Systems and methods for path record handling in a fabric without host stack cooperation in a high performance computing environment. In a case where the subnet manager has determined homogenous subnet/fabric or semi-homogenous subnet/fabric status for the current topology, but is still receiving path queries, the subnet manager can use the relevant status to avoid any route evaluation and generate path record either only based on the configuration status of the requesting port in the homogenous case, or by comparing the configuration status of both ports in the semi-homogenous case.

A system and method for building a data structure for determining if multiple routing or forwarding tables yield the same or different forwarding behaviors. VeriTable uses a single tree/trie traversal to quickly check if multiple forwarding tables are forwarding equivalent, yield routing loops, or black holes. VeriTable also uses the Longest Prefix Matching rule for lookups, e.g., checking if route updates in control plane are consistent with the ones in forwarding plane. VeriTable can be applied to network-wide abnormality diagnosis of network problems, such as scalable and efficient forwarding loop detection and avoidance in the data plane of a network. In addition, VeriTable can be extended to handle incremental updates applied to the forwarding tables in a network.

In one embodiment, a first Protocol Independent Multicast (PIM) router includes port interfaces to receive multicast traffic from a first network and forward the traffic to at least one receiver, which is in a sub-network including other PIM routers, and a routing processor configured, in response to a decision for the first PIM router to relinquish being a designated router, to generate a PIM Hello message with a first option descriptor and a first priority, the first option descriptor indicating a staggered handoff process, and send the PIM Hello message, receipt of the PIM Hello message by the other PIM routers being operative to result in a designated router election electing a new designated router, the new designated router being operative to initiate the staggered handoff process causing the first PIM router to continue forwarding traffic until the new designated router has built a multicast routing tree.

A network node device and method of determining a communication route to one or more other network nodes through a network. The method includes sending current routing information to a network management server (NMS), and receiving new or supplemental routing information from the NMS, this supplemental routing information determined by the NMS based on the current routing information of the network node and of the one or more other network nodes. The supplemental routing information may include lateral route information identifying designated routing nodes that form lateral band(s) of nodes that span the network, each lateral band including gate node(s) as entrances/exits to the lateral band. The method may further include determining, based on the supplemental routing information, a route to one or more of the other network nodes. A lateral band may facilitate a route through a chokepoint or other abnormal topological layout.

An in-band telemetry congestion notification system includes a source device that generates and transmits packets. Each of a plurality of switch devices receives each of the packets, provides in-band telemetry information in each of the packets, and transmits each of the packets to a destination device. When the destination device receives a first subset of the packets, it uses the in-band telemetry information included in each of the first subset of the packets to determine a telemetry information baseline. Subsequent to determining the telemetry information baseline, when the destination device receives a second subset of the packet, it uses the in-band telemetry information included in each of the second subset of the packets to determine a departure from the telemetry information baseline that indicates congestion and, in response, generates a congestion notification and transmits the congestion notification via the plurality of switch devices to the source device.

A first network device joins a redundancy group of a multihomed network element that is connected to a computing device. The first network device detects an active link between the first network device and the computing device, and receives from the computing device, a subscription to a multicast flow. The first network device adds the subscription to a first list of active multicast subscriptions stored on the first network device. The first network device sends a unicast notification to a second network device of the multihomed network element. The unicast notification causes the subscription to be added to a second list of active multicast subscriptions stored on the second network device.

One technique includes receiving, in a first network, a multi-destination packet from a second network, and determining, based on the multi-destination packet, a first multi-destination tree in the first network for forwarding the multi-destination packet. In response to determining that the first multi-destination tree is not rooted on the network device, a second multi-destination tree in the first network is determined, and the multi-destination packet is transmitted using the second multi-destination tree. Another technique includes, upon detecting a first network device joining a network, sending a first indication to a second network device that the first network device is in a state for an amount of time. After the amount of time has elapsed, a second indication that the first network device has exited the state is sent to the second network device. A topology of the network is updated after the first network device has exited the state.

In multicast management, topology information identifying multicast distribution paths is maintained at a router serving as a root of a multicast distribution tree and/or as a first hop router. The router can detect congestion or failures based on messages from other routers. The router can request another router to change the multicast distribution paths as needed to load-balance the traffic, avoid congestion or failure, or eliminate duplicate traffic. In some embodiments, ECMP Redirect functionality is expanded to ECMP paths of arbitrary number of hops. Other features are also provided.

Described herein are systems, methods, and software to identify propagation risk of threats in a computing environment. In one implementation, a management service may identify a connection tree for a computing environment based on forwarding rules for virtual nodes in the computing environment. The management service may further, for each connection in the connection tree, determine a threat value based at least on a protocol associated with the connection. The management service may also identify a threat to a virtual node of the virtual nodes and generate a threat propagation summary for the threat based on the one or more minimum or maximum spanning trees.

A wireless mesh network is described. The mesh network uses a plurality of data communications nodes that are organized in a tree-like structure. The network also includes an access server which communicates with the nodes. The mesh network nodes are one or more root access point nodes having a wired connection to an external network; and one or more mesh access point nodes. Both types of nodes communicate with an external network. Each non-root node automatically connects to an associated parent node selected from one or more nodes within a direct wireless communication range of the node. The node selects a parent node based on one or more parent selection criteria from the access server and establishes a connection to the selected parent node. The node then calculates its routing path to the external network. The nodes include unique identifiers and two or more radios.