| Internet-Draft | Inter-as-OAM | June 2024 |
| Hegde, et al. | Expires 28 December 2024 | [Page] |
The Segment Routing (SR) architecture leverages source routing and can be directly applied to the use of an MPLS data plane. An SR-MPLS network may consist of multiple IGP domains or multiple Autonomous Systems (ASes) under the control of the same organization. It is useful to have the Label Switched Path (LSP) ping and traceroute procedures when an SR end-to-end path traverses multiple ASes or IGP domains. This document outlines mechanisms to enable efficient LSP ping and traceroute in inter-AS and inter-domain SR-MPLS networks through a straightforward extension to the Operations, Administration, and Maintenance (OAM) protocol, relying solely on data plane forwarding for handling echo replies on transit nodes.¶
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 28 December 2024.¶
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 (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.¶
Many network deployments have built their networks consisting of multiple ASes either for the ease of operations or as a result of network mergers and acquisitions. SR can be deployed in such scenarios to provide end-to-end paths, traversing multiple Autonomous systems (ASes).¶
[RFC8660] specifies SR with an MPLS data plane. [RFC8402] describes BGP Peering Segments, and [RFC9087] describes Centralized BGP Egress Peer Engineering, which will help in steering packets from one AS to another. By utilizing these SR capabilities, it is possible to create paths that span multiple ASes.¶
+----------------+
| Controller/PMS |
+----------------+
|---AS1-----| |------AS2------| |----AS3---|
ASBR2----ASBR3 ASBR5------ASBR7
/ \ / \
/ \ / \
PE1----P1---P2 P3---P4---PE4 P5---P6--PE5
\ / \ /
\ / \ /
ASBR1----ASBR4 ASBR6------ASBR8
Autonomous System: AS1, AS2, AS3
Provider Edge: PE1, PE4, PE5
Provider: P1, P2, P3, P4, P5, P6
AS Boundary Router: ASBR1, ASBR2, ASBR3, ASBR4,
ASBR5, ASBR6, ASBR7, ASBR8
For example, Figure 1 describes an inter-AS network scenario consisting of ASes AS1, AS2 and AS3. AS1, AS2, and AS3 are SR enabled and the egress links have PeerNode SID/PeerAdj SID/ PeerSet SID configured and advertised via [RFC9086]. PeerNode SID/PeerAdj SID/PeerSet SID are referred to as Egress Peer Engineering SIDs (EPE-SIDs) in this document. The controller or the head-end can build an end-to-end Traffic-Engineered path consisting of Node-SIDs, Adjacency-SIDs, and EPE-SIDs. It is useful for operators to be able to perform LSP ping and traceroute procedures on these inter-AS SR-MPLS paths, to detect and diagnose failed deliveries and to determine the actual path that traffic takes through the network. LSP ping/traceroute procedures use IP connectivity for echo reply to reach the head-end. In inter-AS networks, IP connectivity may not be there from each router in the path. For example, in Figure 1, P3 and P4 may not have IP connectivity for PE1.¶
It is not always possible to carry out LSP ping and traceroute functionality on these paths to verify basic connectivity and fault isolation using existing LSP ping and traceroute mechanisms([RFC8287] and [RFC8029]). That is because there might not always be IP connectivity from a responding node back to the source address of the ping packet when the responding node is in a different AS from the source of the ping.¶
[RFC8403] describes mechanisms to carry out MPLS ping/traceroute from a Path Monitoring System (PMS). It is possible to build GRE tunnels or static routes to each router in the network to get IP connectivity for the reverse path. This mechanism is operationally very heavy and requires the PMS to be capable of building a huge number of GRE tunnels or installing the necessary static routes, which may not be feasible.¶
[RFC7743] describes an Echo-relay based solution based on advertising a new Relay Node Address Stack TLV containing a stack of Echo-relay IP addresses. These mechanisms can be applied to SR networks as well. The [RFC7743] mechanism requires the return ping packet to be processed on the slow path or as a bump-in-the-wire on every relay node. The motivation of the current document is to provide an alternate mechanism for ping/traceroute in inter-domain SR networks. The definition of the term "domain" as applicable to this document is defined in Section 1.1.¶
This document describes a new mechanism that is efficient and simple and can be easily deployed in SR-MPLS networks. This mechanism uses MPLS paths and no changes are required in the forwarding path. Any MPLS-capable node will be able to forward the echo-reply packet in the fast path. The current document describes a mechanism that uses the Reply Path TLV [RFC7110] to convey the reverse path. Three new sub-TLVs are defined for the Reply path TLV that facilitate encoding SR label stack. The return path can either be derived by a smart application or a controller that has a full topology view or end-to-end view of a section of the topology. This document also proposes mechanisms to derive the return path dynamically during traceroute procedures.¶
This document focuses on the inter-domain use case. The protocol extensions described may also indicate the return path for other use cases, which are outside the scope of this document and are not further detailed here. The SRv6 data plane is also not covered in this document¶
In this document, the term "domain" refers to an IGP domain where every node is visible to every other node for the purpose of shortest path computation, implying an IGP area or level. An Autonomous System (AS) comprises one or more IGP domains. The procedures described herein are applicable to paths constructed across multiple domains, including both inter-area and inter-AS paths. These procedures and deployment scenarios are relevant for inter-AS paths where the participating ASes are under closely coordinating administrations or single ownership. This document pertains to SR-MPLS networks where all nodes within each domain are SR-capable. It also applies to SR-MPLS networks where SR functions as an overlay with SR-incapable underlay nodes. In such networks, the traceroute procedure is executed only on the overlay SR 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.¶
When the network consists of a large number of nodes, the nodes are segregated into multiple IGP domains as shown in Figure 2. The connectivity to the remote PEs can be achieved using BGP-Labeled Unicast (BGP-LU) [RFC8277] or by stacking the labels for each domain as described in [RFC8604].¶
|-Domain 1|-------Domain 2-----|--Domain 3-|
PE1------ABR1--------P--------ABR2------PE4
\ / \ /\ /
-------- ----------------- -------
BGP-LU BGP-LU BGP-LU
It is useful to support MPLS ping and traceroute mechanisms for these networks. The procedures described in this document for constructing Reply Path TLV and its use in echo reply are equally applicable to networks consisting of multiple IGP domains that use BGP-LU or label stacking.¶
Reply Path (RP) TLV is defined in [RFC7110]. SR networks statically assign the labels to nodes and a PMS/head-end may know the entire Link State Database (LSDB) along with assigned SIDs. The reverse path can be built from the PMS/head-end by stacking segments for the reverse path. Reply Path TLV as defined in [RFC7110] is used to carry the return path. Reply mode 5, Reply via Specified Path is defined in section 4.1 of [RFC7110]. While using the procedures described in this document, the reply mode is set to 5 (Reply via Specified Path), and Reply Path TLV is included in the echo request message as described in [RFC7110]. The Reply Path TLV is constructed as per Section 4.2 of [RFC7110]. This document defines three new sub-TLVs to encode the SR path.¶
The type of segment that the head-end chooses to send in the Reply Path TLV is governed by local policy. Implementations may provide Command Line Interface (CLI) input parameters in the form of Labels/IPv4 addresses/IPv6 addresses or a combination of these which get encoded in the Reply Path TLV. Implementations may also provide mechanisms to acquire the LSDB of remote domains and compute the return path based on the acquired LSDB. For traceroute purposes, the return path will have to consider the reply being sent from every node along the path. The return path changes when the traceroute progresses and crosses each domain. One of the ways this can be implemented on the head-end is to acquire the entire LSDB (of all domains) and build a return path for every node along the SR-MPLS path based on the knowledge of the LSDB. Another mechanism is to use a dynamically computed return path as described in Section 5.5¶
Some networks may consist of IPv4-only domains and IPv6-only domains. Handling end-to-end MPLS OAM for such networks is out of the scope of this document. It is recommended to use dual-stack in such cases and use end-to-end IPv6 addresses for MPLS ping and traceroute procedures.¶
Section 4 of [RFC9256] defines various segment types. The types of segments applicable to this document have been defined in this section for the use of MPLS OAM. The intention was to keep the definitions as close to those in [RFC9256] as possible with modifications only when needed. One or more Segment sub-TLVs can be included in the Reply Path TLV. The Segment sub-TLVs included in a Reply Path TLV MAY be of different types.¶
The below types of Segment sub-TLVs apply to the Reply Path TLV. The code points for the sub-TLVs are taken from the IANA registry common to TLVs 1, 16, and 21. This document defines the Type-A, Type-C, and Type-D Segment sub-TLVs usage and processing when it appears in TLV 21(Reply Path TLV). If these sub-TLVs appear in TLVs 1 or 16, appropriate error codes MUST be returned as defined in [RFC8029].¶
Type-A: SID only, in the form of MPLS Label¶
Type-C: IPv4 Node Address with optional SID¶
Type-D: IPv6 Node Address with optional SID for SR MPLS¶
The Type A Segment Sub-TLV encodes a single SID in the form of an MPLS label. The format is as follows:¶
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Flags | RESERVED |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Label | TC |S| TTL |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
where:¶
Type: 2 octects. Carries value TBD1(to be assigned by IANA from the registry "Sub-TLVs for TLV Types 1, 16, and 21").¶
Length: 2 octets. Carries value 8. The length value excludes the length of the Type and Length Fields.¶
Flags: 1 octet of flags as defined in Section 4.4.¶
RESERVED: 3 octets of reserved bits. MUST be set to zero when sending; MUST be ignored on receipt.¶
Label: 20 bits of label value.¶
TC: 3 bits of traffic class. If the originator wants the receiver to choose the TC value, it MUST set the Traffic Class (TC) field to zero.¶
S: 1 bit Reserved.The S bit MUST be zero upon transmission, and MUST be ignored upon reception.¶
TTL: 1 octet of TTL. If the originator wants the receiver to choose the TTL value, it MUST set the TTL field to 255.¶
The label, TC, S, and TTL collectively referred to as SID.¶
The following applies to the Type-A Segment sub-TLV:¶
The receiver MAY override the originator's values for these fields. This would be determined by local policy at the receiver. One possible policy would be to override the fields only if the fields have the default values specified above.¶
The Type-C Segment sub-TLV encodes an IPv4 node address, SR Algorithm, and an optional SID in the form of an MPLS label. The format is as follows:¶
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Flags | RESERVED (MBZ) | SR Algorithm |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| IPv4 Node Address (4 octets) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| SID (optional, 4 octets) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
where:¶
Type: TBD2 (to be assigned by IANA from the registry "Sub-TLVs for TLV Types 1, 16, and 21").¶
Length: 2 octets. Caries value 8 when no optional SID is included or value 12 when optional SID is included.¶
Flags: 1 octet of flags as defined in Section 4.4.¶
RESERVED: 2 octets of reserved bits. MUST be set to zero when sending; MUST be ignored on receipt.¶
SR Algorithm: 1 octet specifying SR Algorithm as described in section 3.1.1 in [RFC8402] or Flexible algorithm as defined in [RFC9350], when A-Flag as defined in Section 4.4 is present. SR Algorithm is used by the receiver to derive the Label. When A-flag is unset, this field has no meaning and thus MUST be set to zero on transmission and ignored on receipt.¶
IPv4 Node Address: 4-octet IPv4 address representing a node. The IPv4 Node Address MUST be present. It should be a stable address belonging to the node (eg:loopback address).¶
SID: optional: 4-octet field containing label, TC, S and TTL as defined in Section 4.1. When the SID field is present, it MUST be used for constructing the Reply Path.¶
The Type-D Segment sub-TLV encodes an IPv6 node address, SR Algorithm and an optional SID in the form of an MPLS label. The format is as follows:¶
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Flags | RESERVED(MBZ) | SR Algorithm |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
// IPv6 Node Address (16 octets) //
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| SID (optional, 4 octets) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
where:¶
Type: TBD3 (to be assigned by IANA from the registry "Sub-TLVs for TLV Types 1, 16, and 21").¶
Length: 2 Octets. Caries value 20 when no optional SID is included or value 24 when optional SID is included.¶
Flags: 1 octet of flags as defined in Section 4.4.¶
RESERVED: 2-octets of reserved bits. MUST be set to zero when sending; MUST be ignored on receipt.¶
SR Algorithm: 1 octet specifying SR-Algorithm as described in section 3.1.1 in [RFC8402] or Flexible algorithm as defined in [RFC9350], when A-Flag as defined in Section 4.4 is present. SR-Algorithm is used by the receiver to derive the label. When A-flag is unset, this field has no meaning and thus MUST be set to zero (MBZ) on transmission and ignored on receipt.¶
IPv6 Node Address: 16-octet IPv6 address of one interface of a node. The IPv6 Node Address MUST be present. It should be a stable address belonging to the node (eg:loopback address).¶
SID: optional: 4-octet field containing label, TC, S and TTL as defined in Section 4.1. The SID is optional and specifies a 4-octet MPLS SID containing label, TC, S, and TTL as defined in Section 4.1. When the SID field is present, it MUST be used for constructing the Reply Path.¶
The Segment Types described above contain the following flags in the "Flags" field (codes to be assigned by IANA from the new registry "Segment Sub-TLV Flags"):¶
where:¶
A-Flag: This flag indicates the presence of SR Algorithm ID in the "SR-Algorithm" field applicable to various segment Types.¶
Unused bits in the Flag octet MUST be set to zero upon transmission and MUST be ignored upon receipt.¶
The following applies to the Segment Flags:¶
A-Flag applies to Segment Type-C and Type-D. If A-Flag appears with Type-A Segment Type, it MUST be ignored.¶
This section uses the term "initiator" for the node that initiates the MPLS ping or MPLS traceroute procedure. The term "responder" is used for the node that receives the echo request and sends the echo reply. The term egress node is used to identify the last node where the MPLS ping or traceroute is destined to. In an MPLS network any node can be initiator or responder or egress.¶
In the inter-AS scenario, the procedures outlined in this document are employed to specify the return path when IP connectivity to the initiator is unavailable. These procedures may also be utilized regardless of the availability of IP connectivity. The LSP ping initiator MUST set the Reply Mode of the echo request to 5 "Reply via Specified Path", and a Reply Path TLV MUST be carried in the echo request message correspondingly. The Reply Path TLV MUST contain the SR Path in the reverse direction encoded as an ordered list of segments. The first segment MUST correspond to the top segment in the MPLS header that the responder MUST use while sending the echo reply.¶
As described in [RFC7110], when Reply mode is set to 5 (Reply via Specified Path), the echo request must contain the Reply path TLV. Absence of Reply Path TLV is treated as a malformed echo request. When an echo request is received, if the responder does not support the Reply Mode 5 defined in [RFC7110], an echo reply with the return code set to "Malformed echo request received" and the Subcode set to zero must be sent back to the initiator according to the rules of [RFC8029]. If the echo request message contains a malformed Segment sub-TLV, such as incorrect length field, an echo reply with return code set to "Malformed echo request received" and the Subcode set to zero must be sent back to the initiator.¶
When a Reply Path TLV is received, the responder that supports processing it, MUST use the segments in Reply Path TLV to build the echo reply. The responder MUST follow the normal FEC validation procedures as described in [RFC8029] and [RFC8287] and this document does not suggest any change to those procedures. When the echo reply has to be sent out the Reply Path TLV MUST be used to construct the MPLS packet to send out.¶
The echo reply message is sent as an MPLS packet with an MPLS label stack. The echo reply message MUST be constructed as described in the [RFC8029]. An MPLS packet is constructed with an echo reply in the payload. The top label MUST be constructed from the first segment of the Reply Path TLV. The remaining labels MUST be constructed by following the order of the segments from the Reply Path TLV. The MPLS header of the Echo Reply MUST be constructed from the segments in Reply Path TLV and MUST NOT add any other label. The S bit is set for the bottom label as per MPLS specifications [RFC3032] The responder MAY check the reachability of the top label in its own Label Forwarding Information Base (LFIB) before sending the echo reply. If the top label is unreachable, the responder SHOULD send the appropriate return code and follow procedures as per section 5.2 of [RFC7110]. The exception case is when the responder does not have IP reachability to the originator, in which case it may not be possible to send an Echo Reply at all. Even if sent (for example by following a default route present on the responder), the Echo Reply might not reach the originator. The node MAY provide necessary log information in case of unreachability. In certain scenarios, the head-end MAY choose to send Type-C/Type-D segments consisting of IPv4 addresses or IPv6 addresses, when it is unable to derive the SID from available topology information. Optionally SID may also be associated with the Type-C/Type-D segment, if such information is available from the controller or via operator input. In such cases, the node sending the echo reply MUST derive the MPLS labels based on Node-SIDs associated with the IPv4/IPv6 addresses. If optional MPLS SID is present in the Type-C/Type-D segments SID MUST be used to encode the echo reply with MPLS labels. If the MPLS SID does not match with the IPv4 or IPv6 address field in the Type-C or Type-D SID, log information should be generated.¶
The reply path return code is set as described in section 7.4 of [RFC7110]. According to section 5.3 of [RFC7110], the Reply Path TLV is included in an echo reply indicating the specified return path that the echo reply message is required to follow.¶
When the node is configured to dynamically create a return path for the next echo request, the procedures described in Section 5.5 MUST be used. The reply path return code MUST be set to TBA1 and the same Reply Path TLV or a new Reply Path TLV MUST be included in the echo reply.¶
The rules and processes defined in Section 4.6 of [RFC8029] and Section 5.4 of [RFC7110] apply here. In addition, if the Reply Path return code is "Use Reply Path TLV in the echo reply for building the next echo request" (defined in this document), the Reply Path TLV from the echo Reply MUST be sent in the next echo request with TTL incremented by 1. If the initiator node does not support the return code "Use Reply Path TLV in the echo reply for building the next echo request", log information should be generated indicating the return code and the operator may choose to specify the return path explicitly or use other mechanisms to verify the SR policy. If the return code is TBA2, "Local policy does not allow dynamic Return Path building", it indicates that the intermediate node does not support building the dynamic return path. Log information should be generated on the initiator receiving this return code and the operator may choose to specify the return path explicitly or use other mechanisms to verify the SR Policy. If the TTL is already 255, the traceroute procedure MUST be ended with an appropriate log message.¶
In some cases, the head-end may not have complete visibility of Inter-AS/Inter-domain topology. In such cases, it can rely on routers in the path to build the reverse path for MPLS traceroute procedures. For this purpose, the Reply Path TLV in the echo reply corresponds to the return path to be used in building the next echo request. A new return code "Use Reply Path TLV in the echo reply for building the next echo request" is defined in this document.¶
Value Meaning
------ ----------------------
TBA1 Use Reply Path TLV in the echo reply
for building the next echo request.
¶
To dynamically build the return Path for the traceroute procedures, the domain border nodes along the path being traced should support the procedures described in this section. Local policy on the domain border nodes should determine whether the domain border node participates in building the return path dynamically during traceroute.¶
The head-end/PMS node may include its node label while initiating the traceroute procedure. When an Area Border Router (ABR) receives the echo request, if the local policy implies building a dynamic return path, ABR should include its Node label in the reply path TLV and send it in the echo reply. If there is a Reply Path TLV included in the received echo request message, the ABR's node label is added before the existing segments. The type of segment added is based on local policy. In cases when SRGB is not uniform across the network which can be inferred from the LSDB, it is RECOMMENDED to add a Type-C or a Type-D segment, but implementations MAY safely use other approaches if they see benefits in doing so. If the existing segment in the Reply Path TLV is a Type-C/Type-D segment, that segment should be converted to a Type-A segment based on the ABR's own SRGB. This is because downstream nodes in the path will not know what SRGB to use to translate the IP address to a label. As the ABR added its own Node label, it is guaranteed that this ABR will be in the return path and will be forwarding the traffic based on the next label after its label.¶
When an ASBR receives an echo request from another AS, and the ASBR is configured to build the return path dynamically, the ASBR should build a Reply Path TLV and include it in the echo reply. The Reply Path TLV should consist of its node label and an EPE-SID to the AS from where the traceroute message was received. A Reply path return code of TBA1 MUST be set in the echo reply to indicate that the next echo request MUST use the return Path from the Reply Path TLV in the echo reply. ASBR should locally decide the outgoing interface for the echo reply packet. Generally, remote ASBR will choose the interface on which the incoming OAM packet was received to send the echo reply out. In case the ASBR identifies multiple paths to reach the initiator, it MUST choose to send one such path in the Reply Path TLV. Reply Path TLV is built by adding two segment sub TLVs. The top segment sub TLV consists of the ASBR's Node SID and the second segment consists of the EPE-SID in the reverse direction to reach the AS from which the OAM packet was received. The type of segment chosen to build Reply Path TLV is a local policy. It is recommended to use the Type-C/Type-D segment for the top segment when the SRGB is not guaranteed to be uniform in the domain.¶
Irrespective of which type of segment is included in the Reply Path TLV, the responder to the echo requests MUST always translate the Reply Path TLV to a label stack and build an MPLS header for the echo reply packet. This procedure can be applied to an end-to-end path consisting of multiple ASes. Each ASBR that receives an echo request from another AS adds its Node-SID and EPE-SID on top of the existing segments in the Reply Path TLV.¶
An ASBR that receives the echo request from a neighbor belonging to the same AS, MUST look at the Reply Path TLV received in the echo request. If the Reply Path TLV consists of a Type-C/Type-D segment, it MUST convert the Type-C/Type-D segment to a Type-A segment by deriving a label from its own SRGB. The ASBR MUST set the reply path return code to TBA1 and send the newly constructed Reply Path TLV in the echo reply.¶
Internal nodes or non-domain border nodes might not set the Reply Path TLV return code to TBA1 in the echo reply message as there is no change in the return Path. In these cases, the head-end node/PMS that initiates the traceroute procedure MUST continue to send the previously sent Reply Path TLV in the echo request message in every subsequent echo request.¶
Note that an ASBR's local policy may prohibit it from participating in the dynamic traceroute procedures. If such an ASBR is encountered in the forward path, dynamic return path-building procedures will fail. In such cases, an ASBR that supports this document MUST set the return code TBA2 to indicate local policies do not allow the dynamic return path building.¶
Value Meaning
------ ---------------------------------------------------
TBA2 Local policy does not allow dynamic Return Path
building.
¶
The procedures described in this document enable LSP ping and traceroute to be executed across multiple IGP domains or multiple ASes that belong to the same administration or closely cooperating administrations. It is assumed that sharing domain internal information across such domains does not pose a security risk. However, procedures described in this document may be used by an attacker to extract the domain's internal information. An operator MUST deploy appropriate filter policies as described in [RFC8029] to restrict the LSP ping/traceroute packets based on origin. It is also RECOMMENDED that an operator deploy security mechanisms such as MACsec [IEEE-802.1AE] on inter-domain links or security-vulnerable links to prevent spoofing attacks.¶
All the security considerations defined in [RFC8029] will be applicable for this document. Appropriate filter policies SHOULD be applied at the edges to prevent attackers from getting into the network. In the event of such a security breach, the network devices MUST have mechanisms to prevent Denial-of-service attacks as described in [RFC8029].¶
IANA should assign three new sub-TLVs from the "sub-TLVs for TLV Types 1, 16, and 21" sub-registry of the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry.¶
Sub-Type Sub-TLV Name Reference
-------- ----------------- ------------
TBD1 SID only in the form of MPLS Section 4.1
label of this document
TBD2 IPv4 Node Address with Section 4.2
optional SID for SR-MPLS of this document
TBD3 IPv6 Node Address with Section 4.3
optional SID for SR-MPLS of this document
¶
The allocation of code points for the segment sub-TLVs should be done from the Standards Action range (0-16383)¶
IANA should create a new "Segment ID Sub-TLV flags" (see Section Section 4.4) registry under the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry.¶
This registry tracks the assignment of 8 flags in the Segment ID sub-TLV flags field. The flags are numbered from 0 (most significant bit, transmitted first) to 7.¶
New entries are assigned by Standards Action. Initial entries in the registry are as follows:¶
Bit number | Name | Reference
------------+----------------------------+--------------
1 | A Flag | Section 4.4
| | of this document
¶
IANA should assign new return codes in the "Reply path return codes" registry under the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry.¶
Value Meaning Reference
-------- ----------------- ------------
TBA1 Use Reply Path TLV This document
from this echo reply
for building next
echo request.
TBA2 Local policy does This document
not allow dynamic
return Path building.
¶
The return codes should be assigned from the Standards Action range (0x0000-0xFFFB).¶
Carlos Pignataro¶
NC State University¶
cpignata@gmail.com¶
Zafar Ali¶
Cisco Systems, Inc.¶
zali@cisco.com¶
This section is to be removed before publishing as an RFC.¶
RFC-Editor: Please clean up the references cited by this section 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 [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.¶
Organization: Juniper Networks¶
Implementation: JUNOS.¶
Description: Implementation for sending Return path TLV with Type-A segment subTLV¶
Maturity Level: Prototype¶
Coverage: Partial. Type-A SIDs in Return Path TLV¶
Contact: shraddha@juniper.net¶
Thanks to Bruno Decraene for suggesting the use of generic Segment sub-TLV. Thanks to Adrian Farrel, Huub van Helvoort, Dhruv Dhody, Dongjie, for careful review and comments. Thanks to Mach Chen for suggesting the use of Reply Path TLV. Thanks to Gregory Mirsky for the detailed review which helped improve the readability of the document to a great extent.¶
This section elaborates examples of the inter-domain ping and Traceroute procedures described in this document.¶
The example topology given in Figure 1 will be used in the below sections to explain LSP ping and traceroute procedures. The PMS/head-end has a complete view of the topology. PE1, P1, P2, ASBR1, and ASBR2 are in AS1. Similarly, ASBR3, ASBR4, P3, P4, and PE4 are in AS2.¶
AS1 and AS2 have SR enabled. IGPs like OSPF/ISIS are used to flood SIDs in each AS. The ASBR1, ASBR2, ASBR3,and ASBR4 advertise BGP EPE-SIDs for the inter-AS links. Topology of AS1 and AS2 are advertised via BGP-Link State (BGP-LS) to the controller/PMS or head-end node. The EPE-SIDs are also advertised via BGP-LS as described in [RFC9086]. The example uses EPE-SIDs for the inter-AS links but the same could be achieved using adjacency-SIDs advertised for a passive IGP link.¶
The description in the document uses below notations for Segment Identifiers (SIDs).¶
Node-SIDs: N-PE1, N-P1, N-ASBR1 etc.¶
Adjacency-SIDs: Adj-PE1-P1, Adj-P1-P2 etc.¶
EPE-SIDs: EPE-ASBR2-ASBR3, EPE-ASBR1-ASBR4, EPE-ASBR3-ASBR2 etc.¶
Consider an SR-MPLS path from PE1 to PE4 consisting of a label stack [N-P1, N-ASBR1, EPE-ASBR1-ASBR4, N-PE4] from Figure 1. In order to perform MPLS ping procedures on this path, the remote end (PE4) needs IP connectivity to head-end PE1, for the echo reply to travel back to PE1. In a deployment that uses a controller-computed inter-domain path, there may be no IP connectivity from PE4 to PE1 as they lie in different ASes.¶
PE1 sends an echo request message to the endpoint PE4 along the path that consists of label stacks [N-P1, N-ASBR1, EPE-ASBR1-ASBR4, N-PE4]. PE1 adds the return path from PE4 to PE1 in the echo request message in the Reply Path TLV. As an example, the Reply Path TLV for PE1 to PE4 for LSP ping is [N-ASBR4, EPE-ASBR4-ASBR1, N-PE1]. This example path provides the entire return path up to the head-end node PE1. The mechanism used to construct the return path is implementation-dependent.¶
An implementation may also build a return Path consisting of labels to reach its own AS. Once the label stack is popped off, the echo reply message will be exposed. The further packet forwarding will be based on IP lookup. An example return Path for this case could be [N-ASBR4, EPE-ASBR4-ASBR1].¶
On receiving an MPLS echo request, PE4 first validates FEC in the echo request. PE4 then builds a label stack to send the response from PE4 to PE1 by copying the labels from the Reply Path TLV. PE4 builds the echo reply packet with the MPLS label stack constructed and imposes MPLS headers on top of the echo reply packet and sends out the packet to PE1. This segment List stack can successfully steer the reply back to the head-end node (PE1).¶
Let us consider a case when P3 node does not have a route to reach N-PE4. On P3 ping packet would be dropped and the head-end node (PE1) will not receive Echo Reply indicating failure.¶
The traceroute procedure involves visiting every node on the path and obtaining echo replies from every node. In this section, we describe the traceroute mechanisms when the head-end/PMS has complete visibility of the LSDB. The head-end/PMS computes the return path from each node in the entire SR-MPLS path that is being tracerouted. The return path computation is implementation-dependent. As the head-end/PMS completely controls the return path, it can use proprietary computations to build the return path.¶
One of the ways the return path can be built is to use the principle of building label stacks by adding each domain border node's Node-SID on the return path label stack as the traceroute progresses. For inter-AS networks, in addition to the border node's Node-SID, EPE-SID in the reverse direction also needs to be added to the label stack.¶
The Inter-domain/inter-AS traceroute procedure uses the TTL expiry mechanism as specified in [RFC8029] and [RFC8287]. Every echo request packet head-end/PMS will include the appropriate return path in the Reply Path TLV. The node that receives the echo request will follow procedures described in Section 5.1 and Section 5.2 to send out an echo reply.¶
For Example:¶
Let us consider a topology from Figure 1. Let us consider an SR-MPLS path [N-P1, N-ASBR1, EPE-ASBR1-ASBR4, N-PE4]. The traceroute is being executed for this inter-AS path for destination PE4. PE1 sends the first echo request with TTL set to 1 and includes a Reply Path TLV consisting of a Type-A segment containing a label derived from its own SR Global Block (SRGB). Note that the type of segment used in constructing the return Path is determined by local policy. If the entire network has the same SRGB configured, Type-A segments can be used. The TTL expires on P1 and P1 sends an echo reply using the return path. Note that implementations may choose to exclude the Reply Path TLV until the traceroute reaches the first domain border as the return IP path to PE1 is expected to be available inside the first domain.¶
The TTL is set to 2 and the next the echo request is sent out. Until the traceroute procedure reaches the domain border node ASBR1, the same return path TLV consisting of a single Label (PE1's node Label) is used. When an echo request reaches the border node ASBR1, and an echo reply is received from ASBR1, the next echo request needs to include an additional label as ASBR1 is a border node. The head-end node has complete visibility of the network LSDB learned via BGP-LS [RFC9552] and [RFC9086] and can derive the details of Autonomous System Boundary Router (ASBR) nodes. The Reply Path TLV is built based on the forward path. As the forward path consists of EPE-ASBR1-ASBR4, an EPE-SID in the reverse direction is included in the Reply Path TLV. The return path now consists of two labels [EPE-ASBR4-ASBR1, N-PE1]. The echo reply from ASBR4 will use this return path to send the reply.¶
The next echo request after visiting the border node ASBR4 will update the return path with the Node-SID label of ASBR4. The return path beyond ASBR4 will be [N-ASBR4, EPE-ASBR4-ASBR1, N-PE1]. This same return path is used until the traceroute procedure reaches the next set of border nodes. When there are multiple ASes the traceroute procedure will continue by adding a set of Node-SIDs and EPE-SIDs as the border nodes are visited.¶
Note that the above return path-building procedure requires the LSDB of all the domains to be available at the head-end/PMS.¶
Let us consider a case when P3 node does not have a route to reach N-PE4. When TTL of the packet is 5, the packet reaches P3 and the packet TTL becomes zero and the packet is sent to the control plane. The FEC validation Procedures are executed and the Echo Reply is sent using the labels in Reply Path TLV which is [N-PE1, EPE-ASBR4-ASBR1, N-ASBR4]. Head-end PE1 increases the TTL to 6 and sends next Echo Request. The packet is dropped at P3 as there is no route on P3 to forward to N-PE4. Traceroute identifies the path [N-P1, N-ASBR1, EPE-ASBR1-ASBR4, N-PE4] is broken at P3.¶
Section 11.1.2.1 assumes the same SRGB is configured on all nodes along the path. The SRGB may differ from one node to another node and the SR architecture [RFC8402] allows the nodes to use different SRGBs. In such scenarios, PE1 finds out the difference in SRGB by looking into the LSDB and sends Type-C (or Type-D in the case of IPv6 networks) segment with the node address of PE1 and with optional MPLS SID associated with the node address. The receiving node derives the label for the return path based on its own SRGB. When the traceroute procedure crosses the border ASBR1, head-end PE1 should send a Type-A segment for N-PE1 based on the label derived from ASBR1's SRGB. This is required because, ASBR4, P3, P4, etc. may not have the topology information to derive SRGB for PE1. After the traceroute procedure reaches ASBR4 the return path will be [N-PE1 (Type-A with label based on ASBR1's SRGB), EPE-ASBR4-ASBR1, N-ASBR4 (Type-C)].¶
If the packet needs to follow return path specific to an algorithm (as defined in [RFC9350]), a Type-C Segment sub-TLV with corresponding algorithm field set should be used. A-flag should be set to indicate that the SID corresponding to the algorithm should be used.¶
To extend the example to 3 or more ASes, let us consider a traceroute from PE1 to PE5 in Figure 1. In this example, the PE1 to PE5 path has to cross 3 domains AS1, AS2, and AS3. Let us consider a path from PE1 to PE5 that goes through [PE1, ASBR1, ASBR4, ASBR6, ASBR8, PE5]. When the traceroute procedure is visiting the nodes in AS1, the Reply Path TLV sent from the head-end consists of [N-PE1]. When the traceroute procedure reaches the ASBR4, the return Path consists of [N-PE1, EPE-ASBR4-ASBR1]. While visiting nodes in AS2, the traceroute procedure consists of Reply Path TLV [N-PE1, EPE-ASBR4-ASBR1, N-ASBR4]. Similarly, while visiting ASBR8, the EPE-SID from ASBR8 to ASBR6 is added to the Reply Path TLV. While visiting nodes in AS3, the Node-SID of ASBR8 would also be added which makes the return Path [N-PE1, EPE-ASBR4-ASBR1, N-ASBR4, EPE-ASBR8-ASBR6, N-ASBR8]¶
Let us consider another example from topology Figure 2. This topology consists of multi-domain IGP with a common border node between the domains. This could be achieved with multi-area or multi-level IGP or multiple instances of IGP deployed on the same node. The return path computation for this topology is similar to the multi-AS computation except that the return path consists of a single border node label.¶
Let us consider a topology from Figure 1. Let us consider an SR policy path built from PE1 to PE4 with the label stack, N-P1, N-ASBR1, EPE-ASBR1-ASBR4, N-PE4. PE1 begins traceroute with the TTL set to 1 and includes [N-PE1] in the Reply Path TLV. The traceroute packet TTL expires on P1 and P1 processes the traceroute as per the procedures described in [RFC8029] and [RFC8287]. P1 sends an echo reply with the same Reply Path TLV with the reply path return code set to 6. The return code of the echo reply itself is set to the return code as per [RFC8029] and [RFC8287]. This traceroute doesn't need any changes to the Reply Path TLV till it leaves AS1. The same Reply Path TLV that is received may be included in the echo reply by P1 and P2 or no Reply Path TLV included so that head-end continues to use the same return path in the echo request that it used to send the previous echo request.¶
When ASBR1 receives the echo request, in the case it receives the Type-C/Type-D segment in the Reply Path TLV in the echo request, it converts that Type-C/Type-D segment to Type-A based on its own SRGB. When ASBR4 receives the echo request, it should form this Reply Path TLV using its Node-SID (N-ASBR4) and EPE-SID (EPE-ASRB4-ASBR1) labels and set the reply path return code to TBA1. Then PE1 should use this Reply Path TLV in subsequent echo requests. In this example, when the subsequent echo request reaches P3, it should use this Reply Path TLV for sending the echo reply. The same Reply Path TLV is sufficient for any router in AS2 to send the reply. Because the first label(N-ASBR4) can direct echo reply to ASBR4 and the second one (EPE-ASBR4-ASBR1) to direct echo reply to AS1. Once the echo reply reaches AS1, normal IP forwarding or the N-PE1 helps it to reach PE1.¶
The example described in the above paragraphs can be extended to multiple ASes by following the same procedure of each ASBR adding Node-SID and EPE-SID on receiving echo requests from neighboring AS.¶
Let us consider a topology from Figure 2. It consists of multiple IGP domains with multiple areas/levels or separate IGP instances. There is a single border node that separates the two domains. In this case, PE1 sends a traceroute packet with TTL set to 1 and includes N-PE1 in the Reply Path TLV. ABR1 receives the echo request and while sending the echo reply adds its node Label to the Reply Path TLV and sets the Reply path return code to TBA1. The Reply Path TLV in the echo reply from ABR1 consists of [N-ABR1, N-PE1]. The next echo request with TTL 2 reaches the P node. It is an internal node so it does not change the return Path. The echo request with TTL 3 reaches ABR2 and it adds its node label so the Reply Path TLV sent in the echo reply will be [N-ABR2, N-ABR1, N-PE1]. echo request with TTL 4 reaches PE4 and it sends an echo reply return code as Egress. PE4 does not include any Reply Path TLV in the echo reply. The above example assumes a uniform SRGB throughout the domain. In the case of different SRGBs, the top segment will be a Type-C/Type-D segment and all other segments will be Type-A. Each border node converts the Type-C/Type-D segment to Type-A before adding its segment to the Reply Path TLV.¶