RSVP IGP synchronization can help improve the convergence time of RSVP and reduce the latency in removing the resource reservations, thereby improving the overall network efficiency.
When an IGP protocol declares a neighbor down, because hello packets are no longer being received, RSVP brings down all the associated LSPs and sessions that are passing through the down neighbor. However, the IGP protocols and RSVP still act independently when bringing a neighbor up.
RSVP sessions are maintained until either the router stops receiving IGP hello packets or the RSVP Path and Resv messages time out or RSVP states are explicitly torn down by the ingress or egress. Configuring a short time for the IS-IS or OSPF hello timers allows these protocols to detect node failures quickly. Also, configuring BFD for IGP interface provides sub-second neighbor down detection time. When quick discovery of a failed neighbor is needed, short IGP (OSPF or IS-IS) hello timers could be configured, or BFD could be enabled on IGP interfaces.
The RSVP IGP synchronization allows RSVP to react to an IGP neighbor down event. It does not allow RSVP to detect that a neighbor node has gone down. For example, when a pair of RSVP/IGP routers are connected with parallel links, detecting one neighbor down does not mean that the entire neighbor node has gone down.
RSVP IGP synchronization is independent of MPLS traffic engineering configurations. Irrespective of MPLS traffic engineering configuration (OSPF or IS-IS), RSVP IGP synchronization allows MPLS (RSVP) to handle IGP neighbor down events and take action, such as tearing down the associated RSVP sessions. For example, when IS-IS is configured as MPLS-TE protocol, the user can still configure MPLS to handle an OSPF neighbor down event (and vice versa).
An IGP neighbor down event is handled only by the RSVP sub-component of MPLS by tearing down the associated sessions. This event is not handled by LDP sub-component of MPLS.
MPLS/RSVP does not keep track of the current state of IGP neighbor. That is, when an IGP neighbor goes down, RSVP tears down all the associated sessions. However, RSVP does not prevent bringing up any session while the IGP neighbor to RSVP next-hop is down (or not yet available). That is, the RSVP session is brought up even when the IGP neighbor to the next-hop does not exist.
An IGP neighbor down is treated as upstream neighbor down or downstream neighbor down event by RSVP, depending upon the direction of the LSP. When a downstream IGP neighbor goes down, it results in an LSP tear down or FRR switchover, whichever is applicable.
MPLS receives and processes an IGP neighbor down event only for the cases when an IGP neighbor goes down because of hellos not received from the peer.
When an IGP neighbor goes down because of an underlying interface down, MPLS does not react to an IGP neighbor down event as RSVP would also receive the interface down event and tears down associated LSPs/sessions. Handling an IGP neighbor down event is redundant in such situations.
When BFD is configured on IGP interfaces, an IGP neighbor down is detected quickly and may help RSVP converge faster.
Bypass LSPs are treated the same way as regular LSPs. Upon an IGP neighbor down, associated bypass LSPs is torn down.
When an IGP neighbor is Nonstop Routing or Graceful Restart (NSR/GR) capable, MPLS does not receive a neighbor down event when NSR is performed on the peer IGP router.
Faster FRR feature is not be triggered when MPLS detects that IGP neighbor is down. Instead, each FRR LSP is processed individually to perform local repair.
It is highly recommended to observe extreme caution when implementing this feature when BFD is enabled for the underlying IGP. Under some circumstances, unnecessary flapping for RSVP sessions/LSPs can occur with this combination.
This command globally enables the handling of an IGP neighbor down event by MPLS. This command can be executed on the fly and takes effect immediately. It is possible to enable handling of neighbor down events for IS-IS.