摘要:
In one embodiment, an apparatus includes a processor for mapping packets associated with network flows to policy profiles independent of congestion level at the apparatus, and enforcing the policy profiles for the packets based on a congestion state. Packets associated with the same network flow are mapped to the same policy profile and at least some of the network flows are protected during network congestion. The apparatus further includes memory for storing the policy profiles. A method for protecting network flows during network congestion is also disclosed.
摘要:
In one embodiment, an apparatus includes a processor for mapping packets associated with network flows to policy profiles independent of congestion level at the apparatus, and enforcing the policy profiles for the packets based on a congestion state. Packets associated with the same network flow are mapped to the same policy profile and at least some of the network flows are protected during network congestion. The apparatus further includes memory for storing the policy profiles. A method for protecting network flows during network congestion is also disclosed.
摘要:
In one embodiment, an intermediate node computes paths for a set of tunnels that do not include a particular link (e.g., and possibly a scaled-down bandwidth for each tunnel), considering all of the tunnels of the set. The intermediate node informs head-end nodes of the tunnels of the computed paths (e.g., and scaled bandwidth) and/or a time to reroute the tunnels.
摘要:
In one embodiment, head-end nodes receive a list of tunnels to be rerouted from a particular link of an intermediate node. If a head-end node is unable to reroute a tunnel for which it is the head-end node using conventional distributed routing, each head-end node executes the same algorithm to compute paths for all tunnels in the list (e.g., potentially applying bandwidth scaling).
摘要:
In one embodiment, a node determines an overload ratio for an output as a ratio of a total rate of received traffic at the output to a preemption threshold of the output. The node also determines a ratio of traffic that is to be marked at the output based on the overload ratio and a ratio of previously marked traffic destined for the output from each input to the total traffic from each input to the output, and whether, for a particular input, the ratio of previously marked traffic is less than the ratio of traffic that is to be marked at the output. If so, the node marks unmarked traffic of the particular input corresponding to a difference between the ratio of traffic that is to be marked at the output and the ratio of previously marked traffic destined for the output from the particular input.
摘要:
A hierarchical multi-rate multi-precedence policer is disclosed. The policer discards packets based on assigned precedence levels. When traffic exceeds an available service rate, the policer drops packets of lower precedence levels to make room for packets of higher precedence levels. In certain implementations, the policer also guarantees bandwidth to each level, thus preventing complete loss of lower precedence traffic when there is a large amount of higher precedence traffic.
摘要:
Methods and systems for preemption in a network having a core device with at least one egress interface are disclosed. In one embodiment, the method includes performing flow-based hash utilizing a plurality of hash-buckets each set to a first state or a second state and computing a load based on a rate measurement that excludes flows which hash into hash-buckets having a state set to the second state. The computed load is compared to a preemption threshold and if the computed load exceeds the preemption threshold, the state of at least one of the hash-buckets is changed from the first state to the second state. An action, such as dropping all packets or marking all packets, is performed on flows hashing in a hash-bucket in the second state.
摘要:
A method and system for call admission and preemption in a network having an ingress node and an egress node defining an ingress/egress node pair, and one or more core nodes located in a path between the ingress node and the egress node and configured to mark packets exceeding an admission threshold. The method includes calculating a sustainable preemption rate based on a rate of traffic received at the egress node from the ingress node that passed through the one or more core nodes without being marked, and a ratio between a preemption threshold and the admission threshold. If a traffic load to the egress node is greater than the sustainable preemption rate, at least some traffic is dropped so that the traffic load does not exceed the sustainable preemption rate for the ingress/egress node pair.
摘要:
In one embodiment, head-end nodes receive a list of tunnels to be rerouted from a particular link of an intermediate node. If a head-end node is unable to reroute a tunnel for which it is the head-end node using conventional distributed routing, each head-end node executes the same algorithm to compute paths for all tunnels in the list (e.g., potentially applying bandwidth scaling).
摘要:
A method and system for call admission and preemption in a network having an ingress node and an egress node defining an ingress/egress node pair, and one or more core nodes located in a path between the ingress node and the egress node and configured to mark packets exceeding an admission threshold. The method includes calculating a sustainable preemption rate based on a rate of traffic received at the egress node from the ingress node that passed through the one or more core nodes without being marked, and a ratio between a preemption threshold and the admission threshold. If a traffic load to the egress node is greater than the sustainable preemption rate, at least some traffic is dropped so that the traffic load does not exceed the sustainable preemption rate for the ingress/egress node pair.