Abstract: A device may receive, from a set of network devices, telemetry information associated with a first set of flows. The device may process, using a machine learning technique, the telemetry information to determine information that permits the set of network devices to identify a set of expected bandwidth values associated with a second set of flows. The device may provide, to a network device of the set of network devices, at least a portion of the information, that permits the set of network devices to identify the set of expected bandwidth values associated with the second set of flows, to permit the network device to determine an expected bandwidth value associated with a flow of the second set of flows and select a link to use when providing network traffic associated with the flow based on the expected bandwidth value.

Abstract: In general, techniques are described for dynamically controlling host-bound traffic by dynamically adding and updating, within the forwarding plane of a network device, network packet policers that each constrains, for one or more packet flows, an amount of host-bound traffic of the packet flows permitted to reach the control plane in accordance with available resources. In one example, a control plane of the network device detects internal congestion in the communication path from the forwarding plane to control plane (the “host-bound path”), identifies packet flows utilizing an excessive amount of host-bound path resources, computes limits for the identified packet flows, and adds “penalty-box policers” configured with the computed limits for the identified packet flows to the forwarding plane. The forwarding plane subsequently applies the policers to the identified packet flows to constrain the amount of traffic of the packet flows allowed to reach the control plane to the computed limits.

Abstract: A first system manager operating on a first node of a distributed computing system, receives data indicating a current state of the distributed computing system. The first system manager may determine, based at least in part on the current state of the distributed computing system and a set of rules for an application, an updated state of the distributed computing system. Furthermore, the first node may send the updated state of the distributed computing system to a second node of the distributed computing system. Responsive to receiving the updated state of the distributed computing system, a second system manager on the second node may modify a state of the second node. Modifying the state of the second node may comprise at least one of: starting the application on the second node, stopping the application on the second node, or modifying a state of the application on the second node.

Abstract: In one example, techniques of this disclosure may enable a point of local repair (PLR) network device to signal availability of link protection or node protection to a merge point (MP) network device and enable a network device to actively determine whether or not it is a merge point router. Based on whether or not the network device determines it is a MP, the network device may selectively clean up LSP states when there is an upstream link or node failure. The RSVP-TE protocol may be extended to enable a network device to send a tear down message to a downstream router, which may enable the downstream router to conditionally delete locale LSP state information. In some instances, a PLR network device may directly send a tear down message to a MP network device even though the PLR network device may not have a working bypass LSP.

Abstract: The disclosed computer-implemented method for preventing tromboning in inter-subnet traffic within data center architectures may include (1) detecting, at a leaf node of a data center, a route advertisement that advertises a route to a spine node of another data center that interfaces with the data center, (2) identifying, at the leaf node, an IP identifier of the spine node included in the route advertisement, (3) determining, at the leaf node, that the route corresponds to the spine node based at least in part on the IP identifier identified in the route advertisement, and then in response to determining that the route corresponds to the spine node, (4) rejecting the route to the spine node at the leaf node such that the leaf node does not learn the route to the spine node. Various other methods, systems, and apparatuses are also disclosed.

Abstract: A device may receive, from a first device associated with a first LAN, network traffic destined for a second LAN. The device may provide the first LAN with access to a core network. The device may not provide the second LAN with access to the core network. The device may identify, based on the network traffic, a Layer 3 address associated with a second device. The second device may be associated with the second LAN. The device may determine that the first device is categorized as a leaf device within an Ethernet Tree provided by the device. The device may determine, based on the Layer 3 address, that the second device is categorized as a leaf device within the Ethernet Tree. The device may drop the network traffic based on determining that the first device and the second device are categorized as leaf devices within the Ethernet Tree.

Abstract: The techniques described herein may dynamically adjust the sampling threshold based on a comparison of a flow export rate to a configured flow export rate. Based on the comparison of the flow export rate and the configured flow export rate, the network device may dynamically adjust the sampling threshold, such as increasing, reducing, or not changing the sampling threshold. Moreover, traffic flows are exported based on the adjusted sampling threshold. For example, if a number of bytes of a packet flow is more than or equal to the adjusted sampling threshold, network devices may export the sampled packet flow with the byte count and packet count of the sampled packet flow reported as-is. When a number of bytes of a packet flow is less than the adjusted sampling threshold, the packet flow will be exported with the byte count and packet count adjusted according to a probability.

Abstract: In one example, techniques of this disclosure may enable a point of local repair (PLR) network device to signal availability of link protection or node protection to a merge point (MP) network device and enable a network device to actively determine whether or not it is a merge point router. Based on whether or not the network device determines it is a MP, the network device may selectively clean up LSP states when there is an upstream link or node failure. The RSVP-TE protocol may be extended to enable a network device to send a tear down message to a downstream router, which may enable the downstream router to conditionally delete locale LSP state information. In some instances, a PLR network device may directly send a tear down message to a MP network device even though the PLR network device may not have a working bypass LSP.

Abstract: A device may cause an optical signal to be transmitted via a network path. The device may receive, from a network device, a link layer discover protocol (LLDP) message. The LLDP message may include signal characteristic information regarding the optical signal. The device may adjust transmission of the optical signal based on receiving the LLDP message. The device may cause an adjusted optical signal to be transmitted via the network path based on adjusting transmission of the optical signal.

Abstract: A first device may receive information that identifies a second device. The second device may be connected to the first device or a third device. The second device may be a source of traffic to be received by the first device. The first device may determine whether the second device is local or remote to the first device based on receiving the information. The first device may store first information or second information based on determining whether the second device is local or remote. The first information may identify a route associated with the second device. The second information may identify a single route associated with multiple second devices. The first device may provide the traffic using the first information or the second information after storing the first information or the second information.

Abstract: In one example, techniques of this disclosure may enable a point of local repair (PLR) network device to signal availability of link protection or node protection to a merge point (MP) network device and enable a network device to actively determine whether or not it is a merge point router. Based on whether or not the network device determines it is a MP, the network device may selectively clean up LSP states when there is an upstream link or node failure. The RSVP-TE protocol may be extended to enable a network device to send a tear down message to a downstream router, which may enable the downstream router to conditionally delete locale LSP state information. In some instances, a PLR network device may directly send a tear down message to a MP network device even though the PLR network device may not have a working bypass LSP.

Abstract: In general, techniques are described for simplifying admission control signaling between subscriber devices, access nodes, and service edge routers to facilitate subscriber-specific admission control for multicast streams. In one example, a service edge router receives a service request and accesses a subscriber profile to determine whether the requesting subscriber is authorized to receive the service. Upon authorization, the service edge router returns the service request to the access node in a substantially similar form in which the service request was received. The access node receives the service request on a service edge router-facing interface, indicating the service edge router has granted authorization for the subscriber device to receive multicast traffic associated with the multicast group identified within the service request.

Abstract: An optical coupling device can couple incident light, propagating orthogonal to a layered structure, into a layer of the layered structure. The device can include a lens having a lens central axis. The lens can focus a first beam to form a converging second beam. The first beam can have a first beam central axis that is offset from the lens central axis. The second beam can have a second beam central axis that is angled with respect to the first beam central axis. A planar grating can redirect the second beam to form a converging third beam. The third beam can have a third beam central axis that is parallel to a plane of the grating. Offsetting the first beam central axis from the lens central axis in this manner can help relax wavelength, manufacturing, and/or alignment tolerances, compared to a configuration in which there is no offset.

Abstract: A first system manager operating on a first node of a distributed routing system, receives data indicating a current state of the distributed routing system. The first system manager may determine, based at least in part on the current state of the distributed routing system and a set of rules for an application, an updated state of the distributed routing system. Furthermore, the first node may send the updated state of the distributed routing system to a second node of the distributed routing system. Responsive to receiving the updated state of the distributed routing system, a second system manager on the second node may modify a state of the second node. Modifying the state of the second node may comprise at least one of: starting the application on the second node, stopping the application on the second node, or modifying a state of the application on the second node.

Abstract: The disclosed computer-implemented method may include (1) receiving, at an upstream router of a multicast distribution tree, a packet that is destined for a receiver within an MPLS network, (2) identifying within the packet (A) a context label that specifies a controller and (B) a tree label that is assigned by the controller, (3) identifying a forwarding table that corresponds to the context label identified within the packet, (4) searching the forwarding table that corresponds to the context label for the tree label that specifies the multicast distribution tree, (5) identifying, based at least in part on the search, a downstream router of the multicast distribution tree that is to receive the packet on the way to the receiver, and then (6) forwarding the packet to the downstream router of the multicast distribution tree on the way to the receiver. Various other methods, systems, and apparatuses are also disclosed.

Abstract: A method performed by a network device may include assembling a multiprotocol label switching (MPLS) echo request, the echo request including an instruction for a transit node to forward the echo request via a bypass path associated with the transit node, and an instruction for an egress node to send an echo reply indicating that the echo request was received on the bypass path. The method may also include sending the MPLS echo request over a functioning label switched path (LSP).

Abstract: The liveness of routing protocols can be determined using a mechanism to aggregate liveness information for the protocols. The ability of an interface to send and receive packets and the forwarding capability of an interface can also be determined using this mechanism. Since liveness information for multiple protocols, the liveness of interfaces, the forwarding capability of interfaces, or both, may be aggregated in a message, the message can be sent more often than could individual messages for each of the multiple protocols. This allows fast detection of failures, and sending connectivity messages for the individual protocols, such as neighbor “hellos,” to be sent less often.

Abstract: An apparatus includes a reconfigurable optical add/drop multiplexer (ROADM) having an input port to receive a first optical signal from a second device. The ROADM also includes a first wavelength selective switch (WSS), in optical communication with the input port, to convert the first optical signal into a second optical signal, a loopback, in optical communication with the first WSS, to transmit the second optical signal, and a second WSS, in optical communication with the loopback, to convert the second optical signal to a third optical signal and direct the third optical signal back to the second device via the input port.

Abstract: Methods and systems are presented for heterogeneous integration of photonics and electronics with atomic layer deposition (ALD) bonding. One method includes operations for forming a compound semiconductor and for depositing (e.g., via atomic layer deposition) a continuous film of a protection material (e.g., Al2O3) on a first surface of the compound semiconductor. Further, the method includes an operation for forming a silicon on insulator (SOI) wafer, with the SOI wafer comprising one or more waveguides. The method further includes bonding the compound semiconductor at the first surface to the SOI wafer to form a bonded structure and processing the bonded structure. The protection material protects the compound semiconductor from acid etchants during further processing of the bonded structure.

Abstract: In some embodiments, a non-transitory processor-readable medium stores code representing instructions to be executed by a processor that cause the processor to receive a data unit having a header portion. The code causes the processor to select a route based on the header portion and modify, based on the route being associated with a tunnel, the data unit to define a tunnel data unit including a tunnel header. The code causes the processor to select a loopback link from a set of loopback links of a loopback link aggregation group (LAG). The code further causes the processor to receive via the loopback link, the tunnel data unit and send the tunnel data unit via the tunnel based on the tunnel header.