BIER Working Group
Internet-Draft Y. Liu
Intended status: Standards Track China Mobile
Expires: December 27, 2024 C. Lin
New H3C Technologies
Zh. Zhang
ZTE Corporation
Y. Qiu
New H3C Technologies
June 28, 2024
BIER Loop Avoidance using Segment Routing
draft-liu-bier-uloop-04
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
Internet-Drafts are draft documents valid for a maximum of six
months and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet-Drafts as
reference material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html
This Internet-Draft will expire on December 27, 2024.
Copyright Notice
Copyright (c) 2024 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with
respect to this document. Code Components extracted from this
document must include Simplified BSD License text as described in
Liu, et al. Expires December, 2024 [Page 1]
Internet-Draft BIER loop Avoidance June 2024
Section 4.e of the Trust Legal Provisions and are provided without
warranty as described in the Simplified BSD License.
Abstract
This document provides a mechanism leveraging SR-MPLS/SRv6 to ensure
that BIER messages can be forwarded loop-freeness during the IGP
reconvergence process following a link-state change event.
Table of Contents
1. Introduction ................................................ 3
1.1. Requirements Language .................................. 4
2. Loop-free convergence process ............................... 5
3. Computing Loop-avoiding Path ................................ 5
3.1. Explicit Path of Loop-avoiding ......................... 5
3.2. Calculation Method of Explicit Path .................... 6
4. Example Application ......................................... 7
5. IANA Considerations ......................................... 9
6. Security Considerations ..................................... 9
7. References .................................................. 9
7.1. Normative References ................................... 9
7.2. Informative References ................................. 9
8. Acknowledgments ............................................. 9
Authors' Addresses ............................................ 10
liu, et al. Expires December, 2024 [Page 2]
Internet-Draft BIER loop Avoidance June 2024
1. Introduction
Forwarding loops happen during the convergence of the IGP, as a
result of transient inconsistency among forwarding states of the
nodes of the network.
When the network topology changes, loops may appear on new
forwarding paths due to the different convergence speeds of each
node's routing.
During the multicast packet forwarding process, when the upstream
BFR senses that its BFR-NBR is not reachable, the upstream BFR as a
PLR node can quickly switch multicast traffic to backup path through
the BIER FRR mechanism [I-D.ietf-bier-frr]. If the network fails to
recover, multicast traffic will switch back from the backup path to
the primary path.
As shown in Figure 1 below, R1 is connected to the multicast source,
and all IGP links are symmetric metric. Except for the link cost
between R7 and R8, which is 100, the cost of all other links is 1.
The multicast data packet sent from R1 to R9 is initially forwarded
along the path R1->R2->R3->R4->R9. When the link between R2 and R3
fails, node R3 fails, or the link failure between R2 and R3 is
restored, there may be a loop in packet forwarding between R2 and R7
during the routing convergence process.
SRC --- R1 ----- R2 ------ R3----- R4
| | | \
| | | \
| | | \
R5 ----- R6 ------ R7------R8----- R9----Receiver
100
Figure 1
Both of the following scenarios may result in a loop in the
forwarding of BIER messages.
* Scenario 1: When a link or node fails, a micro-loop may occur
during routing convergence.
Taking Figure 1 as an example, the fault process is as follows:
1) When the link between R2 and R3 fails or R3 fails, R2 will
detect that the route to R3 is unreachable. If BIER FRR is
enabled on R2, based on BIER FRR, R2 will first choose to send
messages to backup neighbor R6. If R2 does not enable BIER FRR,
R2 will lose packets. At this stage, FRR can enhance network
reliability.
liu, et al. Expires December, 2024 [Page 3]
Internet-Draft BIER loop Avoidance June 2024
2) R2 floods topology change event through IGP. After receiving
IGP messages, each network node recalculates the SPF tree.
3) After calculating the new unicast SPF tree, issue the new
unicast forwarding table entries and update the BIFTs of
multicast.
Normally, after R2 completes the routing convergence to R9, the
BFR NBR from R2 to R9 becomes R6. When R2 receives a BIER message
sent to R9, R2 searches for BIFT and forwards the message to R6.
If the convergence speed of R6's routing is slower than R2, when
R2 has completed convergence, R6 is not yet complete. During the
convergence process of R6, the BFR BNR recorded in BIFT on R6 to
R9 will still be R2. Therefore, after receiving the BIER message,
R6 will send the message back to R2 according to the Bitstring in
the BIER header. An instant micro-loop between R2 and R6 appears.
* Scenario 2: After fault recovery, a micro-loop may also occur
during routing convergence.
In Figure 1, if the link between R2 and R3 fails, after the
entire network routing converges, the BIER packet forwarding path
from R1 to R9 becomes R1->R2->R6->R7->R8->R9.
When the link failure between R2 and R3 is recovered, during the
routing convergence process, if the routing of R6 converges
faster than R2, before R2 completes the routing convergence,
because R2 still records the BFR NBR to R9 as R6, when R2
receives the BIER message sent to R9, it will still forward the
message to R6. Resulting in a short period of micro-loop.
This document provides a mechanism leveraging SR-MPLS/SRv6 to ensure
that BIER messages can be forwarded loop-freeness during the IGP
reconvergence process following a topology change event.
1.1. Requirements Language
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
"OPTIONAL" in this document are to be interpreted as described in
BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
capitals, as shown here.
liu, et al. Expires December, 2024 [Page 4]
Internet-Draft BIER loop Avoidance June 2024
2. Loop-free convergence process
Upon a topology change, when a BFR converging for BFERs does not
trust the loop-freeness of its post-convergence paths for BFERs, it
performs convergence processing as follows.
After computing the new path to BFER, for a predetermined amount of
time C (Corresponding to an avoiding uloop timer), BFR installs an
explicit path through which packets can be steered to BFER without
loop. For example, forwarding through an explicit SR-MPLS/SRv6 path
or through a P2MP path. C should be greater than or equal to the
worst-case convergence time of a node, network-wide. The
determination of "C" is outside the scope of this document. The
forwarding path is computed when the event occurs.
After C elapses, BFR installs the normal post-convergence forwarding
entry for BFER that ensure the loop-free property.
Taking R1 in Figure 1 as an example, When R1 senses a topology
change event between R2 and R3 through unicast routing, it
recalculates the forwarding path of multicast packets for the
affected nodes. Within the interval C, R1 specifies an explicit path
to send multicast packets to R9 along the link between R7-R8. Before
the routing convergence is completed, the multicast traffic is
forwarded along the path R1->R6->R7->R8->R9.
3. Computing Loop-avoiding Path
When the link between R2 and R3 fails, unicast routing converges,
and all BFRs affected by topology change recalculate the SPF tree.
Before the routing of all nodes completes convergence, BFR steers
the packet targeting R9 to an explicit path, forwarding it based on
the specified node label or SRv6 SID in the explicit path.
When the link failure between R2 and R3 is restored, the process of
avoiding micro-loop is also similar.
3.1. Explicit Path of Loop-avoiding
The explicit path of loop-avoiding can be (but not limited to):
* SR policy TE path. Treat each node or adjacent SID on the explicit
path as a segment on the SR Policy TE path. During the convergence
process, add SRv6 encapsulation to the BIER message, specify the
SRH Segment List, and send it to the endpoint of the explicit
path. After reaching the endpoint, decapsulate the outer layer SR-
MPLS/SRv6 packet header, restore the original BIER packet, and
continue forwarding according to the BIER header.
liu, et al. Expires December, 2024 [Page 5]
Internet-Draft BIER loop Avoidance June 2024
* BIER-TE forwarding path. During the convergence process, each node
on the explicit path is treated as a BIER-TE node and forwarded
through BIER-TE. If the explicit paths to different BFERs require
passing through some identical replication nodes, the BFR nodes on
these explicit paths can be arranged as BIER-TE forwarding paths.
During routing convergence, BIER packets are first forwarded along
the BIER-TE path to the endpoint of the avoiding micro-loop path,
then unpacked and the inner layer BIER packets are restored, and
finally forwarded according to the BIER header.
The bit position of nodes on the BIER-TE path can be arranged into
the bitstring of the original BIER header, or the BIER-TE header
can be encapsulated outside the BIER message.
* P2MP policy forwarding path. If the explicit paths to different
BFERs require passing through some identical replication nodes,
the BFR nodes on these explicit paths can be arranged as P2MP
policy forwarding paths. During the convergence process, multicast
messages are forwarded through the path specified by the P2MP
policy.
This document focuses on using the SR-MPLS and SRv6 TE path as the
path for multicast avoidance loop.
3.2. Calculation Method of Explicit Path
There are currently two methods to calculate the nodes included in
the explicit path.
Method 1: Similar to [RFC7490], using the concept of P-Space and Q-
Space for TI-LFA generate explicit SR/SRv6-based path from P to Q.
The repair list is expressed generally as {P node (NODE SID), all
ADJ/End.X SIDs from P node to Q node}.
1) Using BFER as the destination node, BFR calculates the optimal
convergence path tree. That is, when the topology changes,
unicast routing converges, and BFR calculates a new SPF tree.
2) Find the Q node. On the new SPF tree, traverse the parent nodes
starting from the BFER destination node until finding the
farthest node from BFER, which is not affected by the link
failure and can reach BFER, as the Q node.
3) Find the P node. On the new SPF tree, traverse the parent nodes
starting from the Q node until finding a node that is not
affected by the topology change on the path from current BFR to
that node. This node will be considered as the P node.
liu, et al. Expires December, 2024 [Page 6]
Internet-Draft BIER loop Avoidance June 2024
4) Calculate the repair segment list path. The repair segment list
path is found by computing the explicit SR/SRv6-based path from
P to Q when these nodes are not adjacent along the convergence
path.
* For SR-MPLS, the repair list is expressed {Node_SID(P),
AdjSID(P->Q)}.
* For SRv6, the repair list is expressed {END_SID(P), END.X
(P->Q)}.
5) After BFR completes routing convergence locally, start an
avoiding uloop timer.
6) Before the timeout of the avoiding uloop timer, if BFR receives
the BIER message, it sends the BIER message to the endpoint
along the explicit path indicated by the repair list.
7) After the endpoint receives the message, remove the explicit
path encapsulation, restore the BIER message, and then continue
forwarding according to the BIER header.
Method 2: Directly generate a strict explicit path from current BFR
to Q node.
* For SR-MPLS, the repair list is expressed {AdjSID(S->Q)}.
* For SRv6, the repair list is expressed {END.X(S->Q)}.
If the P nodes and Q nodes of different BFERs are the same, which
means that multicast packets can be forwarded through the same path,
it is necessary to merge the multicast forwarding paths to avoid
headend replication. Try to place the multicast replication point on
the node closest to the multicast receivers.
4. Example Application
SRC --- R1 ----- R2 ------ R3----- R4----- R10----Receiver1
| | | \
| | | \
| | | \
R5 ----- R6 ------ R7------R8----- R9----Receiver2
100
Figure 2
As an example, in Figure 2, R1 is connected to the multicast source,
R9 and R10 are BFERs. All IGP links are symmetric metric. Except for
the link cost between R7 and R8, the cost of all other links is 1.
liu, et al. Expires December, 2024 [Page 7]
Internet-Draft BIER loop Avoidance June 2024
The multicast data packet sent from R1 to R9 and R10 is initially
forwarded along the path R1->R2->R3->R4->R9/R10. When the link
between R2 and R3 fails, the forwarding path of BIER messages from
R1 to R9/R10 becomes R1->R2->R6->R7->R8->R9/R10 after the
convergence of the entire network routing is completed.
When the link between R2 and R3 recovers, R2 will perceive the
topology change and recalculate the SPF tree, and inform IGP
neighbors of the topology change event.
All BFRs that receive topology change event need to recalculate the
SPF tree. To avoid micro loop during routing convergence, these BFRs
also need to calculate the repair path and repair list for each
BFER.
Taking BFIR node R1 as an example, the process is as follows:
1) When R1 receives a topology change event between R2 and R3, R1
calculates the new SPF tree and the repair path for each BFER.
2) R1 calculates the Q node for R9, and the result is [R8].
3) R1 calculates the P node for R9, and the result is [R7].
4) The path between P node and Q node is the repair path of R9.
5) R1 repeats the above process to calculate the repair path for R10.
The P and Q nodes are also R7 and R8.
BFERs with the same P and Q nodes use the same repair path. Because
R9 and R10 have the same repair path, in the avoiding uloop
interval, the BIER message is first forwarded to the endpoint of the
repair path, and then is replicated at the endpoint.
To R9 and R10, both need to go through the R7->R8 link.
* For SR-MPLS, the repair list of R1 for R9 and R10 considering the
fault recovery of link between R2 and R3 or of node R3 is:
<NodeSID(R7), AdjSID_R7R8>.
* For SRv6, the repair list of R1 for R9 and R10 considering the
fault recovery of link between R2 and R3 or of node R3 is:
<End.X_R7R8>.
When the link state between R2 and R3 changes to UP, R1 converges
routing and sends the BIER message to R8 along the SR-MPLS/SRv6 path
indicated by the repair list in the avoiding uloop interval.
liu, et al. Expires December, 2024 [Page 8]
Internet-Draft BIER loop Avoidance June 2024
After R8 receives the message, remove the outer SR-MPLS/SRv6
encapsulation, restore the BIER message, and then continue forwarding
according to the BIER header.
5. IANA Considerations
No requirements for IANA.
6. Security Considerations
The behavior described in this document is internal functionality to
a router that result in the ability to explicitly steer traffic over
the post convergence path after a remote topology change in a manner
that guarantees loop freeness. Because the behavior serves to
minimize the disruption associated with a topology change, it can be
seen as a modest security enhancement.
7. References
7.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, DOI
10.17487/RFC2119, March 1997, <https://www.rfc-
editor.org/info/rfc2119>.
[RFC7490] Bryant, S., Filsfils, C., Previdi, S., Shand, M., So, N.,
"Remote Loop-Free Alternate (LFA) Fast Reroute (FRR)", BCP
14, RFC 8174, DOI 10.17487/RFC7490, April 2015,
<https://www.rfc-editor.org/info/rfc7490>.
[RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
May 2017, <https://www.rfc-editor.org/info/rfc8174>.
7.2. Informative References
TBD
8. Acknowledgments
The authors would like to thank the following for their valuable
contributions of this document:
TBD
liu, et al. Expires December, 2024 [Page 9]
Internet-Draft BIER loop Avoidance June 2024
Authors' Addresses
Yisong Liu
China Mobile
Email: liuyisong@chinamobile.com
Changwang Lin
New H3C Technologies
Email: linchangwang.04414@h3c.com
Zheng Zhang
ZTE Corporation
Email: zhang.zheng@zte.com.cn
Yuanxiang Qiu
New H3C Technologies
Email: qiuyuanxiang@h3c.com
liu, et al. Expires December, 2024 [Page 10]