Abstract: A network switch having hardware thereon for transmitting probes to neighbor devices for exercising forwarding states (e.g., layer 2 and layer 3) on the switch. A light-weight agent resides on one or both of neighbor network devices and can be used to control the testing. Probe allocation can be managed locally on a source device based on a layer 3 routing table. One or more probes originating from the source network device (device A) from a local CPU are routed on the same network device A in hardware and sent out on a link towards a peer device (device B). Peer device B captures the probe using an Access Control List (ACL) hardware, and reflects the probe back to network device A on the ingress port. Network device A can then capture the reflected probe using ACL hardware and verifies the actual forwarding behavior applied on the probe packet.

Abstract: A monitoring solution allows for testing of forwarding states on network devices. In a particular example, an agent on a router A directs a probe packet to a first neighbor device. The first neighbor device reflects the probe back to the router A. The reflected probe packet undergoes a forwarding state lookup similar to other network traffic and is routed to a second neighbor network device. The second neighbor network device reflects the probe back to router A, which can then intercept the packet and redirect it to an agent on router A for verification whether the lookup was performed correctly.

Abstract: In accordance with one example embodiment, a system configured for programming a network layer multicast address entry in a routing table of an ingress line card module is disclosed. The network layer multicast address entry includes a network layer address associated with at least one egress line card. The system is further configured for programming a data link layer multicast routing address entry in a routing table of a fabric card module in which the data link layer multicast routing address entry corresponds to the network layer multicast address entry.

Abstract: In accordance with one example embodiment, a system configured for programming a network layer multicast address entry in a routing table of an ingress line card module is disclosed. The network layer multicast address entry includes a network layer address associated with at least one egress line card. The system is further configured for programming a data link layer multicast routing address entry in a routing table of a fabric card module in which the data link layer multicast routing address entry corresponds to the network layer multicast address entry.

Abstract: A source network device can inject a probe into any one of multiple pipelines by supplying control bits for controlling a demultiplexer through which the probe passes. The control bits of the probe are coupled to the demultiplexer select lines so as to couple an input port of the demultiplexer to one of multiple output ports. As a result, an agent operating on the source network device can test any desired pipeline. In a one-hop embodiment, the source network device redirects the probe back to the agent without transmission to a neighbor. A two-hop embodiment uses a neighbor device to reflect the probe back to the source device in order to test forwarding states of the source network device.

Abstract: A deterministic model is described that is used for testing networks by exercising forwarding rules (e.g., layer 2 and layer 3) on network devices. Probe detection on a peer network device B can be implemented using a global register available to all input pipelines. The global register can be used to check a source and destination port in a User Datagram Protocol (UDP) header. If there is a match, the packet is considered a probe, and a probe detection signal is transmitted to an output pipeline to redirect the probe back to the input port. Network device A can then capture the reflected probe using layer 2 hardware and redirect it to the CPU in order to verify the actual forwarding behavior applied on the probe packet. In an alternative embodiment, probe detection logic can be incorporated in an ACL at a beginning of a pipeline for switching packets.

Abstract: A deterministic model is described that is used for testing networks by exercising forwarding rules (e.g., layer 2 and layer 3) on network devices. Within a single network hop, a light-weight agent can be used to control the testing. One or more probe packets can be injected into an ingress pipeline of a network device using the agent executing on a local processor. The probes are detected after performing at least layer 2 and layer 3 lookups. Hardware in switching logic redirects the probes to the local processor in order to verify the actual forwarding behavior applied on the probe packet.

Abstract: A network switch having hardware thereon for transmitting probes to neighbor devices for exercising forwarding states (e.g., layer 2 and layer 3) on the switch. A light-weight agent resides on one or both of neighbor network devices and can be used to control the testing. One or more probes originating from the source network device (device A) from a local CPU are routed on the same network device A in hardware and sent out on a link towards a peer device (device B). Peer device B captures the probe using an Access Control List (ACL) hardware, and transmits the probes to a local agent for processing. The agent then transmits the probe back to the source network device and the source network device can determine whether the FIB lookup was correctly performed.

Abstract: A network switch having hardware thereon for transmitting probes to neighbor devices for exercising forwarding states (e.g., layer 2 and layer 3) on the switch. A light-weight agent resides on one or both of neighbor network devices and can be used to control the testing. One or more probes originating from the source network device (device A) from a local CPU are routed on the same network device A in hardware and sent out on a link towards a peer device (device B). The probes purposefully include a time-to-live (TTL) parameter designed to expire when the probe reaches the peer. Peer device B captures the probe using layer 3 hardware because of the TTL parameter generates an error, which causes the probe to be transmitted to a CPU. The CPU transmits the probe back to network device A as a TTL expiration error.

Abstract: Disclosed are systems and methods for scaling Massively Scalable Data Center (MSDC) networks with a large number of end-point tunnels utilizing Equal-cost multi-path routing (ECMP). The systems and methods can use the NO-OP label operations to maintain single ECMP objects to switch a set of segment routing tunnels that share the same ECMP links. The forwarding engine can determine the use of the NO-OP label operation and update a received packet to enable the use of the single ECMP objects of the set of segment routing tunnels.

Abstract: A network device is described that injects probes for exercising its own forwarding rules (e.g., layer 2 and layer 3). One or more probe packets can be injected into an ingress pipeline of the network device using the agent executing on a local processor. The probes are detected before or after performing at least layer 2 and layer 3 lookups. Hardware in switching logic redirects the probes to an external testing system in order to verify the actual forwarding behavior applied to the probe packet. In order to deliver the probes to the external server, the network device performs an additional layer 3 lookup and generates an encapsulated packet incorporating results of a lookup on the probe packet. In this way, the external server can analyze the lookup on the probe packet to determine whether the lookup was performed correctly.

Abstract: In one illustrative example, an IP network media data router includes a spine and leaf switch architecture operative to provide IP multicast delivery of media data from source devices to receiver devices. The architecture may include K spine switches, K sets of L leaf switches, M data links between each leaf switch and each spine switch where each data link has a maximum link bandwidth of BWL, and a plurality of bidirectional data ports connected to each leaf switch. Notably, the router is provided or specified with a number of bidirectional data ports N=(a/K)×(BWL/BWP) for a guaranteed non-blocking IP multicast delivery of data at a maximum port bandwidth of BWP, where “a” is a fixed constant greater than or equal to K. The architecture may be reconfigurable or expandable to include C additional spine switches and C additional sets of L leaf switches.

Abstract: Various embodiments are disclosed for increasing Layer-3 LPM (longest prefix match) routing database in a network platform. In some embodiments, chipsets in fabric modules (FMs) can be partitioned into multiple banks. Network traffic can be directed towards a corresponding bank in the FMs by using a LPM table on a line card (LC). Entries in the LPM table on the LC can be programmed either statically or dynamically based upon LPM routes that are dynamically learned.

Abstract: In accordance with one example embodiment, a system configured for programming a network layer multicast address entry in a routing table of an ingress line card module is disclosed. The network layer multicast address entry includes a network layer address associated with at least one egress line card. The system is further configured for programming a data link layer multicast routing address entry in a routing table of a fabric card module in which the data link layer multicast routing address entry corresponds to the network layer multicast address entry.

Abstract: Disclosed are systems and methods for scaling Massively Scalable Data Center (MSDC) networks with a large number of end-point tunnels utilizing Equal-cost multi-path routing (ECMP). The systems and methods can use the NO-OP label operations to maintain single ECMP objects to switch a set of segment routing tunnels that share the same ECMP links. The forwarding engine can determine the use of the NO-OP label operation and update a received packet to enable the use of the single ECMP objects of the set of segment routing tunnels.

Abstract: Aspects of the subject technology relate to systems for arbitrating direct forwarder (“DF”) instantiation between VPC peers used to facilitating the transport of bidirectional multicast traffic over a L2/L3 network boundary. In some aspects, arbitration of DF instantiation on a given VPC peer can include determining a first set of metrics for a first VPC switch, determining a second set of metrics for a second VPC switch, and determining, at the first VPC switch, whether to instantiate a designated forwarder (DF) operation based on a comparison of the first set of metrics and the second set of metrics. Methods and machine-readable media are also provided.

Abstract: In accordance with one example embodiment, a system configured for programming a network layer multicast address entry in a routing table of an ingress line card module is disclosed. The network layer multicast address entry includes a network layer address associated with at least one egress line card. The system is further configured for programming a data link layer multicast routing address entry in a routing table of a fabric card module in which the data link layer multicast routing address entry corresponds to the network layer multicast address entry.

Abstract: Disclosed are systems and methods for scaling Massively Scalable Data Center (MSDC) networks with a large number of end-point tunnels utilizing Equal-cost multi-path routing (ECMP). The systems and methods can use the NO-OP label operations to maintain single ECMP objects to switch a set of segment routing tunnels that share the same ECMP links. The forwarding engine can determine the use of the NO-OP label operation and update a received packet to enable the use of the single ECMP objects of the set of segment routing tunnels.

Abstract: Various embodiments are disclosed for increasing Layer-3 LPM (longest prefix match) routing database in a network platform. In some embodiments, chipsets in fabric modules (FMs) can be partitioned into multiple banks. Network traffic can be directed towards a corresponding bank in the FMs by using a LPM table on a line card (LC). Entries in the LPM table on the LC can be programmed either statically or dynamically based upon LPM routes that are dynamically learned.

Abstract: Disclosed are systems and methods for scaling Massively Scalable Data Center (MSDC) networks with a large number of end-point tunnels utilizing Equal-cost multi-path routing (ECMP). The systems and methods can use the NO-OP label operations to maintain single ECMP objects to switch a set of segment routing tunnels that share the same ECMP links. The forwarding engine can determine the use of the NO-OP label operation and update a received packet to enable the use of the single ECMP objects of the set of segment routing tunnels.