| Internet-Draft | SR Replication Segment | July 2023 |
| Voyer, Ed., et al. | Expires 1 February 2024 | [Page] |
This document describes the Segment Routing Replication segment for Multi-point service delivery. A Replication segment allows a packet to be replicated from a Replication node to Downstream nodes.¶
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.¶
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). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
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."¶
This Internet-Draft will expire on 1 February 2024.¶
Copyright (c) 2023 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 (https://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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
Replication segment is a new type of segment for Segment Routing (SR) [RFC8402], which allows a node (henceforth called a Replication node) to replicate packets to a set of other nodes (called Downstream nodes) in a Segment Routing Domain. Replication segments provide building blocks for Point-to-Multipoint service (P2MP) delivery via SR Point-to-Multipoint (SR P2MP) policy. A Replication segment can replicate packets to directly connected nodes or to downstream nodes (without need for state on the transit routers). This document focuses on the Replication segment building block for both Segment Routing with Multiprotocol Label Switching (SR-MPLS) [RFC8660] and Segment Routing with IPv6 (SRv6) [RFC8986]. The examples in the Appendix illustrate the behavior of Replication Segment in SR domain. The use of one or more stitched Replication segments constructed for SR P2MP Policy tree is specified in [I-D.ietf-pim-sr-p2mp-policy]. This document focuses on specifying replication behavior in an SR domain. The management of IP multicast groups, building IP multicast trees, and performing multicast congestion control are out of scope of this document.¶
This section defines terms introduced and used frequently in this document. Refer to Terminology sections of [RFC8402], [RFC8754] and [RFC8986] for other terms used in Segment Routing.¶
In the simplest use case, a single Replication segment includes the ingress (Root) node of a multi-point service and the egress (Leaf) nodes of the service as all the Downstream nodes. This achieves Ingress Replication [RFC7988] that has been widely used for Multicast VPN (MVPN) [RFC6513] and Ethernet VPN (EVPN)[RFC7432] BUM (Broadcast, Unknown and Multicast) traffic.¶
Replication segments can also be used as building blocks for replication trees when Replication segments on the Root, intermediate Replication nodes and Leaf nodes are stitched together to achieve efficient replication. That is specified in [I-D.ietf-pim-sr-p2mp-policy].¶
In a Segment Routing Domain, a Replication segment is a logical construct which connects a Replication node to a set of Downstream nodes. A Replication segment is a local segment instantiated at a Replication node. It can be either provisioned locally on a node or programmed by a Path Computation Element (PCE).¶
A Replication segment is identified by the tuple <Replication-ID, Node-ID>, where:¶
Replication-ID is a variable length field. In simplest case, it can be a 32-bit number, but it can be extended or modified as required based on specific use of a Replication segment. This is out of scope for this document. The length of Replication-ID is specified in the signaling mechanism used for Replication segment. Examples of such signaling and extensions are described in [I-D.ietf-pim-sr-p2mp-policy]. When the PCE signals a Replication segment to its node, the <Replication-ID, Node-ID> tuple identifies the segment.¶
A Replication segment includes the following elements:¶
The Downstream nodes and Replication state of a Replication segment can change over time, depending on the network state and Leaf nodes of a multi-point service that the segment is part of.¶
Replication SID identifies the Replication segment in the forwarding plane. At a Replication node, the Replication SID operates on Replication state of the Replication segment.¶
Replication state is a list of replication branches to the Downstream nodes. In this document, each branch is abstracted to a <Downstream node, Downstream Replication SID> tuple. <Downstream node> represents the reachability from the Replication node to the Downstream node. In its simplest form, this MAY be specified as an interface or next-hop if downstream node is adjacent to the Replication node. The reachability may be specified in terms of Flexible Algorithm path (including the default algorithm) [RFC9350], or specified by an SR explicit path represented either by a SID-list (of one or more SIDs) or by a Segment Routing Policy [RFC9256]. Downstream Replication SID is the Replication SID of the Replication segment at the Downstream node.¶
A packet is steered into a Replication segment at a Replication node in two ways:¶
In either case, the packet is replicated to each Downstream node in the associated Replication state.¶
If a Downstream node is an egress (Leaf) of the multi-point service, i.e. no further replication is needed, then that Leaf node's Replication segment will not have any Replication state i.e. the list of Replication branches is empty. The Replication segment will have an indicator role of the node is Leaf. The operation performed on incoming Replication SID is NEXT https://www.rfc-editor.org/rfc/rfc8402#section-2. At an egress node, the Replication SID MAY be used to identify that portion of the multi-point service. Notice that the segment on the Leaf node is still referred to as a Replication segment for the purpose of generalization.¶
A node can be a Bud node, i.e. it is a Replication node and a Leaf node of a multi-point service [I-D.ietf-pim-sr-p2mp-policy]. Replication segment of a Bud node has a list of Replication Branches as well as Leaf role indicator.¶
In principle it is possible for different Replication segments to replicate packets to the same Replication segment on a Downstream node. However, such usage is intentionally left out of scope of this document.¶
When the Active Segment is a Replication SID, the processing results in a POP https://www.rfc-editor.org/rfc/rfc8402#section-2 operation and lookup of the associated Replication state. For each replication in the Replication state, the operation is a PUSH https://www.rfc-editor.org/rfc/rfc8402#section-2 of the downstream Replication SID and an optional segment list on to the packet to steer the packet to the Downstream node.¶
For Leaf/Bud nodes local delivery off tree is per local configuration. For some usages, this may involve looking at the next SID for example to get the necessary context.¶
When the Root of a multi-point service steers a packet to a Replication segment, it results in a replication to each Downstream node in the associated replication state. The operation is a PUSH of the replication SID and an optional segment list on to the packet which is forwarded to the downstream node.¶
The following applies to Replication SID in MPLS encapsulation:¶
For SRv6 [RFC8986], this document specifies “Endpoint with replication” behavior (End.Replicate for short) to replicate a packet and forward the replicas according to a Replication state.¶
When processing a packet destined to a local Replication SID, the packet is replicated according to the associated Replication state to Downstream nodes and/or locally delivered off tree when this is a Leaf/Bud node. IPv6 Hop Limit MUST be decremented and MUST be non-zero to replicate an incoming packet. For replication, the outer header is re-used, and the Downstream Replication SID, from Replication state, is written into the outer IPv6 header destination address. If required, an optional segment list may be used on some branches using H.Encaps.Red https://www.rfc-editor.org/rfc/rfc8986.html#name-hencapsred-hencaps-with-red (while some other branches may not need that). Note that this H.Encaps.Red is independent from the replication segment – it is just used to steer the replicated packet on a traffic engineered path to a Downstream node. The penultimate segment in encapsulating IPv6 header will execute Ultimate Segment Decapsulation (USD) flavor [RFC8986] of End/End.X behavior and forward the inner (replicated) packet to the Downstream node. If H.Encaps.Red is used to steer a replicated packet to a Downstream node, the operator must ensure the MTU on path to the Downstream node is sufficient to account for additional SRv6 encapsulation.¶
The above also applies when the Replication segment is for the Root node, whose upstream node has placed the Replication-SID in the header. A local application, for e.g. Multicast VPN (MVPN) [RFC6513] or Ethernet VPN (EVPN) [RFC7432], may also apply H.Encaps.Red and then steer the resulting traffic into the segment. Again, note that the H.Encaps.Red is independent of the Replication segment – it is the action of the application (e.g. MVPN/EVPN service). If the service is on a Root node, the two H.Encaps mentioned, one for the service and other in the previous paragraph for replication to Downstream node SHOULD be combined for optimization (to avoid extra IPv6 encapsulation).¶
For Leaf/Bud nodes local delivery off the tree is per Replication SID or next SID (if present in SRH). For some usages, this may involve getting the necessary context either from the next SID (e.g., MVPN with shared tree) or from the replication SID itself (e.g., MVPN with non-shared tree). In both cases, the context association is achieved with signaling and is out of scope of this document.¶
The following applies to Replication SID in SRv6 encapsulation:¶
The "Endpoint with replication and/or decapsulate behavior (End.Replicate for short) is variant of End behavior. The pseudo-code in this section follows the convention introduced in RFC 8986 [RFC8986].¶
A Replication state conceptually contains the following elements:¶
Replication state:
{
Node-Role: {Head, Transit, Leaf, Bud};
# On Leaf, replication list is zero length
Replication-List:
{
Downstream node: <Node-Identifier>;
Downstream Replication SID: R-SID;
# Segment-List may be empty
Segment-List: [SID-1, .... SID-N];
}
}
¶
Below is the Replicate function on a packet for Replication state (RS).¶
S01. Replicate(RS, packet)
S02. {
S03. For each Replication R in RS.Replication-List {
S04. Make a copy of the packet
S05. Set IPv6 DA = RS.R-SID
S06. If RS.Segment-List is not empty {
S07. # Head node may optimize below encapsulation and
S08. # the encapsulation of packet in a single encapsulation
S09. Execute H.Encaps or H.Encaps.Red with RS.Segment-List
on packet copy #RFC 8986 Section 5.1, 5.2
S10. }
S11. Submit the packet to the egress IPv6 FIB lookup and
transmission to the new destination
S12. }
S13. }
¶
Notes:¶
When N receives a packet whose IPv6 DA is S and S is a local End.Replicate SID, N does:¶
S01. Lookup FUNCT portion of S to get Replication state RS
S02. If (IPv6 Hop Limit <= 1) {
S03. Discard the packet
S04. # ICMPv6 Time Exceeded is not permitted (Section 2.2.3 below)
S05. }
S06. If RS is not found {
S07. Discard the packet
S08. }
S09. Decrement IPv6 Hop Limit by 1
S10. If (IPv6 NH == SRH and SRH TLVs present) {
S11. Process SRH TLVs if allowed by local configuration
S12. }
S13. Call Replicate(RS, packet)
S14. If (RS.Node-Role == Leaf OR RS.Node-Role == Bud) {
S15. If (IPv6 NH == SRH and Segments Left > 0) {
S16. Derive packet processing context(PPC) from Segment List
S17. If (Segments Left != 0) {
S18. Discard the packet
S19. # ICMPv6 Parameter Problem with Code 0
S20. # (Erroneous header field encountered)
S21. # is not permitted (Section 2.2.3 below)
S22. }
S23. } Else {
S24. Derive packet processing context(PPC)
from FUNCT of Replication SID
S25. }
S26. Process the next header
S27. }¶
The processing of Upper-Layer header of a packet matching End.Replicate SID at Leaf/Bud node is as follows:¶
S01. If (Upper-Layer header type == 4(IPv4) OR
Upper-Layer header type == 4(IPv6) ) {
S02. Remove the outer IPv6 header with all its extension headers
S03. Process the packet in context of PPC
S04. } Else If (Upper-Layer header type == 143(Ethernet) ) {
S05. Remove the outer IPv6 header with all its extension headers
S06. Process the Ethernet Frame in context of PPC
S07. } Else If (Upper-Layer header type is allowed
by local configuration) {
S08. Proceed to process the Upper-Layer header
S09. } Else {
S10. Discard the packet
S11. # ICMPv6 Parameter Problem with Code 4
S12. # (SR Upper-layer Header Error)
S13. # is not permitted (Section 2.2.3 below)
S14. }¶
Notes:¶
Processing the Replication SID may modify, if configured to process TLVs, the "variable-length data" of TLV types that change en route. Therefore, TLVs that change en route are mutable. The remainder of the SRH (Segments Left, Flags, Tag, Segment List, and TLVs that do not change en route) are immutable while processing this SID.¶
If a Root node encodes a context SID in SRH with an optional HMAC SRH TLV [RFC8754], it MUST set the 'D' bit as defined in Section 2.1.2 because the Replication SID is not part of the segment list in SRH.¶
HMAC generation and verification is as specified in Section 2.1.2.1 of RFC 8754. Verification of HMAC TLV is determined by local configuration. If verification fails, an implementation of Replication SID MUST NOT originate an ICMPv6 error message (parameter problem, code 0). The failure SHOULD be logged (rate limited) and the packet SHOULD be discarded.¶
RFC 9259 [RFC9259] specifies procedures for OAM operations like ping and traceroute on SRv6 SIDs.¶
It is possible to ping a Replication SID of a Leaf/Bud node, assuming the source node knows the Replication SID a priori, directly by putting it in the IPv6 destination address without a SRH or in a SRH as the last segment. While it is not possible to ping a Replication SID of a transit node because transit nodes do not process upper layer headers, it is still possible to ping a Replication SID of Leaf/Bud node of a tree via the Replication SID of intermediate transit nodes. The source of ping MUST compute the ICMPv6 Echo Request checksum using the Replication SID of Leaf/Bud as destination address. The source can then send the Echo Request packet to a transit node's Replication SID. The transit nodes replicate the packet by replacing the IPv6 destination address till the packet reaches the Leaf/Bud node which responds with an ICMPv6 Echo Reply. Appendix A.2.1 illustrates examples of ping to a Replication SID.¶
Traceroute to a Leaf/Bud node Replication SID is not possible due to restriction prohibiting origination of ICMPv6 Time Exceeded error message for a Replication SID as described in the section below.¶
ICMPv6 RFC [RFC4443] Section 2.4 states an ICMPv6 error message MUST NOT be originated as a result of receiving a packet destined to an IPv6 multicast address. This is to prevent a storm of ICMPv6 error messages resulting from replicated IPv6 packets from overwhelming a source node. There are two exceptions (1) the Packet Too Big message for Path MTU discovery, and (2) Parameter Problem Message, Code 2 reporting an unrecognized IPv6 option.¶
An implementation of Replication segment for SRv6 MUST enforce these same restrictions and exceptions. Though this specification does not use any extension header a future extension may do so and MUST support the exception (2) above.¶
Note to the RFC Editor: Please remove this section and reference to RFC 7942 before publication.¶
This section records the status of known implementations of the protocol defined by this specification at the time of posting of this Internet-Draft, and is based on a proposal described in RFC 7942 [RFC7942]. The description of implementations in this section is intended to assist the IETF in its decision processes in progressing drafts to RFCs. Please note that the listing of any individual implementation here does not imply endorsement by the IETF. Furthermore, no effort has been spent to verify the information presented here that was supplied by IETF contributors. This is not intended as, and must not be construed to be, a catalog of available implementations or their features. Readers are advised to note that other implementations may exist. According to RFC 7942 [RFC7942], "this will allow reviewers and working groups to assign due consideration to documents that have the benefit of running code, which may serve as evidence of valuable experimentation and feedback that have made the implemented protocols more mature. It is up to the individual working groups to use this information as they see fit".¶
There are two known implementations of this draft by Cisco and Nokia. Interoperability reports for the implementations are not applicable since this draft does not specify any inter-operable elements of Replication segments.¶
Cisco Implementation uses Replication segments defined in this draft as a basis for PCE to compute and establish P2MP trees in SR domain to provide multi-point services. The implementation, based on latest version of this draft, is in production and supports all MUST and SHOULD clauses for SR-MPLS Replication segments. The documentation is available at Cisco documentation and the point of contact is Rishabh Parekh (riparekh@cisco.com).¶
Nokia has implemented replication SID as defined in this draft to establish P2MP tree in segment routing domain. The implementation supports SR-MPLS encapsulation and has all the MUST and SHOULD clause in this draft. The implementation is at general availability maturity and is compliant with the latest version of the draft. The documentation for implementation can be found at Nokia help and the point of contact is hooman.bidgoli@nokia.com.¶
IANA has assigned the following codepoint for End.Replicate behavior in the "SRv6 Endpoint Behaviors" registry in the "Segment Routing" registry group.¶
| Value | Hex | Endpoint behavior | Reference |
|---|---|---|---|
| 75 | 0x004B | End.Replicate | [This.ID] |
The SID behaviors defined in this document are deployed within an SR domain [RFC8402]. An SR domain needs protection from outside attackers as described in [RFC8754] and following is a brief reminder of the same:¶
For SR-MPLS deployments:¶
For SRv6 deployments:¶
Additionally, an iACL may be applied to all nodes (k) provisioning SIDs as defined in this specification:¶
Failure to protect the SR MPLS domain by correctly provisioning MPLS support per interface permits attackers from outside the domain to send packets that use the replication services provisioned within the domain.¶
Failure to protect the SRv6 domain with IACLs on external interfaces, combined with failure to implement BCP 38 [RFC2827]or apply IACLs on nodes provisioning SIDs, permits attackers from outside the SR domain to send packets that use the replication services provisioned within the domain.¶
Given the definition of the Replication segment in this document, an attacker subverting ingress filter above cannot take advantage of a stack of replication segments to perform amplification attacks nor link exhaustion attacks. Replication segment trees always terminate at a Leaf or Bud node resulting in a decapsulation. This however does allow an attacker to inject traffic to the receivers within a P2MP service.¶
Incorrect provisioning of Replication segments can result in a chain of Replication segments forming a loop. This can happen if Replication segments are provisioned on SR nodes without using a control plane. In this case, replicated packets can create a storm till MPLS TTL (for SR-MPLS) or IPv6 Hop Limit (for SRv6) decrements to zero. A control plane, for example PCE, can be used to prevent loops. Loop prevention is out of scope of this document. The control plane protocols (like PCEP, BGP, etc.) used to instantiate Replication segments can leverage their own security mechanisms such as encryption, authentication filtering etc.¶
For SRv6, Section 5.2 describes how the Parameter Problem Message, code 2 exception for ICMPv6 Error message originated for IPv6 multicast destination can be used by a malicious node to cause a denial-of-service attack. Although this specification does not use any extension headers, any future extension doing so is susceptible to the same security concern.¶
The authors would like to acknowledge Siva Sivabalan, Mike Koldychev, Vishnu Pavan Beeram, Alexander Vainshtein, Bruno Decraene, Thierry Couture, Joel Halpern, Ketan Talaulikar and Darren Dukes for their valuable inputs.¶
Clayton Hassen Bell Canada Vancouver Canada¶
Email: clayton.hassen@bell.ca¶
Kurtis Gillis Bell Canada Halifax Canada¶
Email: kurtis.gillis@bell.ca¶
Arvind Venkateswaran Cisco Systems, Inc. San Jose US¶
Email: arvvenka@cisco.com¶
Zafar Ali Cisco Systems, Inc. US¶
Email: zali@cisco.com¶
Swadesh Agrawal Cisco Systems, Inc. San Jose US¶
Email: swaagraw@cisco.com¶
Jayant Kotalwar Nokia Mountain View US¶
Email: jayant.kotalwar@nokia.com¶
Tanmoy Kundu Nokia Mountain View US¶
Email: tanmoy.kundu@nokia.com¶
Andrew Stone Nokia Ottawa Canada¶
Email: andrew.stone@nokia.com¶
Tarek Saad Cisco Systems Inc. Canada¶
Email:tsaad@cisco.com¶
Kamran Raza Cisco Systems, Inc. Canada¶
Email:skraza@cisco.com¶
This section illustrates an example of a single Replication segment. Examples showing Replication segment stitched together to form P2MP tree (based on SR P2MP policy) are in [I-D.ietf-pim-sr-p2mp-policy].¶
Consider the following topology:¶
R3------R6
/ \
R1----R2----R5-----R7
\ /
+--R4---+
In this example, the Node-SID of a node Rn is N-SIDn and Adjacency-SID from node Rm to node Rn is A-SIDmn. Interface between Rm and Rn is Lmn. The state representation uses "R-SID->Lmn" to represent a packet replication with outgoing replication SID R-SID sent on interface Lmn.¶
Assume a Replication segment identified with R-ID at Replication node R1 and downstream nodes R2, R6 and R7. The Replication SID at node n is R-SIDn. A packet replicated from R1 to R7 has to traverse R4.¶
The Replication segment state at nodes R1, R2, R6 and R7 is shown below. Note nodes R3, R4 and R5 do not have state for the Replication segment.¶
Replication segment at R1:¶
Replication segment <R-ID,R1>: Replication SID: R-SID1 Replication state: R2: <R-SID2->L12> R6: <N-SID6, R-SID6> R7: <N-SID4, A-SID47, R-SID7>¶
Replication to R2 steers packet directly to R2 on interface L12. Replication to R6, using N-SID6, steers packet via shortest path to that node. Replication to R7 is steered via R4, using N-SID4 and then adjacency SID A-SID47 to R7.¶
Replication segment at R2:¶
Replication segment <R-ID,R2>: Replication SID: R-SID2 Replication state: R2: <Leaf>¶
Replication segment at R6:¶
Replication segment <R-ID,R6>: Replication SID: R-SID6 Replication state: R6: <Leaf>¶
Replication segment at R7:¶
Replication segment <R-ID,R7>: Replication SID: R-SID7 Replication state: R7: <Leaf>¶
When a packet is steered into the Replication segment at R1:¶
For SRv6 , we use SID allocation scheme, reproduced below, from Illustrations for SRv6 Network Programming [I-D.filsfils-spring-srv6-net-pgm-illustration]¶
Each node k has:¶
Assume a Replication segment identified with R-ID at Replication node R1 and downstream nodes R2, R6 and R7. The Replication SID at node k, bound to an End.Replicate function, is 2001:db8:cccc:k:Fk::/128. A packet replicated from R1 to R7 has to traverse R4.¶
The Replication segment state at nodes R1, R2, R6 and R7 is shown below. Note nodes R3, R4 and R5 do not have state for the Replication segment. The state representation uses "R-SID->Lmn" to represent a packet replication with outgoing replication SID R-SID sent on interface Lmn. "SL" represents and optional segment list used to steer a replicated packet on a specific path to a Downstream node.¶
Replication segment at R1:¶
Replication segment <R-ID,R1>: Replication SID: 2001:db8:cccc:1:F1::0 Replication state: R2: <2001:db8:cccc:2:F2::0->L12> R6: <2001:db8:cccc:6:F6::0> R7: <2001:db8:cccc:4:C7::0>, SL: <2001:db8:cccc:7:F7::0>¶
Replication to R2 steers packet directly to R2 on interface L12. Replication to R6, using 2001:db8:cccc:6:F6::0, steers packet via shortest path to that node. Replication to R7 is steered via R4, using H.Encaps.Red with End.X SID 2001:db8:cccc:4:C7::0 at R4 to R7.¶
Replication segment at R2:¶
Replication segment <R-ID,R2>: Replication SID: 2001:db8:cccc:2:F2::0 Replication state: R2: <Leaf>¶
Replication segment at R6:¶
Replication segment <R-ID,R6>: Replication SID: 2001:db8:cccc:6:F6::0 Replication state: R6: <Leaf>¶
Replication segment at R7:¶
Replication segment <R-ID,R7>: Replication SID: 2001:db8:cccc:7:F7::0 Replication state: R7: <Leaf>¶
When a packet, (A,B2), is steered into the Replication segment at R1:¶
R1 has to steer packet to Downstream node R7 via node R4. It can do this in one of two ways:¶
This section illustrates ping of a Replication SID.¶
Node R1 pings replication SID of node R6 directly by sending the following packet:¶
Node R1 pings Replication SID of R7 via R4 by sending the following packet with SRH:¶
Assume node R4 is a transit Replication node with Replication SID 2001:db8:cccc:4:F4::0 replicating to R7. Node R1 pings Replication SID of R7 via Replication SID of R4 as follows:¶