Topology-Independent Loop-Free Alternates (TI-LFA)
The Juniper cloud-native router (JCNR) supports Topology-Independent Loop-Free Alternates (TI-LFA) with fast reroute (FRR) for SR-MPLS implementations. It supports protection against link-failure that is detected by the Poll Mode Drivers (PMD).
TI-LFA Overview
Fast Reroute (FRR)
Fast Reroute (FRR) technology is used in modern routing protocols to minimize disruption and packet loss in the event of a network failure. Unlike traditional routing protocols, FRR speeds up the convergence process by proactively pre-computing backup paths for critical traffic flows. The backup paths are readily available and instantly activated upon failure, ensuring minimum loss of data communication.
Loop-Free Alternates (LFAs) play a critical role in FRR implementations. LFAs are pre-calculated, loop-free backup paths that are leveraged by FRR protocols to quicky switch traffic to the chosen backup path.
There are two key aspects to FRR:
-
Detection:
-
Fault Detection—Rapidly identify link or node failures within a network through continuous monitoring of network elements.
-
Detection Methods—Applying one of the various methods for fault detection, including proactive methods such as periodic link probing or reactive methods that sense failure when it occurs. In some cases Bidirectional Forwarding Detection (BFD) may be used for quick fault detection. Cloud-Native Router detects failure by monitoring the link status events from the Poll Mode Driver (PMD).
-
-
Handling:
-
Precomputed Backup Paths—Using predetermined and optimised backup paths that are quickly activated when a failure is detected.
-
Fast Switchover—Network devices participating in the FRR process quickly switching to precomputed backup paths to minimize the impact on the network's forwarding capabilities.
-
Topology-Independent Loop-Free Alternates (TI-LFA)
IS-IS and OSPF can calculate LFA backup paths in a plain IP network. However, the LFA feature requires any backup path to be guaranteed loop-free. For this reason, in a plain IP network, LFA cannot offer backup paths to every single known destination. LFA only offers partial topology coverage. TI-LFA is a topology independent implementation of LFA. TI-LFA can push a Segment ID stack that can navigate around any potential loops along the backup path. In other words, backup paths can be calculated independent of the topology. It defines LFAs based on the traffic flow itself and can function effectively regardless of the underlying network layout. TI-LFA uses the pre-computed post-convergence path of the routing protocol. TI-LFA finds the path that would be calculated in the event of a particular link or node failure, and uses that exact path as the backup path. This is not always possible in regular LFA, due to the requirement for a loop-free backup path. By using the post-convergence path, TI-LFA reduces jitter during failover and the network operator only needs to ensure the network has enough capacity to carry the traffic on the post-convergence path after a failure. TI-LFA has multiple advantages:
-
Simplified configuration—TI-LFA automatically computes backup paths, eliminating the need to manually configurate LFAs for each network element.
-
Faster failover—TI-LFA uses the pre-computed post-convergence path of the routing protocol that enables it to activate backup paths significantly faster than traditional LFAs.
-
Improved scalability—TI-LFA scales efficiently in large and complex networks because it is topology independent in contrast to LDP and RSVP which require additional state to create backup paths.
TI-LFA Implementation in JCNR
An IGP identifies the primary and post-convergence (backup for TI-LFA) paths for a prefix based on its criterion. The paths are associated with a weight metric to signify priority (numerically lower the weight, higher the priority). The Cloud-Native Router control plane (cRPD) sends the primary and backup path to the data plane via the vRouter agent. The vRouter data plane implements FRR by identifying the primary path and quickly switching over to backup path if a link failure was detected.
Key points to note about the Cloud-Native Router TI-LFA implementation:
-
TI-LFA is supported for SR-MPLS on IS-IS implementations.
-
TI-LFA is supported for SR-MPLS on OSPF implementations (Juniper Technology Preview Feature).
-
TI-LFA is supported when Cloud-Native Router is deployed as head-end, transit or egress node in an SR-MPLS domain.
- Only TI-LFA protection against link failure is supported.
-
FRR is triggered based on link status events detected by the Poll Mode Drivers (PMD).
-
One primary and one backup path is supported.
-
Both physical and bond interfaces are supported (FRR is triggered for bond interfaces only if all the links in the bond are down).
TI-LFA Configuration (SR-MPLS on IS-IS)
Steps to configure TI-LFA
The following example requires that you navigate various levels in the configuration hierarchy. For information about navigating the CLI, see Using the CLI Editor in Configuration Mode in the Junos OS CLI User Guide.
TI-LFA Configuration (SR-MPLS on OSPF)
This is a Juniper Technology Previews (Tech Previews) Feature.
Steps to configure TI-LFA
The following example requires that you navigate various levels in the configuration hierarchy. For information about navigating the CLI, see Using the CLI Editor in Configuration Mode in the Junos OS CLI User Guide.