Fast Reroute (FRR) is a mechanism for protecting MPLS TE LSPs from link and node failures by locally repairing the LSPs at the point of failure, allowing data to continue to flow on them while their headend routers attempt to establish new end-to-end LSPs to replace them. FRR locally repairs the protected LSPs by rerouting them over backup tunnels that bypass failed links or node.

Backup tunnels that bypass only a single link of the LSP’s path provide link protection. They protect LSPs if a link along their path fails by rerouting the LSP’s traffic to the next hop (bypassing the failed link). These are referred to as next-hop (NHOP) backup tunnels because they terminate at the LSP’s next hop beyond the point of failure. The figure below illustrates an NHOP backup tunnel.

Figure 1. NHOP Backup Tunnel

FRR provides node protection for LSPs. Backup tunnels that bypass next-hop nodes along LSP paths are called next-next-hop (NNHOP) backup tunnels because they terminate at the node following the next-hop node of the LSP paths, thereby bypassing the next-hop node. They protect LSPs if a node along their path fails by enabling the node upstream of the failure to reroute the LSPs and their traffic around the failed node to the next-next hop. FRR supports the use of RSVP Hellos to accelerate the detection of node failures. NNHOP backup tunnels also provide protection from link failures, because they bypass the failed link and the node.

The figure below illustrates an NNHOP backup tunnel.

Figure 2. NNHOP Backup Tunnel

If an LSP is using a backup tunnel and something changes so that the LSP is no longer appropriate for the backup tunnel, the LSP is torn down. Such changes are the following:

• Backup bandwidth of the backup tunnel is reduced.

• Backup bandwidth type of backup tunnel is changed to a type that is incompatible with the primary LSP.

• Primary LSP is modified so that FRR is disabled. (The no mpls traffic-eng fast-reroute command is entered.)

NHOP and NNHOP backup tunnels can be used to provide bandwidth protection for rerouted LSPs. This is referred to as backup bandwidth. You can associate backup bandwidth with NHOP or NNHOP backup tunnels. This informs the router of the amount of backup bandwidth a particular backup tunnel can protect. When a router maps LSPs to backup tunnels, bandwidth protection ensures that an LSP uses a given backup tunnel only if there is sufficient backup bandwidth. The router selects which LSPs use which backup tunnels in order to provide maximum bandwidth protection. That is, the router determines the best way to map LSPs onto backup tunnels in order to maximize the number of LSPs that can be protected. For information about mapping tunnels and assigning backup bandwidth, see the "Backup Tunnel Selection Procedure" section.

LSPs that have the “bandwidth protection desired” bit set have a higher right to select backup tunnels that provide bandwidth protection; that is, those LSPs can preempt other LSPs that do not have that bit set. For more information, see the "Prioritizing Which LSPs Obtain Backup Tunnels with Bandwidth Protection" section.

RSVP Hello enables RSVP nodes to detect when a neighboring node is not reachable. This provides node-to-node failure detection. When such a failure is detected, it is handled in a similar manner as a link-layer communication failure.

RSVP Hello can be used by FRR when notification of link-layer failures is not available (for example, with Fast Ethernet), or when the failure detection mechanisms provided by the link layer are not sufficient for the timely detection of node failures.

A node running Hello sends a Hello Request to a neighboring node every interval. If the receiving node is running Hello, it responds with Hello Ack. If four intervals pass and the sending node has not received an Ack or it receives a bad message, the sending node declares that the neighbor is down and notifies FRR.

There are two configurable parameters:

• Hello interval--Use the ip rsvp signalling hello refresh interval command.

• Number of acknowledgment messages that are missed before the sending node declares that the neighbor is down--Use the ip rsvp signalling hello refresh misses command

A Hello instance implements RSVP Hello for a given router interface IP address and remote IP address. A large number of Hello requests are sent; this puts a strain on the router resources. Therefore, create a Hello instance only when it is necessary and delete it when it is no longer needed.

There are two types of Hello instances:

Backup tunnel support has the following capabilities:

Backup bandwidth protection has the following capabilities:

RSVP Hello enables a router to detect when a neighboring node has gone down but its interface to that neighbor is still operational. This feature is useful when next-hop node failure is not detectable by link layer mechanisms, or when notification of link-layer failures is not available (for example, Gigabit Ethernet). This allows the router to switch LSPs onto its backup tunnels and avoid packet loss.

For a more detailed description of RSVP Hello, see the RSVP Hello Operation.

This section assumes that you want to add FRR protection to a network in which MPLS TE LSPs are configured.

backup bandwidth --The usage of NHOP and NNHOP backup tunnels to provide bandwidth protection for rerouted LSPs.

backup tunnel --An MPLS TE tunnel used to protect other (primary) tunnels’ traffic when a link or node failure occurs.

bandwidth --The available traffic capacity of a link.

Cisco Express Forwarding --A means for accelerating the forwarding of packets within a router, by storing route lookup.

enterprise network --A large and diverse network connecting most major points in a company or other organization.

Fast Reroute --Procedures that enable temporary routing around a failed link or node while a new LSP is being established at the headend.

global pool --The total bandwidth allocated to an MPLS traffic engineering link or node.

headend --The router that originates and maintains a given LSP. This is the first router in the LSP’s path.

hop --Passage of a data packet between two network nodes (for example, between two routers).

instance --A Hello instance implements the RSVP Hello extensions for a given router interface address and remote IP address. Active Hello instances periodically send Hello Request messages, expecting Hello ACK messages in response. If the expected ACK message is not received, the active Hello instance declares that the neighbor (remote IP address) is unreachable (that is, it is lost). This can cause LSPs crossing this neighbor to be fast rerouted.

interface --A network connection.

Intermediate System-to-Intermediate System --IS-IS. Link-state hierarchical routing protocol that calls for intermediate system (IS) routers to exchange routing information based on a single metric to determine network topology.

link --A point-to-point connection between adjacent nodes. There can be more than one link between adjacent nodes. A link is a network communications channel consisting of a circuit or transmission path and all related equipment between a sender and a receiver. Sometimes referred to as a line or a transmission link.

limited backup bandwidth --Backup tunnels that provide bandwidth protection.

load balancing --A configuration technique that shifts traffic to an alternative link if a certain threshold is exceeded on the primary link. Load balancing is similar to redundancy in that if an event causes traffic to shift directions, alternative equipment must be present in the configuration. In load balancing, the alternative equipment is not necessarily redundant equipment that operates only in the event of a failure.

LSP --label-switched path. A connection between two routers in which MPLS forwards the packets.

merge point --The backup tunnel’s tail.

MPLS --Multiprotocol Label Switching. Packet-forwarding technology, used in the network core, that applies data link layer labels to tell switching nodes how to forward data, resulting in faster and more scalable forwarding than network layer routing normally can do.

MPLS global label allocation --There is one label space for all interfaces in the router. For example, label 100 coming in one interface is treated the same as label 100 coming in a different interface.

NHOP --next hop. The next downstream node along an LSP’s path.

NHOP backup tunnel --next-hop backup tunnel. Backup tunnel terminating at the LSP’s next hop beyond the point of failure, and originating at the hop immediately upstream of the point of failure. It bypasses a failed link, and is used to protect primary LSPs that were using this link before the failure.

NNHOP --next-next hop. The node after the next downstream node along an LSP’s path.

NNHOP backup tunnel --next-next-hop backup tunnel. Backup tunnel terminating at the LSP’s next-next hop beyond the point of failure, and originating at the hop immediately upstream of the point of failure. It bypasses a failed link or node, and is used to protect primary LSPs that were using this link or node before the failure.

node --Endpoint of a network connection or a junction common to two or more lines in a network. Nodes can be interconnected by links, and serve as control points in the network. Nodes can be processors, controllers, or workstations.

OSPF --Open Shortest Path First. A link-state hierarchical Interior Gateway Protocol routing algorithm, derived from the IS-IS protocol. OSPF features include least-cost routing, multipath routing, and load balancing.

primary LSP --The last LSP originally signaled over the protected interface before the failure. The primary LSP is the LSP before the failure.

primary tunnel --Tunnel whose LSP may be fast rerouted if there is a failure. Backup tunnels cannot be primary tunnels.

promotion --Conditions, such as a new backup tunnel comes up, cause a reevaluation of a backup tunnel that was chosen for an LSP. If the reevaluation is successful, it is called a promotion.

protected interface --An interface that has one or more backup tunnels associated with it.

redundancy --The duplication of devices, services, or connections so that, in the event of a failure, the redundant devices, services, or connections can perform the work of those that failed.

RSVP --Resource Reservation Protocol. A protocol used for signaling requests (setting up reservations) for Internet services by a customer before that customer is permitted to transmit data over that portion of the network.

scalability --An indicator showing how quickly some measure of resource usage increases as a network gets larger.

SRLG --shared risk link group. Sets of links that are likely to go down together.

state --Information that a router must maintain about each LSP. The information is used for rerouting tunnels.

sub-pool --The more restrictive bandwidth in an MPLS traffic engineering link or node. The subpool is a portion of the link or node’s overall global pool bandwidth.

tailend --The router upon which an LSP is terminated. This is the last router in the LSP’s path.

topology --The physical arrangement of network nodes and media within an enterprise networking structure.

tunnel --Secure communications path between two peers, such as two routers.

unlimited backup bandwidth --Backup tunnels that provide no bandwidth (best-effort) protection (that is, they provide best-effort protection).