FAST REROUTE (FRR)
The secondary path provides the backup for primary path. By configuring the secondary path as standby, we make sure that the secondary path will pre-establish and ready to use, and shouldn't wait for primary path failure. However, as we already discuss this in our earlier post "MPLS Traffic Protection (Part 1)" the ingress router will switch over to secondary path only when it receives the ResvTear message from downstream routers. When an intermediate link or node fails the downstream node that detects the failure will send the message to ingress router along the LSP path. Upon receiving this message ingress router will clear the primary path and switch over the traffic to already establish secondary path.
By default, the fast reroute detours will inherit only the administrative groups (affinity) settings from the original LSP. This means the detour path must contain the affinity values that are define in LSP configuration however by default detour path will not consider the bandwidth requirements define in LSP settings therefor it is possible for a detour path to have subsequently less bandwidth than was specified in original LSP.
The following fast reroute parameters can be configured
Let's understand this with the help of following figure
Fast Reroute Configuration
root@PE1# show protocols mpls
label-switched-path pe1-to-pe2 {
from 192.168.100.1;
to 192.168.200.1;
fast-reroute;
primary primary-path {
bandwidth 300m;
}
secondary second-path {
bandwidth 50m;
standby;
}
}
path primary-path;
path second-path;
interface ge-0/0/0.0;
![]() |
| Figure-1: Primary LSP Failure and RESVTEAR Message |
Here
we have a link state path with primary and secondary paths establish.
Suppose the link between nodes P3 and P6 went down. The node P3 will
detect this failure and send RESVTEAR message to node P2 along
the LSP path. P2 will process the message and send it to P1, P1 also
process the message and send it to ingress router PE1. After PE1
receives the RESVTEAR message PE1 will clear the primary path of LSP and transfer the data to already establish secondary path.
Here comes a central problem that is the data packets will have lost
in between the time when a failure occurs and the time when ingress router will
notify about the failure. That time depends on number of hops the RESVTEAR message will
traverse.
The solution for this problem is to use the Fast Reroute for LSP. With Fast Reroute each router along the primary path of LSP will establish the detour paths.
These detour paths enable the LSP to keep delivering the traffic via
detour paths when a primary path link or node failure occurs. When a
path failure occurs the router that detect the failure will immediately
forwards the traffic for the LSP along the detour path. It will minimize
the traffic lost during primary path failure and intimation of that
failure to the ingress router. Note that Fast Reroute is a short-term
solution whereas the secondary path (with or without standby) is a
long-term solution.
When
we enable the fast reroute, ingress router will add an object in RSVP
message telling the downstream routers to establish the detour paths if
possible. When a primary path failure occurs and detour path is
available, the router that detect the failure immediately forwards the
traffic along the detour path, also generates the PATHERR message
and send that message to ingress router. This error message informs the
ingress router about a failure of primary path and that the detour is
being use to forward the traffic. If any downstream node fails to find
the detour path it sends the RESVTEAR message instead of PATHERR message to ingress node to intimate about the path failure. After
the ingress node receive PATHERR message and if secondary path is
already established the ingress router switch over the traffic on
secondary path. The ingress node will clear the primary path with its
detour paths and attempts to locate new primary path that match the
primary path constraints.
The same process occurs when secondary path is available but not pre-established (without standby), in that case ingress node will first signals and establish the secondary path, moves traffic to secondary path and then removes the primary path and its detour paths. If no secondary path is defined, the ingress router examines the TED to locate new primary path according to the defined primary path constraints. If ingress router finds the new primary path it creates new primary path for LSP and moves traffic on it and then clear the previous fail primary path and its detour paths. One more possibility is that there may be no secondary paths defined and CSPF algorithm might not able to find a new primary path that meets the path constraints. When this happens the fast reroute paths will stay up indefinitely and the ingress router continue to forward traffic using the detour paths.
The same process occurs when secondary path is available but not pre-established (without standby), in that case ingress node will first signals and establish the secondary path, moves traffic to secondary path and then removes the primary path and its detour paths. If no secondary path is defined, the ingress router examines the TED to locate new primary path according to the defined primary path constraints. If ingress router finds the new primary path it creates new primary path for LSP and moves traffic on it and then clear the previous fail primary path and its detour paths. One more possibility is that there may be no secondary paths defined and CSPF algorithm might not able to find a new primary path that meets the path constraints. When this happens the fast reroute paths will stay up indefinitely and the ingress router continue to forward traffic using the detour paths.
By default, the fast reroute detours will inherit only the administrative groups (affinity) settings from the original LSP. This means the detour path must contain the affinity values that are define in LSP configuration however by default detour path will not consider the bandwidth requirements define in LSP settings therefor it is possible for a detour path to have subsequently less bandwidth than was specified in original LSP.
The following fast reroute parameters can be configured
- Bandwidth
- Hop-limit
- Include and exclude administrative groups (enable by default)
Let's understand this with the help of following figure
![]() |
Figure-2: Example Network Diagram
|
Fast Reroute Configuration
root@PE1# show protocols mpls
label-switched-path pe1-to-pe2 {
from 192.168.100.1;
to 192.168.200.1;
fast-reroute;
primary primary-path {
bandwidth 300m;
}
secondary second-path {
bandwidth 50m;
standby;
}
}
path primary-path;
path second-path;
interface ge-0/0/0.0;
![]() |
| Figure-3: Detour Paths |
As we already discuss, the detour path will inherit only the affinity
values from original LSP. Apart from that the detour path will establish
with shortest CSPF metric value. Figure-3 shows
the detour paths for LSP pe1-to-pe2. Here the detour paths establish for P1, P2 and P3 routers. It Starts with the ingress router signals the fast reroute to all downstream nodes along primary path of LSP. Although not shown in above figure, since the secondary path is pre-establish using the standby option the detour paths also establish for the secondary path of LSP.
The node PE1 try to calculate and establish fast reroute path to node PE2 that avoids the P1 node and its links. The PE1 router can't bypass P1 router thus can't establish the detour path for PE1 as there is only a single path available between node PE1 and node P1. The sample output in Figure-4 we can see that Fast Reroute desire object is included in PATH message.
In Record Route flag = 9 is indicates that both the node and link protection available, and flag =1
indicates only link protection is available for the downstream
node. In Figure-4 the primary path of LSP 172.16.1.2 (P1), 172.16.11.2
(P2) has node protection and 172.16.13.1 (P3) has only link protection.
Same with secondary path of LSP 172.16.1.2 (P1), 172.16.21.2 (P4) has
node protection and 172.16.22.2 (P5) has only link protection. For both
primary (172.16.13.1) and secondary (172.16.23.1) the P6 node has no
flags means P6 has neither the node protection nor the link protection.
The sample output in Figure-5 Node P1 calculates and establish a detour
path around P2 to PE2. Node P1 successfully establish the detour path
and avoids P2 node and its links. The detour path is establish through P4(172.16.21.2),
P5(172.16.22.2), P6(172.16.23.1) to PE2 (172.16.2.1) with CSPF metric
value is 4. Detour path MPLS label is 299888.
The sample output in Figure-6 shows that a detour is established on node
P2, it avoids P3 node and P3 links. The detour path is established through P4
(172.16.31.2), P5 (172.16.22.2) and P6(172.16.23.1) to PE2 (172.16.2.1). The detour path MPLS label is 299920.
The node PE1 try to calculate and establish fast reroute path to node PE2 that avoids the P1 node and its links. The PE1 router can't bypass P1 router thus can't establish the detour path for PE1 as there is only a single path available between node PE1 and node P1. The sample output in Figure-4 we can see that Fast Reroute desire object is included in PATH message.
![]() |
| Figure-4: PE1 show mpls lsp extensive |
![]() |
| Figure-5: P1 show mpls lsp extensive |
![]() |
| Figure-6: P2 show mpls lsp extensive |
![]() |
| Figure-7: P3 show mpls lsp extensive |
![]() |
| Figure-8: P4 show mpls lsp extensive |
![]() |
| Figure-9: P5 show mpls lsp extensive |
Since P5 is use as transit router for P3 (merger of P1 and P2 detour) and P4 and the next-hop for these detours LSP are same that is node P6, therefor P5 can merge the detour receive from nodes P4 and P3, two detour paths together as indicated by the identical label out value 299840. This means that whether P5 receives traffic from P3 with a label value of 299824 or receive traffic from P2 with a label value of 299840, P4 will forward the packet to P5 with a label value 299840. The Label out value in the detour branch from 172.16.31.1 is 299840, the same value as label out value on node P4.
![]() |
| Figure-10: P6 show mpls lsp extensive |
~~~~ THE END ~~~~










No comments:
Post a Comment