cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1037
Views
0
Helpful
3
Replies

MPLS Label Distribution Protocol Keepalive Messages

Mitrixsen
Level 1
Level 1

Hello, everyone!


I've built a simple small

MPLS topology

which consists of two routers only to see how LDP works in action.

Mitrixsen_0-1694621818650.png

Once enabled using the

mpls ip 

command, the routers exchanged LDP multicast hellos and setup an adjacency.

My question is, why are there periodical unicast LDP Keepalives that are being sent between the neighbors while they are also both sending LDP multicast messages?

Mitrixsen_1-1694621978947.png

Wouldn't the multicast messages act as a sufficient keepalive by themselves? Why is there the need for the LDP Unicast Keepalive?

Thank you in advance.


David

 

 

2 Accepted Solutions

Accepted Solutions

The multicast UDP hellos are used to discover potential LDP peers on the local link. Unless you are using targeted LDP, the local node has no knowledge of potential peers (ie, it does not have them explicitly configured).

Exchanging hellos establishes a hello adjacency and provides each peer with the IP address of the other, allowing them to open a unicast TCP session to exchange labels. The unicast keepalive message is intended to monitor the integrity of this particular TCP session between the LDP peers, while the multicast LDP hello messages have no particular intended recipient and are agnostic to any established TCP session.

Disclaimers: I am long in CSCO. Bad answers are my own fault as they are not AI generated.

View solution in original post

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello @Mitrixsen ,

the multicast Hellos are just for new neighbor discovery.

The unicast UDP keepalives are intended to verify the usability of the specific link for the LDP TCP session between loopbacks.

In general multiple links in ECMP can exist between the two LSR nodes and this UDP unicast keepalive checks the health of each physical link between the two nodes.

The TCP keepalive between loopbacks can be re-routed in case of failure however each device needs to know how many paths are usable to the MPLS LDP peer.

It could rely on IGP but this unicast keepalives provide an indipendent way to check.

Hope to help

Giuseppe

 

View solution in original post

3 Replies 3

The multicast UDP hellos are used to discover potential LDP peers on the local link. Unless you are using targeted LDP, the local node has no knowledge of potential peers (ie, it does not have them explicitly configured).

Exchanging hellos establishes a hello adjacency and provides each peer with the IP address of the other, allowing them to open a unicast TCP session to exchange labels. The unicast keepalive message is intended to monitor the integrity of this particular TCP session between the LDP peers, while the multicast LDP hello messages have no particular intended recipient and are agnostic to any established TCP session.

Disclaimers: I am long in CSCO. Bad answers are my own fault as they are not AI generated.

M02@rt37
VIP
VIP

Hello @Mitrixsen,

Unicast messages are preferred for detecting session failures because they provide a more direct and reliable means of communication between the routers involved in the LDP session. This helps ensure rapid detection and recovery from any issues that may occur in the network.

Unicast messages can be targeted specifically to the remote LDP peer, providing a direct communication path. This allows the routers to quickly detect if the peer becomes unresponsive. Multicast messages are broadcast to a group of potential peers and are typically used during the session initialization process when the exact peer is not yet known.

Best regards
.ı|ı.ı|ı. If This Helps, Please Rate .ı|ı.ı|ı.

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello @Mitrixsen ,

the multicast Hellos are just for new neighbor discovery.

The unicast UDP keepalives are intended to verify the usability of the specific link for the LDP TCP session between loopbacks.

In general multiple links in ECMP can exist between the two LSR nodes and this UDP unicast keepalive checks the health of each physical link between the two nodes.

The TCP keepalive between loopbacks can be re-routed in case of failure however each device needs to know how many paths are usable to the MPLS LDP peer.

It could rely on IGP but this unicast keepalives provide an indipendent way to check.

Hope to help

Giuseppe