SR-MPLS FRR ExtensionFutureweiBoston, MAUSAHuaimo.chen@futurewei.comHuawei TechnologiesHuawei Bld., No.156 Beiqing Rd.Beijing100095Chinahuzhibo@huawei.comChina TelecomBeiqijia Town, Changping DistrictBeijing102209Chinawangaj3@chinatelecom.cnChina Mobile510000Chinaliuyisong@chinamobile.comVerizon Inc.13101 Columbia PikeSilver SpringMD 20904USA301 502-1347gyan.s.mishra@verizon.com
The current SR FRR such as TI-LFA provides fast re-route protection
for the failure of a node on an SR-MPLS path by the neighbor
upstream node as point of local repair (PLR) of the failed node.
However, once the IGP converges,
the SR FRR is no longer sufficient to forward traffic of the path
around the failure, since the non-neighbor upstream node
of the failed node will no longer have a route to the failed node.
This document describes a simple mechanism to extend the
fast re-route protection
for the failure on an SR-MPLS path after the IGP converges.
The mechanism protects the node SID, adjacency SID and
binding SID of the failed node on the path.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in
when, and only when, they appear in all capitals, as shown here. describes
an SR FRR mechanism that provides fast re-route protection for
the failure of a node on an SR-TE path by the neighbor
upstream node as point of local repair (PLR) of the failed node.
However, once the IGP converges,
the SR FRR is no longer sufficient to forward traffic of the path
around the failure, since the non-neighbor upstream node
of the failed node will no longer have a route to the failed
node and drop the traffic.
proposes a solution in which
a hold-down timer is configured on every node in a network.
After the IGP converges on a node failure,
when a node is going to delete the route to the failed node,
instead of programming a route delete,
it programs a tunnel/path to the node
consisting of the Node SID of the
nearside neighbor of the failed node
followed by the original path in the packet.
The modified path
will be in force until the hold-down timer expires.
This document describes
a simple mechanism to extend the fast re-route protection
for the failure on an SR-MPLS path after the IGP converges.
The mechanism protects the node SID, adjacency SID and
binding SID of the failed node on the path.This section illustrates the extension to SR-MPLS FRR
for the failure on SR-MPLS paths after the IGP converges
through examples.
It shows the procedure on every related node on each path
without any failure, with a failure
before and after the IGP converges on the failure. shows an example topology
with two SR-MPLS paths: Path 1 and Path 2 for explaining the
extension. They go through the same nodes and links, but
represented differently. The former does not have any
binding SID (BSID).
The latter has a BSID of node N (BSID-N).
P1->N->Q1->C, indicated at A by {SID-P1,SID-N,SID-Q1,SID-C}
Path 2:A->P1->N->Q1->C, indicated at A by {SID-P1,SID-N,BSID-N} and
BSID-N associated with SID list {SID-Q1,SID-C}
]]>
SR-MPLS Path 1: A->P1->N->Q1->C
is indicated at node A by node SID of P1 (SID-P1), node SID of N (SID-N),
node SID of Q1 (SID-Q1) and node SID of C (SID-C).
SR-MPLS Path 2
is indicated at node A by {SID-P1,SID-N,BSID-N},
where BSID-N (binding SID of N)
is associated with SID list {SID-Q1,SID-C}.
shows the
result of executing procedure on each related node
of SR-MPLS path 1 without any failure.
P1->N->Q1->C, indicated at A by {SID-P1,SID-N,SID-Q1,SID-C}
]]>
The results from nodes A, P1, P3, N, Q1 and C
are as follows.
Node A as the ingress of the path adds
SID-P1, SID-N, SID-Q1, and SID-C
into a packet (Pkt) received from CE1 and sends the packet
with the SIDs to node P1.
The packet sent to P1 is represented by
"1.{SID-P1,SID-N,SID-Q1,SID-C}Pkt".
Node P1 pops its SID-P1 from the packet received,
sends the packet with top SID (SID-N) to P3
along the IGP shortest path to N
according to its FIB entry for SID-N.
The packet sent to P3 is represented by
"2.{SID-N,SID-Q1,SID-C}Pkt".
Node P3 sends the packet with top SID (SID-N) to N
along the IGP shortest path to N.
The packet sent to N is represented by
"3.{SID-N,SID-Q1,SID-C}Pkt".
Node N pops its SID-N from the packet received,
sends the packet with top SID (SID-Q1) to Q1
along the IGP shortest path to Q1
according to its FIB entry for SID-Q1.
The packet sent to Q1 is represented by
"4.{SID-Q1,SID-C}Pkt".
Node Q1 pops its SID-Q1 from the packet received,
sends the packet with top SID (SID-C) to C
along the IGP shortest path to C.
The packet sent to C is represented by
"5.{SID-C}Pkt".
Node C pops its SID-C and gets the packet without any SIDs,
which is represented by
"6.Pkt".
shows the
result of executing procedure on each related node
of SR-MPLS path 1 with no BSID
when node N failed and
before the IGP converges on the failure.
P1->N->Q1->C, indicated at A by {SID-P1,SID-N,SID-Q1,SID-C}
]]>
The results from nodes A, P1, Q1 and C are the same as
those described in .
The results from neighbor upstream node P3 of N and node N1
are as follows.
After detecting the failure of N,
the neighbor upstream node P3 of N pops SID-N
from the packet received,
re-routes the packet to node Q1 via node N1 without
going through failed N.
The packet sent to N1 is represented by
"3.{SID-Q1,SID-C}Pkt".
Node N1
sends the packet to Q1 according to the top SID (SID-Q1)
in the packet.
The packet sent to Q1 is represented by
"4.{SID-Q1,SID-C}Pkt".
shows the
result of executing procedure on each related node
on SR-MPLS path 1 with no BSID
when node N failed and
after the IGP converges on the failure.
P1->N->Q1->C, indicated at A by {SID-P1,SID-N,SID-Q1,SID-C}
]]>
The results from nodes A, N1, Q1 and C are the same as
those described above in .
The results from nodes P1 and P4
are as follows.
Since non-neighbor upstream
node P1 of N on
the SR path does not have a FIB entry for SID-N as the top
SID of the packet, node P1
pops SID-N from the packet, and
sends the packet to P4 according to the top SID (SID-Q1)
in the packet
along the IGP shortest path to Q1.
The packet sent to P4 is represented by
"2.{SID-Q1,SID-C}Pkt".
Node P4 sends the packet to N1 according to
the top SID (SID-Q1) in the packet received.
The packet sent to N1 is represented by
"3.{SID-Q1,SID-C}Pkt".
shows the
result of executing procedure on each related node
of SR-MPLS path 2 with BSID-N without any failure.
P1->N->Q1->C, indicated at A by {SID-P1,SID-N,BSID-N} and
BSID-N associated with SID list {SID-Q1,SID-C}
]]>
The results from nodes A, P1, P3, and N
are as follows.
Node A as the ingress of the path adds
SID-P1, SID-N, and BSID-N
into a packet (Pkt) received from CE1 and sends the packet
with the SIDs to node P1.
The packet sent to P1 is represented by
"1.{SID-P1,SID-N,BSID-N}Pkt".
Node P1 pops its SID-P1 from the packet received,
sends the packet with top SID (SID-N) to P3
along the IGP shortest path to N
according to its FIB entry for SID-N.
The packet sent to P3 is represented by
"2.{SID-N,BSID-N}Pkt".
Node P3 sends the packet with top SID (SID-N) to N
along the IGP shortest path to N.
The packet sent to N is represented by
"3.{SID-N,BSID-N}Pkt".
Node N pops its SID-N from the packet received,
replaces its BSID-N with SID list {SID-Q1,SID-C}, and
sends the packet with top SID (SID-Q1) to Q1
along the IGP shortest path to Q1
according to its FIB entry for SID-Q1.
The packet sent to Q1 is represented by
"4.{SID-Q1,SID-C}Pkt".
The results from nodes Q1 and C are the same as
those described in .
shows the
result of executing procedure on each related node
of SR-MPLS path 2 with BSID-N
when node N failed and
before the IGP converges on the failure.
P1->N->Q1->C, indicated at A by {SID-P1,SID-N,BSID-N} and
BSID-N associated with SID list {SID-Q1,SID-C}
]]>
The results from nodes A, P1, Q1 and C are the same as
those described in .
The results from neighbor upstream node P3 of N and node N1
are as follows.
After detecting the failure of N,
the neighbor upstream node P3 of N pops SID-N
from the packet received,
replaces BSID-N in the packet with SID list {SID-Q1,SID-C},
and re-routes the packet to node Q1 via node N1 without
going through failed N.
The packet sent to N1 is represented by
"3.{SID-Q1,SID-C}Pkt".
Node N1
sends the packet to Q1 according to the top SID (SID-Q1)
in the packet.
The packet sent to Q1 is represented by
"4.{SID-Q1,SID-C}Pkt".
shows the
result of executing procedure on each related node
on SR-MPLS path 2 with BSID-N
when node N failed and
after the IGP converges on the failure.
P1->N->Q1->C, indicated at A by {SID-P1,SID-N,BSID-N} and
BSID-N associated with SID list {SID-Q1,SID-C}
]]>
The results from nodes A, Q1 and C are the same as
those described above in .
The results from nodes P4 and N1 are the same as
those described above in .
The result from node P1
is as follows.
Since non-neighbor upstream
node P1 of N on
the SR path does not have a FIB entry for SID-N as the top
SID of the packet, node P1
pops SID-N from the packet,
replaces BSID-N in the packet with SID list {SID-Q1,SID-C}
and
sends the packet to P4 according to the top SID (SID-Q1)
in the packet
along the IGP shortest path to Q1.
The packet sent to P4 is represented by
"2.{SID-Q1,SID-C}Pkt".
This section presents the procedures on a neighbor upstream node and
a non-neighbor upstream node of
node N on an SR MPLS path.For an SR-MPLS path with the node SID of node N (SID-N),
suppose that node X is an non-neighbor upstream node of node N
along the path, wherein SID-N follows
the node SID of node X (SID-X)
or the adjacency SID to node X in the packet
to be transported by the path.
For example, SR-MPLS Path 1 and Path 2
in
are from A to P1 to N to Q1 to C.
Node P1 is the non-neighbor upstream node of N.
At ingress node A of the path,
node A adds the SIDs in the SID list
into the packet to be transported by the path.
Without any failure
or from the failure of node N to IGP convergence on the failure,
the non-neighbor upstream node (such as P1) of node N pops
its SID from the packet if any and
sends the packet to the next hop node toward node N
along the IGP shortest path to N.
After node N failed and from the IGP convergence on the failure
to global reroute,
the non-neighbor upstream node (such as P1) of node N
pops its SID (such as SID-P1) from the packet if any,
pops SID-N from the packet
and does one of the following:
If the current top SID in the
packet is a node SID of a node named Nx,
sends the packet toward Nx along the IGP shortest
path to Nx.
If the current top SID in the
packet is an adjacency SID of node N,
obtains the remote node of the adjacency from the adjacency SID,
replaces the adjacency SID with the node SID of the remote node,
and sends the packet toward the remote node along the IGP
shortest path to the remote node.
If the current top SID in the
packet is a Binding SID (BSID) of node N,
replaces the BSID in the packet with the SID list
associated with the BSID, and
does a. or b.
according to the current top SID in the packet
(i.e., does a. if it is a node SID;
does b. if it is an adjacency SID of N).
Note: Distributing the information about the BSID of N
(including the BSID, the SID list, the ID of N)
to upstream nodes of N
is out of scope of this document.
Suppose that node Y is the neighbor upstream node of node N
on an SR-MPLS path.
Without any failure,
node Y sends the packet received from the path
to node N according to the top SID in the packet.
For example, SR-MPLS Path 1 and Path 2
in
are from A to P1 to N to Q1 to C.
Node P3 is the neighbor upstream node of N
suppose that the shortest path from P1 to N is from P1 to P3 to N.
When node Y detects the failure of node N,
node Y pops SID-N from the packet
if SID-N is the top SID of the packet,
and does one of the following:
If the current top SID in the packet is node SID of node Nx,
sends the packet toward node Nx without going through failed N
(i.e., {SID-Nx,...} in the packet would be {RL(Q),SID-Nx,...},
where RL(Q) is the repair list redirecting the packet to node Q,
whose path to SID-Nx is not affected by the failure).
If the current top SID in the packet is an adjacency SID of N,
gets remote node R of the adjacency from the adjacency SID,
replaces the top SID in the packet with node SID of R, and
sends the packet to node R according to the top SID
in the packet
without going through failed N
(i.e., {adj(N-R),...} in the
packet would be {RL(Q),SID-R,...}, where RL(Q) is the repair list
redirecting the packet to node Q, whose path to SID-R is not
affected by the failure).
If the current top SID in the packet is a BSID of N,
replaces the BSID with the SID list associated with the BSID,
and does a. or b.
according to the current top SID in the packet
(i.e., does a. if it is a node SID;
does b. if it is an adjacency SID of N).
TBD.No requirements for IANA.AcknowledgmentsThe authors would like to thank people
for their comments to this work.