摘要:
In one embodiment, an apparatus generally comprises one or more input interfaces for receiving a plurality of flows, a plurality of output interfaces, and a processor operable to identify large flows and select one of the output interfaces for each of the large flows to load-balance the large flows over the output interfaces. The apparatus further includes memory for storing a list of the large flows, a pinning mechanism for pinning the large flows to the selected interfaces, and a load-balance mechanism for selecting one of the output interfaces for each of the remaining flows. A method for local placement of large flows to assist in load-balancing is also disclosed.
摘要:
In one embodiment, an apparatus generally comprises one or more input interfaces for receiving a plurality of flows, a plurality of output interfaces, and a processor operable to identify large flows and select one of the output interfaces for each of the large flows to load-balance the large flows over the output interfaces. The apparatus further includes memory for storing a list of the large flows, a pinning mechanism for pinning the large flows to the selected interfaces, and a load-balance mechanism for selecting one of the output interfaces for each of the remaining flows. A method for local placement of large flows to assist in load-balancing is also disclosed.
摘要:
A scheduling method and system for a multi-level class hierarchy are disclosed. The hierarchy includes a root node linked to at least two groups. One of the groups has priority over the other of the groups and comprises at least one high priority queue and at least one low priority queue. The method includes receiving traffic at the root node, directing traffic received at the root node to one of the groups, and directing traffic received at the priority group to one of the high priority and low priority queues. Packets are accepted at the high priority queue or the low priority queue if a specified rate is not exceeded at the high and low priority queues and at least some packets are dropped at the low priority queue if the specified rate is exceeded at the high and low priority queues.
摘要:
Real-time customer packet traffic is instrumented to determine measured delays between two or more points along a path actually traveled by a packet, such as within or external to one or more packet switching devices. These measurements may include delays within a packet switching device other than the ingress and egress time of a packet. These measured delays can be used to determine whether or not the performance of a packet switching device or network meets desired levels, especially for complying with a Service Level Agreement.
摘要:
Disclosed are, inter alia, methods, apparatus, computer-readable media, mechanisms, and means for instrumenting real-time customer packet traffic. These measured delays can be used to determine whether or not the performance of a packet switching device and/or network meets desired levels, especially for complying with a Service Level Agreement.
摘要:
In one embodiment, an Internet Protocol (IP) routing information base of a packet switching device is filtered to produce a significantly smaller subset of IP routes that are installed in one or more forwarding information bases for forwarding of IP packets. In one embodiment, these smaller forwarding information bases are located in memory local to a network processor to more quickly perform lookup operations thereon. In one embodiment, one or more of these forwarding information bases is used only for exact matching of addresses (not longest prefix matching). In one embodiment, the IP routes in these smaller forwarding information bases substantially correspond to packet switching devices in a network (e.g., core and edge routers), such as in contrast to including all the IP routes for devices external to the network.
摘要:
In one embodiment, an Internet Protocol (IP) routing information base of a packet switching device is filtered to produce a significantly smaller subset of IP routes that are installed in one or more forwarding information bases for forwarding of IP packets. In one embodiment, these smaller forwarding information bases are located in memory local to a network processor to more quickly perform lookup operations thereon. In one embodiment, one or more of these forwarding information bases is used only for exact matching of addresses (not longest prefix matching). In one embodiment, the IP routes in these smaller forwarding information bases substantially correspond to packet switching devices in a network (e.g., core and edge routers), such as in contrast to including all the IP routes for devices external to the network.
摘要:
A method, performed in a network packet routing element, comprises establishing a forwarding information base (FIB) lacking a hierarchical data structure but in which one or more dependent FIB entries are associated with a parent FIB entry; establishing a plurality of strict priority queues, each having an associated priority; receiving a change to the parent FIB entry; for each of the dependent FIB entries, selecting one of the queues and enqueuing the dependent FIB entries in the selected queues for re-resolution; dequeuing the dependent FIB entries for re-resolution, according to a priority order of the queues.
摘要:
In one embodiment, a list of border node next hop options is maintained in a memory. The list of border node next hop options includes one or more of border nodes that may be utilized to reach one or more prefixes. An index value is associated with each border node of the list of border node next hop options. A list of labels is also maintained in the memory. The index value of each border node is associated with a corresponding label for a path to reach that border node. When a change to the one or more border nodes is detected, the list of border node next hop options is updated to remove a border node. However, a label for the path to reach the border node is maintained in the list of labels for at least a period of time.
摘要:
In one embodiment, one or more virtual private network (VPN) prefixes may be grouped at a network node into sets having shared network border node next-hop options, where each border node has a defined index value associated therewith. Also, a list of VPN labels associated with each VPN prefix may be maintained by the network node, where each VPN label is associated with a border node of a particular set by a corresponding index value. Further, the network node may determine a particular border node for traffic to be forwarded, along with the defined index value. The network node may then apply the index value to select an associated VPN label, and may affix the selected VPN label to the traffic for forwarding.