private network. The value is therefore associated with a source/destination pair within the private network. The value can be derived by performing a hash operation on the private network header. After the public network header containing the value derived from the private network header is attached to the packet, it can be forwarded across the public network. A logical operation such as a hash operation can be performed on the public network header to select one of a plurality of possible paths on the public network to forward the packet. As a result, each source/destination pair within the private network will be associated with a path within the public network. Traffic from the private network can therefore be distributed over multiple paths in the public network, thus reducing or eliminating traffic congestion and overload.
摘要:
An apparatus and method for encapsulating and forwarding packets on a network are disclosed. The network can include a first subnetwork such as a virtual private network connected to a larger public network such as the Internet. An encapsulating header is attached to a packet to be transferred across the public network from a source node on the private network to a destination node on the private network, such that the packet can be transferred across the public network. The encapsulating header includes a value which is derived from the private header on the packet used to transfer the packet along the private network. The value is therefore associated with a source/destination pair within the private network. The value can be derived by performing a hash operation on the private network header. After the public network header containing the value derived from the private network header is attached to the packet, it can be forwarded across the public network. A logical operation such as a hash operation can be performed on the public network header to select one of a plurality of possible paths on the public network to forward the packet. As a result, each source/destination pair within the private network will be associated with a path within the public network. Traffic from the private network can therefore be distributed over multiple paths in the public network, thus reducing or eliminating traffic congestion and overload.
摘要:
An apparatus and method for forwarding data on a network are described. A label-switching subnetwork within the network includes an ingress node and an egress node coupled to source and destination nodes, respectively, on the network. The ingress node sends a signal along a route within the subnetwork through a plurality of subnetwork nodes to the egress node. In response, the subnetwork nodes transmit response signals back along the route toward the ingress node which define the route through the subnetwork and simultaneously allocate a plurality of paths within the route. A single path can be selected for forwarding of data packets associated with a source/destination pair, ensuring that data packets arriving at the destination are not misaligned.
摘要:
In general, techniques are described for selectively invoking graceful restart procedures when a route reflector member of a redundant route cluster fails. In one example, a method is provided that includes determining, by a provider edge router that supports graceful restart procedures, that a first router forms a redundant group with at least a second router. The method also includes detecting a failure of the first router and determining that at least the second router in the redundant group is operating approximately while the first router is failed. The method further includes overriding graceful restart procedures with respect to the failed first router when at least the second router is operating. The method also includes forwarding one or more data packets according to route information provided via the second router.
摘要:
Network address translation (NAT) translates between globally unique addresses used within a global network and a local network. A method, for example, includes mapping a first set of globally non-routable global addresses to a second set of globally routable global addresses, and forwarding packets in accordance with the mapping. The method may further include assigning the first set of addresses to devices of a local network, and forwarding packets between the devices of the local network and a global network. These techniques may significantly reduce the demand placed on routing devices in a global network.
摘要:
Techniques are described for distributing network traffic across parallel data paths. For example, a router may perform a hash on routing information of the packet to generate a hash value corresponding to the packet flow associated with the packet. The router may map the hash value of the packet to a forwarding element associated with a data path. The router may dynamically update the mapping of hash values to forwarding elements in accordance with traffic flow statistics. In this manner, the router may distribute the packet flows from data paths with high volumes of traffic to data paths with smaller volumes of traffic. The router may further prevent out of sequence delivery of packets by updating the mapping upon a gap in the packet flow exceeding a threshold gap. For example, the router may update the mapping when a packet for a packet flow associated with the particular hash value has not been received for at least a defined time interval.
摘要:
In general, techniques are described for selectively invoking graceful restart procedures when a route reflector member of a redundant route cluster fails. In one example, a method is provided that includes determining, by a provider edge router that supports graceful restart procedures, that a first router forms a redundant group with at least a second router. The method also includes detecting a failure of the first router and determining that at least the second router in the redundant group is operating approximately while the first router is failed. The method further includes overriding graceful restart procedures with respect to the failed first router when at least the second router is operating. The method also includes forwarding one or more data packets according to route information provided via the second router.
摘要:
A network device coordinates with other devices in a network to create a distributed filtering system. The device detects an attack in the network, such as a distributed denial of service attack, and forwards attack information to the other devices. The devices may categorize data into one or more groups and rate limit the amount of data being forwarded based on rate limits for the particular categories. The rate limits may also be updated based on the network conditions. The rate limits may further be used to guarantee bandwidth for certain categories of data.
摘要:
A network comprises a plurality of switching nodes interconnected by communication links for transferring digital packets. At least one switching node in the network pre-establishes a bypass virtual circuit through the network to bypass an element of the network, such as a switching node or a communication link, in the network. The bypass virtual circuit defines a path to another switching node in the network. The first switching node uses the bypass virtual circuit so constructed in forwarding of a packet in the event of a failure or other malfunction of the element if the first switching node would otherwise transfer the packet over that element.
摘要:
An apparatus and method for efficient hashing uses both an identifying portion of a data packet, e.g., source and destination ID, and an identifying value of the node, e.g., the IP address of the node, to generate a hash result. By inserting a unique value into the hash operation at each node, the invention effectively provides for a different hash implementation at each node. As a result, in situations where multiple paths or multiple links within a path are available to forward packets, traffic can be split over the multiple paths and links. Inefficient utilization of network links found in prior systems which use the same hash operation at each node are eliminated.