RFC 8611

Label Switched Path (LSP) Ping and Traceroute Multipath Support for Link Aggregation Group (LAG) Interfaces

πŸ“… published June 2019, from ietf

Title Block and Vibe Check πŸŽ€

okay bestie, let's get into this πŸ’…βœ¨ RFC 8611 is titled Label Switched Path (LSP) Ping and Traceroute Multipath Support for Link Aggregation Group (LAG) Interfaces, and it's an Internet Standards Track document from June 2019 🌸. The authors are N. Akiya (Big Switch Networks), G. Swallow (SETC), S. Litkowski and B. Decraene (Orange), J. Drake (Juniper Networks), and M. Chen (Huawei) 🫢.

It's marked as Updates: 8029, meaning it's an add-on to RFC 8029, which is the base LSP Ping and Traceroute spec πŸ¦‹. It doesn't obsolete anything, it just levels up an existing spec πŸ”₯.

The ISSN is 2070-1721 πŸ“š.


Abstract πŸ’œ

okay so the whole point of this doc is adding extensions to MPLS LSP Ping and Traceroute (from RFC 8029) so they can find and test specific paths over Layer 2 ECMP across LAG interfaces 🎯. It also defines a way for a Label Switching Router (LSR) to tell you what capabilities it supports πŸ™Œ.

And yes, it updates RFC 8029 πŸ’€.


Status of This Memo πŸ“œ

This is an Internet Standards Track document 🌈. It's a product of the IETF, it represents the community's consensus, got public review, and was approved by the IESG 🫢. More info about Internet Standards lives in Section 2 of RFC 7841 ✨.

You can check the current status, errata, and where to send feedback at https://www.rfc-editor.org/info/rfc8611 πŸ’–.


Copyright (c) 2019 IETF Trust and the document authors, all rights reserved πŸ’…. This doc is subject to BCP 78 and the IETF Trust's Legal Provisions (https://trustee.ietf.org/license-info) as of publication πŸ“. Read them carefully because they explain your rights 🧚. Code extracted from this doc has to include Simplified BSD License text, as described in Section 4.e of the Trust Legal Provisions, and comes with no warranty πŸ”₯.


Table of Contents πŸ—‚οΈ

1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
  1.1.  Background  . . . . . . . . . . . . . . . . . . . . . . .   3
  1.2.  Terminology . . . . . . . . . . . . . . . . . . . . . . .   4
  1.3.  Requirements Language . . . . . . . . . . . . . . . . . .   4
2.  Overview of Solution  . . . . . . . . . . . . . . . . . . . .   4
3.  LSR Capability Discovery  . . . . . . . . . . . . . . . . . .   6
  3.1.  Initiator LSR Procedures  . . . . . . . . . . . . . . . .   7
  3.2.  Responder LSR Procedures  . . . . . . . . . . . . . . . .   7
4.  Mechanism to Discover L2 ECMP . . . . . . . . . . . . . . . .   7
  4.1.  Initiator LSR Procedures  . . . . . . . . . . . . . . . .   7
  4.2.  Responder LSR Procedures  . . . . . . . . . . . . . . . .   8
  4.3.  Additional Initiator LSR Procedures . . . . . . . . . . .  10
5.  Mechanism to Validate L2 ECMP Traversal . . . . . . . . . . .  11
  5.1.  Incoming LAG Member Links Verification  . . . . . . . . .  11
    5.1.1.  Initiator LSR Procedures  . . . . . . . . . . . . . .  11
    5.1.2.  Responder LSR Procedures  . . . . . . . . . . . . . .  12
    5.1.3.  Additional Initiator LSR Procedures . . . . . . . . .  12
  5.2.  Individual End-to-End Path Verification . . . . . . . . .  14
6.  LSR Capability TLV  . . . . . . . . . . . . . . . . . . . . .  14
7.  LAG Description Indicator Flag: G . . . . . . . . . . . . . .  15
8.  Local Interface Index Sub-TLV . . . . . . . . . . . . . . . .  16
9.  Remote Interface Index Sub-TLV  . . . . . . . . . . . . . . .  17
10. Detailed Interface and Label Stack TLV  . . . . . . . . . . .  17
  10.1.  Sub-TLVs . . . . . . . . . . . . . . . . . . . . . . . .  19
    10.1.1.  Incoming Label Stack Sub-TLV . . . . . . . . . . . .  19
    10.1.2.  Incoming Interface Index Sub-TLV . . . . . . . . . .  20
11. Rate-Limiting on Echo Request/Reply Messages  . . . . . . . .  21
12. Security Considerations . . . . . . . . . . . . . . . . . . .  21
13. IANA Considerations . . . . . . . . . . . . . . . . . . . . .  22
  13.1.  LSR Capability TLV . . . . . . . . . . . . . . . . . . .  22
    13.1.1.  LSR Capability Flags . . . . . . . . . . . . . . . .  22
  13.2.  Local Interface Index Sub-TLV  . . . . . . . . . . . . .  22
    13.2.1.  Interface Index Flags  . . . . . . . . . . . . . . .  22
  13.3.  Remote Interface Index Sub-TLV . . . . . . . . . . . . .  23
  13.4.  Detailed Interface and Label Stack TLV . . . . . . . . .  23
    13.4.1.  Sub-TLVs for TLV Type 6  . . . . . . . . . . . . . .  23
    13.4.2.  Interface and Label Stack Address Types  . . . . . .  25
  13.5.  DS Flags . . . . . . . . . . . . . . . . . . . . . . . .  25
14. References  . . . . . . . . . . . . . . . . . . . . . . . . .  25
  14.1.  Normative References . . . . . . . . . . . . . . . . . .  25
  14.2.  Informative References . . . . . . . . . . . . . . . . .  26
Appendix A.  LAG with Intermediate L2 Switch Issues . . . . . . .  27
  A.1.  Equal Numbers of LAG Members  . . . . . . . . . . . . . .  27
  A.2.  Deviating Numbers of LAG Members  . . . . . . . . . . . .  27
  A.3.  LAG Only on Right . . . . . . . . . . . . . . . . . . . .  27
  A.4.  LAG Only on Left  . . . . . . . . . . . . . . . . . . . .  28
Acknowledgements  . . . . . . . . . . . . . . . . . . . . . . . .  28
Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  29

1. Introduction ✨

1.1. Background 🌷

okay so the MPLS LSP Ping and Traceroute mechanisms in [RFC8029] are super powerful for diagnosing every available Layer 3 (L3) path of an LSP, including L3 Equal-Cost Multipath (ECMP) πŸ’–. In a lot of MPLS networks, Link Aggregation Groups (LAGs), defined in [IEEE802.1AX], give you Layer 2 (L2) ECMP and people use them all the time πŸ“.

Here's the catch though 😭. MPLS LSP Ping and Traceroute were never built to find and test specific paths across L2 ECMP. So this happens when an LSP goes over a LAG:

  • Label switching works over some of the LAG's member links, but fails over others πŸ’€.
  • The MPLS echo request for that LSP gets load-balanced onto one of the member links that's working fine πŸ™ˆ.

So even though there's a broken member link, LSP Ping and Traceroute won't catch the label-switching failure 🫠. Basically, missing L2 ECMP coverage means these tools can be totally blind to failures on a bad LAG interface 😱. That's why it's desirable to extend MPLS LSP Ping and Traceroute so they deterministically cover LAG interfaces πŸ‘‘.

This work came out of problems people actually hit in live networks 🌈.

1.2. Terminology πŸ“š

Here are the acronyms and terms used in this doc 🧚:

  • MPLS - Multiprotocol Label Switching.
  • LSP - Label Switched Path.
  • LSR - Label Switching Router.
  • ECMP - Equal-Cost Multipath.
  • LAG - Link Aggregation Group.
  • Initiator LSR - The LSR that sends the MPLS echo request message.
  • Responder LSR - The LSR that receives the MPLS echo request message and sends the MPLS echo reply message.

1.3. Requirements Language πŸ”₯

okay this is the important bit πŸŽ€. The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" are to be read as described in BCP 14 [RFC2119] [RFC8174], but only when they show up in all capitals, exactly like this πŸ’….

2. Overview of Solution πŸ¦‹

2.0. Quick Recap ✨

okay so the big idea is a brand new TLV that lets you discover what a responder LSR can do πŸ’–, plus extensions for MPLS LSP Ping and Traceroute that describe Multipath Information for each individual LAG member link 🫢. That's how these tools can find and test specific L2 ECMP paths over LAG interfaces 🎯. You're expected to already know the Downstream Detailed Mapping TLV (DDMAP) from Section 3.4 of [RFC8029] 🌸.

Here's how the solution plays out 🌈. The initiator's MPLS echo request carries a DDMAP TLV, plus the new LSR Capability TLV to say "hey, I want separate load-balancing info for each L2 next hop over the LAG in the reply" πŸ’…. The responder LSR puts that same LSR Capability TLV in its MPLS echo reply as an acknowledgement πŸ™Œ. It also adds load-balancing info for each downstream LAG member, meaning multipath information and an interface index πŸ¦‹.

This works for every type of LSP that can cross LAG interfaces πŸ‘‘. A lot of LAGs are built from peer-to-peer links, where router X and router X+1 connect directly and have the same number of LAG members πŸ’–. But you can also build LAGs asymmetrically by putting Ethernet switches between two routers 🧚. Appendix A lists some use cases where the mechanisms here might not apply πŸ™ˆ. Also, the mechanisms in this doc don't change anything for LSPs pinned to one particular LAG member, meaning the LAG isn't treated as one logical interface by the LSP 🫢.

Here's an example LDP network 🌸:

  <----- LDP Network ----->

          +-------+
          |       |
  A-------B=======C-------E
          |               |
          +-------D-------+

  ---- Non-LAG
  ==== LAG comprising of two member links

                    Figure 1: Example LDP Network

When node A runs LSP Traceroute to node E, node B sends back load-balancing info for these entries πŸ¦‹:

  1. Downstream C over Non-LAG (upper path).
  2. First Downstream C over LAG (middle path).
  3. Second Downstream C over LAG (middle path).
  4. Downstream D over Non-LAG (lower path).

And here's what this doc defines πŸŽ€:

  • in Section 3, a mechanism to discover capabilities of responder LSRs πŸ”;
  • in Section 4, a mechanism to discover L2 ECMP information πŸ’œ;
  • in Section 5, a mechanism to validate L2 ECMP traversal βœ…;
  • in Section 6, the LSR Capability TLV 🌷;
  • in Section 7, the LAG Description Indicator flag πŸ“;
  • in Section 8, the Local Interface Index Sub-TLV 🧚;
  • in Section 9, the Remote Interface Index Sub-TLV 🌈; and
  • in Section 10, the Detailed Interface and Label Stack TLV πŸ‘‘.
That's Part 1 of 5, bestie 🫢. Part 2 starts at Section 3, so stay tuned ✨.

3. LSR Capability Discovery ✨

Okay so, MPLS Ping works like this: an initiator LSR sends an MPLS echo request message and gets a matching MPLS echo reply message back from a responder LSR. MPLS Traceroute is basically the same energy, except the initiator LSR can potentially send multiple MPLS echo request messages, each with an incrementing TTL value. πŸ¦‹

There have been a bunch of extensions to MPLS Ping and Traceroute over the years πŸ’… So it's often useful, and sometimes necessary, for the initiator LSR to be able to tell these apart for sure:

  • The responder LSR sent the MPLS echo reply message with contents C because it has feature X, Y, and Z implemented. ✨
  • The responder LSR sent the MPLS echo reply message with contents C because it has a subset of features X, Y, and Z (i.e., not all of them) implemented. 🌸
  • The responder LSR sent the MPLS echo reply message with contents C because it does not have features X, Y, or Z implemented. πŸ’€

To help the initiator LSR figure out which one it's dealing with, this document defines the LSR Capability TLV (described in Section 6). πŸŽ€ When the initiator LSR wants to discover the capabilities of the responder LSR, it includes the LSR Capability TLV in the MPLS echo request message. When the responder LSR gets an MPLS echo request message with the LSR Capability TLV included, and it knows the LSR Capability TLV, then it MUST include the LSR Capability TLV in the MPLS echo reply message, describing the features and extensions supported by the local LSR. 🫢 Otherwise, an MPLS echo reply must be sent back to the initiator LSR with the return code set to "One or more of the TLVs was not understood", following the rules defined in Section 3 of [RFC8029]. Then the initiator LSR can send another MPLS echo request without including the LSR Capability TLV. 🧚

It is RECOMMENDED that implementations supporting the LAG multipath extensions defined in this document include the LSR Capability TLV in MPLS echo request messages. πŸ‘‘

3.1. Initiator LSR Procedures 🌈

If an initiator LSR doesn't know what capabilities a responder LSR can support, it can send an MPLS echo request message and carry the LSR Capability TLV to the responder, so it can discover what the responder LSR can support. πŸ“

3.2. Responder LSR Procedures πŸ’œ

When a responder LSR receives an MPLS echo request message that carries the LSR Capability TLV, these procedures are used:

If the responder knows how to process the LSR Capability TLV, the following procedures are used:

  • The responder LSR MUST include the LSR Capability TLV in the MPLS echo reply message. πŸ’–
  • If the responder LSR understands the LAG Description Indicator flag:
    • Set the Downstream LAG Info Accommodation flag if the responder LSR is capable of describing the outgoing LAG member links separately; otherwise, clear the Downstream LAG Info Accommodation flag. πŸ”₯
    • Set the Upstream LAG Info Accommodation flag if the responder LSR is capable of describing the incoming LAG member links separately; otherwise, clear the Upstream LAG Info Accommodation flag. ✨

4. Mechanism to Discover L2 ECMP πŸ¦‹

4.1. Initiator LSR Procedures 🌸

Through LSR Capability Discovery as defined in Section 3, the initiator LSR can figure out whether the responder LSR can describe incoming/outgoing LAG member links separately in the DDMAP TLV. πŸ‘€

Once the initiator LSR knows that a responder can support this mechanism, it sends an MPLS echo request carrying a DDMAP TLV with the LAG Description Indicator flag (G) set to the responder LSR. The LAG Description Indicator flag (G) says that separate load-balancing information for each L2 next hop over a LAG is wanted in the MPLS echo reply. The new LAG Description Indicator flag is described in Section 7. 🫢

4.2. Responder LSR Procedures πŸ’…

When a responder LSR receives an MPLS echo request message with the LAG Description Indicator flag set in the DDMAP TLV, and the responder LSR understands the LAG Description Indicator flag and is capable of describing outgoing LAG member links separately, the following procedures are used, regardless of whether or not the outgoing interfaces include LAG interfaces: πŸŽ€

  • For each downstream interface that is a LAG interface:
    • The responder LSR MUST include a DDMAP TLV when sending the MPLS echo reply. There is a single DDMAP TLV for the LAG interface, with member links described using sub-TLVs. πŸ‘‘
    • The responder LSR MUST set the LAG Description Indicator flag in the DS Flags field of the DDMAP TLV. πŸ’œ
    • In the DDMAP TLV, the Local Interface Index Sub-TLV, Remote Interface Index Sub-TLV, and Multipath Data Sub-TLV are used to describe each LAG member link. All other fields of the DDMAP TLV are used to describe the LAG interface. 🧚
    • For each LAG member link of the LAG interface:
      • The responder LSR MUST add a Local Interface Index Sub-TLV (described in Section 8) with the LAG Member Link Indicator flag set in the Interface Index Flags field. It describes the interface index of this outgoing LAG member link (the local interface index is assigned by the local LSR). 🌈
      • The responder LSR MAY add a Remote Interface Index Sub-TLV (described in Section 9) with the LAG Member Link Indicator flag set in the Interface Index Flags field. It describes the interface index of the incoming LAG member link on the downstream LSR (this interface index is assigned by the downstream LSR). How the local LSR gets the interface index of the LAG member link on the downstream LSR is outside the scope of this document. πŸ“
      • The responder LSR MUST add a Multipath Data Sub-TLV for this LAG member link, if the received DDMAP TLV requested multipath information. πŸ’–

Based on the procedures above, every LAG member link will have a Local Interface Index Sub-TLV and a Multipath Data Sub-TLV entry in the DDMAP TLV. The order of the sub-TLVs in the DDMAP TLV for a LAG member link MUST be Local Interface Index Sub-TLV immediately followed by Multipath Data Sub-TLV, except as follows. A LAG member link MAY also have a corresponding Remote Interface Index Sub-TLV. When a Local Interface Index Sub-TLV, a Remote Interface Index Sub-TLV, and a Multipath Data Sub-TLV are placed in the DDMAP TLV to describe a LAG member link, they MUST be placed in the order of Local Interface Index Sub-TLV, Remote Interface Index Sub-TLV, and Multipath Data Sub-TLV. The blocks of Local Interface Index, Remote Interface Index (optional), and Multipath Data Sub-TLVs for each member link MUST appear adjacent to each other and be in order of increasing local interface index. πŸ”₯

A responder LSR with a LAG interface that has two member links would send the following DDMAP for this LAG interface:

   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
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  ~  DDMAP fields describing LAG interface (DS Flags with G set)  ~
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  | Local Interface Index Sub-TLV of LAG member link #1           |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  | Remote Interface Index Sub-TLV of LAG member link #1          |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  | Multipath Data Sub-TLV LAG member link #1                     |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  | Local Interface Index Sub-TLV of LAG member link #2           |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  | Remote Interface Index Sub-TLV of LAG member link #2          |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  | Multipath Data Sub-TLV LAG member link #2                     |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                       Label Stack Sub-TLV                     |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

            Figure 2: Example of DDMAP in MPLS Echo Reply

When none of the received multipath information maps to a particular LAG member link, the responder LSR MUST still place the Local Interface Index Sub-TLV and the Multipath Data Sub-TLV for that LAG member link in the DDMAP TLV. The value of the Multipath Length field of the Multipath Data Sub-TLV is set to zero. 🧁

4.3. Additional Initiator LSR Procedures πŸ¦‹

The procedures in Section 4.2 let an initiator LSR:

  • Find out whether or not the responder LSR can describe outgoing LAG member links separately, by looking at the LSR Capability TLV. πŸ‘€
  • Use the value of the LAG Description Indicator flag in DS Flags to tell whether each received DDMAP TLV describes a LAG interface or a non-LAG interface. πŸ’–
  • Get multipath information that is expected to traverse the specific LAG member link described by the corresponding interface index. ✨

When an initiator LSR receives a DDMAP containing LAG member information from a downstream LSR with TTL=n, the subsequent DDMAP sent by the initiator LSR to the downstream LSR with TTL=n+1 through a particular LAG member link MUST be updated according to the following procedures: πŸ’…

  • The Local Interface Index Sub-TLVs MUST be removed in the sending DDMAP.
  • If the Remote Interface Index Sub-TLVs were present and the initiator LSR is traversing over a specific LAG member link, then the Remote Interface Index Sub-TLV corresponding to the LAG member link being traversed SHOULD be included in the sending DDMAP. All other Remote Interface Index Sub-TLVs MUST be removed from the sending DDMAP. 🌸
  • The Multipath Data Sub-TLVs MUST be updated to include just one Multipath Data Sub-TLV. The initiator LSR MAY just keep the Multipath Data Sub-TLV corresponding to the LAG member link being traversed, or combine the Multipath Data Sub-TLVs for all LAG member links into a single Multipath Data Sub-TLV when diagnosing further downstream LSRs. πŸŽ€
  • All other fields of the DDMAP are to comply with procedures described in [RFC8029]. 🫢

Figure 3 is an example that shows how to use the DDMAP TLV to send a notification about which member link (link #1 in the example) will be chosen to send the MPLS echo request message to the next downstream LSR: 🧚

   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
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  ~  DDMAP fields describing LAG interface (DS Flags with G set)  ~
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |[OPTIONAL] Remote Interface Index Sub-TLV of LAG member link #1|
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |             Multipath Data Sub-TLV LAG member link #1         |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                       Label Stack Sub-TLV                     |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

            Figure 3: Example of DDMAP in MPLS Echo Request

5. Mechanism to Validate L2 ECMP Traversal ✨

okay so Section 4 is where the responder LSR builds a DDMAP for a downstream LAG πŸŽ€. The Remote Interface Index Sub-TLV, which tells you which incoming LAG member links the downstream LSR has, is optional πŸ™ˆ because the responder often just can't get that info. So when it's missing, you check the traversal with the stuff in Section 5.1. If the LSRs can hand over Remote Interface Index Sub-TLVs, you use Section 5.2 instead πŸ’….

bestie, without Remote Interface Index Sub-TLVs in the DDMAP, checking LAG member link traversal means the initiator LSR has to go through every available LAG member link and then run its own extra logic on the results 😭. This section explains how the initiator gets the extra info from the downstream LSRs and what logic it runs to validate the L2 ECMP traversal 🧚.

5.1.1. Initiator LSR Procedures

An MPLS echo request carrying a DDMAP TLV with the Interface and Label Stack Object Request flag and LAG Description Indicator flag set is sent to ask for a Detailed Interface and Label Stack TLV that includes extra LAG member link information (i.e., interface index) in the MPLS echo reply πŸ’œ.

5.1.2. Responder LSR Procedures

When it gets an echo request with the LAG Description Indicator flag set, a responder LSR that understands that flag and can describe the incoming LAG member link SHOULD use the following procedures, no matter whether the incoming interface was a LAG interface or not 🌸:

  • When the I flag (Interface and Label Stack Object Request flag) of the DDMAP TLV in the received MPLS echo request is set:
    • The responder LSR MUST add the Detailed Interface and Label Stack TLV (described in Section 10) in the MPLS echo reply.
    • If the incoming interface is a LAG, the responder LSR MUST add the Incoming Interface Index Sub-TLV (described in Section 10.1.2) in the Detailed Interface and Label Stack TLV. The LAG Member Link Indicator flag MUST be set in the Interface Index Flags field, and the Interface Index field is set to the LAG member link that received the MPLS echo request.

These procedures let the initiator LSR use the Incoming Interface Index Sub-TLV in the Detailed Interface and Label Stack TLV to figure out, if the incoming interface is a LAG, which incoming LAG member it was πŸ¦‹.

5.1.3. Additional Initiator LSR Procedures

Along with the procedures in Section 4, the procedures in this section let an initiator LSR find out:

  • The expected load-balance information of every LAG member link, at LSR with TTL=n.
  • With specific entropy, the expected interface index of the outgoing LAG member link at TTL=n.
  • With specific entropy, the interface index of the incoming LAG member link at TTL=n+1.

Depending on the LAG traffic division algorithm, the messages may or may not traverse different member links. The expectation is that there's a relationship between the interface index of the outgoing LAG member link at TTL=n and the interface index of the incoming LAG member link at TTL=n+1 for all entropies examined. In other words, the messages with a set of entropies that load-balances to outgoing LAG member link X at TTL=n should all reach the next hop on the same incoming LAG member link Y at TTL=n+1 πŸ“.

With some extra logic, the initiator LSR can run the following checks in a scenario where it (a) knows there's a LAG with two LAG members between TTL=n and TTL=n+1, and (b) has the multipath information to traverse the two LAG member links πŸ”₯.

The initiator LSR sends two MPLS echo request messages to traverse the two LAG member links at TTL=n+1:

  • Success case:
    • One MPLS echo request message reaches TTL=n+1 on LAG member link 1.
    • The other MPLS echo request message reaches TTL=n+1 on LAG member link 2.

The two MPLS echo request messages sent by the initiator LSR reach the immediate downstream LSR from two different LAG member links ✨.

  • Error case:
    • One MPLS echo request message reaches TTL=n+1 on LAG member link 1.
    • The other MPLS echo request message also reaches TTL=n+1 on LAG member link 1.
    • One or both MPLS echo request messages cannot reach the immediate downstream LSR on whichever link.

One or two MPLS echo request messages sent by the initiator LSR cannot reach the immediate downstream LSR, or the two MPLS echo request messages reach the immediate downstream LSR from the same LAG member link 😭.

Note that the procedures above give a deterministic result for LAG interfaces that are back-to-back connected between LSRs (i.e., no L2 switch in between). If there's an L2 switch between the LSR at TTL=n and the LSR at TTL=n+1, there's no guarantee that every incoming interface at TTL=n+1 can be traversed, even when traversing every outgoing LAG member link at TTL=n. Issues from a LAG with an L2 switch in between are described further in Appendix A. LAG provisioning models in operator networks should be considered when analyzing the output of LSP Traceroute that is exercising L2 ECMPs πŸ’€.

5.2. Individual End-to-End Path Verification

When the Remote Interface Index Sub-TLVs are available from an LSR with TTL=n, the validation of LAG member link traversal can be done by the downstream LSR at TTL=n+1. The initiator LSR follows the procedures described in Section 4.3 🌈.

The DDMAP validation procedures for the downstream responder LSR are then updated to compare the incoming LAG member link to the interface index described in the Remote Interface Index Sub-TLV in the DDMAP TLV. If this comparison fails, the return code is set to "Downstream Mapping Mismatch (5)" 🫢.

6. LSR Capability TLV πŸŽ€

This document defines a new TLV called the LSR Capability TLV. It MAY be included in the MPLS echo request message and the MPLS echo reply message. An MPLS echo request message and an MPLS echo reply message MUST NOT include more than one LSR Capability TLV. Having an LSR Capability TLV in an MPLS echo request message is a request that a responder LSR includes an LSR Capability TLV in the MPLS echo reply message, with that TLV describing the features and extensions the responder LSR supports πŸ§šβ€β™€οΈ.

The format of the LSR Capability TLV is shown below:

LSR Capability TLV Type is 4. Length is 4. The LSR Capability TLV has the following format:

   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             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                      LSR Capability Flags                     |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                    Figure 4: LSR Capability TLV

Where:

  • The Type field is 2 octets in length, and the value is 4.
  • The Length field is 2 octets in length, and the value is 4.
  • The LSR Capability Flags field is 4 octets in length; this document defines the following flags:
   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
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                 Reserved (Must Be Zero)                   |U|D|
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

This document defines two flags. The unallocated flags MUST be set to zero when sending and ignored on receipt. Both the U and the D flag MUST be cleared in the MPLS echo request message when sending and ignored on receipt. Zero, one, or both of the flags (U and D) MAY be set in the MPLS echo reply message πŸ¦‹.

   Flag  Name and Meaning
   ----  ----------------

      U  Upstream LAG Info Accommodation

         An LSR sets this flag when the LSR is capable of describing
         a LAG member link in the Incoming Interface Index Sub-TLV
         in the Detailed Interface and Label Stack TLV.

      D  Downstream LAG Info Accommodation

         An LSR sets this flag when the LSR is capable of describing
         LAG member links in the Local Interface Index Sub-TLV and
         the Multipath Data Sub-TLV in the Downstream Detailed
         Mapping TLV.

7. LAG Description Indicator Flag: G πŸ“

This document defines a new flag, the G flag (LAG Description Indicator), in the DS Flags field of the DDMAP TLV 🌸.

The G flag in the MPLS echo request message asks the responder LSR for detailed LAG information. In the MPLS echo reply message, the G flag MUST be set if the DDMAP TLV describes a LAG interface. It MUST be cleared otherwise πŸ’–.

The G flag is defined as below:

The Bit Number is 3.

    0 1 2 3 4 5 6 7
   +-+-+-+-+-+-+-+-+
   | MBZ |G|E|L|I|N|
   +-+-+-+-+-+-+-+-+
   Flag  Name and Meaning
   ----  ----------------

      G  LAG Description Indicator

         When this flag is set in the MPLS echo request, the responder
         LSR is requested to respond with detailed LAG information.
         When this flag is set in the MPLS echo reply, the corresponding
         DDMAP TLV describes a LAG interface.

8. Local Interface Index Sub-TLV πŸ’…

The Local Interface Index Sub-TLV describes the interface index that the local LSR assigned to an egress interface. One or more Local Interface Index sub-TLVs MAY appear in a DDMAP TLV 🧚.

The format of the Local Interface Index Sub-TLV is below:

   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             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                     Local Interface Index                     |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                 Figure 5: Local Interface Index Sub-TLV

Where:

  • The Type field is 2 octets in length, and the value is 4.
  • The Length field is 2 octets in length, and the value is 4.
  • The Local Interface Index field is 4 octets in length; it is an interface index assigned by a local LSR to an egress interface. It's normally an unsigned integer and in network byte order. 🫢

9. Remote Interface Index Sub-TLV πŸ¦‹

okay so the Remote Interface Index Sub-TLV is an optional TLV, and it's giving "tell me who's downstream" πŸ’… It describes the interface index that a downstream LSR assigned to an ingress interface. One or more Remote Interface Index sub-TLVs MAY appear in a DDMAP TLV. 🌸

The format of the Remote Interface Index Sub-TLV looks like this:

   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             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                    Remote Interface Index                     |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 6: Remote Interface Index Sub-TLV πŸŽ€

Where:

  • The Type field is 2 octets in length, and the value is 5. πŸ’œ
  • The Length field is 2 octets in length, and the value is 4. πŸ’œ
  • The Remote Interface Index field is 4 octets in length; it is an interface index assigned by a downstream LSR to an ingress interface. It's normally an unsigned integer and in network byte order. ✨

10. Detailed Interface and Label Stack TLV 🧚

bestie, this one is the detail queen πŸ‘‘ The Detailed Interface and Label Stack TLV MAY be included in an MPLS echo reply message to report the interface on which the MPLS echo request message was received, plus the label stack that was on the packet when it got there. A responder LSR MUST NOT insert more than one instance of this TLV into the MPLS echo reply message. It lets the initiator LSR see the exact interface and label stack info as it looked at the responder LSR. 😭✨

The Type is 6. The Length is K + Sub-TLV Length (sum of Sub-TLVs). K is the sum of all the fields of this TLV before the list of Sub-TLVs, but the length of K depends on the Address Type. The details are described below. Here's the format:

   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             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  | Address Type  |             Reserved (Must Be Zero)           |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                   IP Address (4 or 16 octets)                 |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                   Interface (4 or 16 octets)                  |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  .                                                               .
  .                      List of Sub-TLVs                         .
  .                                                               .
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 7: Detailed Interface and Label Stack TLV πŸŽ€

The Detailed Interface and Label Stack TLV format is derived from the Interface and Label Stack TLV format (from [RFC8029]). Two changes were made. The first is that the label stack is converted into a sub-TLV. The second is that a new sub-TLV is added to describe an interface index. The other fields of the Detailed Interface and Label Stack TLV have the same use and meaning as in [RFC8029]. A quick summary of those fields:

  • Address Type

The Address Type indicates if the interface is numbered or unnumbered. It also determines the length of the IP Address and Interface fields. The resulting total length of the initial part of the TLV is listed as "K Octets". The Address Type is set to one of the following values:

   Type #        Address Type           K Octets
   ------        ------------           --------
        1        IPv4 Numbered                16
        2        IPv4 Unnumbered              16
        3        IPv6 Numbered                40
        4        IPv6 Unnumbered              28
  • IP Address and Interface

IPv4 addresses and interface indices are encoded in 4 octets; IPv6 addresses are encoded in 16 octets.

If the interface upon which the echo request message was received is numbered, then the Address Type MUST be set to IPv4 Numbered or IPv6 Numbered, the IP Address MUST be set to either the LSR's Router ID or the interface address, and the Interface MUST be set to the interface address.

If the interface is unnumbered, the Address Type MUST be either IPv4 Unnumbered or IPv6 Unnumbered, the IP Address MUST be the LSR's Router ID, and the Interface MUST be set to the index assigned to the interface.

Note: Usage of IPv6 Unnumbered has the same issue as [RFC8029], which is described in Section 3.4.2 of [RFC7439]. A solution should be considered and applied to both [RFC8029] and this document. πŸ™ˆ

10.1. Sub-TLVs 🌈

This section defines the sub-TLVs that MAY be included as part of the Detailed Interface and Label Stack TLV. Two sub-TLVs are defined:

        Sub-Type    Sub-TLV Name
        ---------   ------------
          1         Incoming Label Stack
          2         Incoming Interface Index

10.1.1. Incoming Label Stack Sub-TLV πŸ“

The Incoming Label Stack Sub-TLV contains the label stack as received by an LSR. If any TTL values have been changed by this LSR, they SHOULD be restored. πŸ’…

The Type is 1. The Length is variable, and the format looks like this:

   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             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                 Label                 | TC  |S|      TTL      |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  .                                                               .
  .                                                               .
  .                                                               .
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                 Label                 | TC  |S|      TTL      |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 8: Incoming Label Stack Sub-TLV πŸŽ€

10.1.2. Incoming Interface Index Sub-TLV πŸ¦‹

The Incoming Interface Index Sub-TLV MAY be included in a Detailed Interface and Label Stack TLV. It describes the index that a local LSR assigned to the interface that received the MPLS echo request message. ✨

The Type is 2. The Length is 8, and the format is:

   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             |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |    Interface Index Flags      |       Reserved (Must Be Zero) |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |                   Incoming Interface Index                    |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 9: Incoming Interface Index Sub-TLV πŸŽ€

Interface Index Flags

The Interface Index Flags field is a bit vector with the following format:

   0                   1
   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |   Reserved (Must Be Zero)   |M|
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

One flag is defined: M. The remaining flags MUST be set to zero when sending and ignored on receipt. πŸ’œ

  Flag  Name and Meaning
  ----  ----------------

     M  LAG Member Link Indicator

        When this flag is set, the interface index described in this
        sub-TLV is a member of a LAG.

Incoming Interface Index

An Index assigned by the LSR to this interface. It's normally an unsigned integer and in network byte order. 🌸

11. Rate-Limiting on Echo Request/Reply Messages 🚦

okay so an LSP may be over several LAGs, and each LAG may have many member links. To exercise all the links, many echo request/reply messages will be sent in a short period. Those messages may traverse a common path as a burst, and under some circumstances that could cause congestion on the common path. To avoid potential congestion, it is RECOMMENDED that implementations randomly delay the echo request and reply messages at the initiator LSRs and responder LSRs. Rate-limiting of ping traffic is further specified in Section 5 of [RFC8029] and Section 4.1 of [RFC6425], which apply to this document as well. 🫢

12. Security Considerations πŸ”

This document extends the LSP Traceroute mechanism [RFC8029] to discover and exercise L2 ECMP paths, so operators can find problematic member link(s) of a LAG. These on-demand diagnostic mechanisms are used by an operator within an MPLS control domain. 🧚

[RFC8029] reviews the possible attacks and approaches to mitigate possible threats when using these mechanisms. πŸ’€

To prevent leakage of vital information to untrusted users, a responder LSR MUST only accept MPLS echo request messages from designated trusted sources via filtering the source IP address field of received MPLS echo request messages. As noted in [RFC8029], spoofing attacks only have a small window of opportunity. If an intermediate node hijacks these messages (i.e., causes non-delivery), the use of these mechanisms will determine the data plane is not working as it should. Hijacking of a responder node such that it provides a legitimate reply would involve compromising the node itself and the MPLS control domain. [RFC5920] provides additional MPLS network-wide operation recommendations to avoid attacks. Please note that source IP address filtering provides only a weak form of access control and is not, in general, a reliable security mechanism. Nonetheless, it is required here in the absence of any more robust mechanisms that might be used. 🚨

13. IANA Considerations πŸ“‹

13.1. LSR Capability TLV 🌟

IANA has assigned value 4 (from the range 0-16383) for the LSR Capability TLV from the "TLVs" registry under the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry [IANA-MPLS-LSP-PING].

  Type    TLV Name                                    Reference
  -----   --------                                    ---------
    4     LSR Capability                              RFC 8611

13.1.1. LSR Capability Flags πŸŽ€

IANA has created a new "LSR Capability Flags" registry. The initial contents are as follows:

  Value   Meaning                                     Reference
  -----   -------                                     ---------
    31    D: Downstream LAG Info Accommodation        RFC 8611
    30    U: Upstream LAG Info Accommodation          RFC 8611
  0-29    Unassigned

Assignments of LSR Capability Flags are via Standards Action [RFC8126]. πŸ’–

13.2. Local Interface Index Sub-TLV ✨

Okay bestie, IANA gave the Local Interface Index Sub-TLV the value 4 (picked from the 0-16383 range) πŸŽ€ It lives in the "Sub-TLVs for TLV Type 20" subregistry, which sits under the "TLVs" registry of the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry [IANA-MPLS-LSP-PING].

     Sub-Type   Sub-TLV Name                             Reference
     --------   ------------                             ---------
        4       Local Interface Index                    RFC 8611

13.2.1. Interface Index Flags πŸ’…

IANA made a brand new "Interface Index Flags" registry 🌈 The initial contents look like this:

    Bit Number Name                                      Reference
    ---------- --------------------------------          ---------
         15    M: LAG Member Link Indicator              RFC 8611
       0-14    Unassigned

Assignments of Interface Index Flags are via Standards Action [RFC8126]. πŸ”₯

Note that this registry is used by the Interface Index Flags field of the following sub-TLVs:

  • The Local Interface Index Sub-TLV, which may be present in the Downstream Detailed Mapping TLV.
  • The Remote Interface Index Sub-TLV, which may be present in the Downstream Detailed Mapping TLV.
  • The Incoming Interface Index Sub-TLV, which may be present in the Detailed Interface and Label Stack TLV.

13.3. Remote Interface Index Sub-TLV 🧚

IANA has assigned value 5 (from the range 0-16383) to the Remote Interface Index Sub-TLV, so it's now officially in the "Sub-TLVs for TLV Type 20" subregistry of the "TLVs" registry in the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry [IANA-MPLS-LSP-PING]. Iconic πŸ‘‘

     Sub-Type   Sub-TLV Name                             Reference
     --------   ------------                             ---------
       5        Remote Interface Index                   RFC 8611

13.4. Detailed Interface and Label Stack TLV πŸ“

IANA has assigned value 6 (from the range 0-16383) to the Detailed Interface and Label Stack TLV from the "TLVs" registry in the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry [IANA-MPLS-LSP-PING]. The main character of this section, no cap πŸ’–

    Type    TLV Name                                    Reference
    -----   --------                                    ---------
      6     Detailed Interface and Label Stack          RFC 8611

13.4.1. Sub-TLVs for TLV Type 6 🌸

Plot twist first: RFC 8029 changed the registration procedures for TLV and sub-TLV registries for LSP Ping, so that's the context here ✨

IANA has created a new "Sub-TLVs for TLV Type 6" subregistry under the "TLVs" registry of the "Multiprotocol Label Switching (MPLS) Label Switched Paths (LSPs) Ping Parameters" registry [IANA-MPLS-LSP-PING].

This registry conforms with RFC 8029. 🫢

The registration procedures for this sub-TLV registry are:

    Range        Registration Procedure   Note
    -----        ----------------------   -----
    0-16383      Standards Action         This range is for mandatory
                                          TLVs or for optional TLVs that
                                          require an error message if
                                          not recognized.
    16384-31743  RFC Required             This range is for mandatory
                                          TLVs or for optional TLVs that
                                          require an error message if
                                          not recognized.
    31744-32767  Private Use              Not to be assigned
    32768-49161  Standards Action         This range is for optional TLVs
                                          that can be silently dropped if
                                          not recognized.
    49162-64511  RFC Required             This range is for optional TLVs
                                          that can be silently dropped if
                                          not recognized.
    64512-65535  Private Use              Not to be assigned

The initial allocations for this registry are:

    Sub-Type     Sub-TLV Name             Reference Comment
    --------     ------------             --------- -------
    0            Reserved                 RFC 8611
    1            Incoming Label Stack     RFC 8611
    2            Incoming Interface Index RFC 8611
    3-31743      Unassigned
    31744-32767                           RFC 8611  Reserved for
                                                    Private Use
    32768-64511  Unassigned
    64512-65535                           RFC 8611  Reserved for
                                                    Private Use

Note: IETF does not prescribe how the Private Use sub-TLVs are handled; however, if a packet containing a sub-TLV from a Private Use ranges is received by an LSR that does not recognize the sub-TLV, an error message MAY be returned if the sub-TLV is from the range 31744-32767, and the packet SHOULD be silently dropped if it is from the range 64511-65535. πŸ’€ The IETF stays out of how Private Use sub-TLVs get handled, but the rules above are the ones that apply.

13.4.2. Interface and Label Stack Address Types πŸ¦‹

The Detailed Interface and Label Stack TLV shares the Interface and Label Stack Address Types with the Interface and Label Stack TLV. To reflect this, IANA has updated the name of the registry from "Interface and Label Stack Address Types" to "Interface and Label Stack and Detailed Interface and Label Stack Address Types". Shared registry glow-up πŸ’œ

13.5. DS Flags 🌈

IANA has assigned a new bit number from the "DS Flags" subregistry of the "Multiprotocol Label Switched Paths (LSPs) Ping Parameters" registry [IANA-MPLS-LSP-PING]. Rude of me to skip the full name, but you get it πŸ™ˆ

Note: the "DS Flags" subregistry was created by [RFC8029].

     Bit number Name                                        Reference
     ---------- ----------------------------------------    ---------
          3     G: LAG Description Indicator                RFC 8611

14. References πŸ“š

14.1. Normative References

14.2. Informative References


Appendix A. LAG with Intermediate L2 Switch Issues πŸŽ€

Several flavors of provisioning models that use a "LAG with L2 switch" and the corresponding MPLS data-plane ECMP traversal validation issues are described in this appendix. Get ready for the spicy stuff 🌢️

A.1. Equal Numbers of LAG Members

   R1 ==== S1 ==== R2

The issue with this LAG provisioning model is that packets traversing a LAG member from Router 1 (R1) to intermediate L2 switch (S1) can get load-balanced by S1 towards Router 2 (R2). Therefore, MPLS echo request messages traversing a specific LAG member from R1 to S1 can actually reach R2 via any of the LAG members, and the sender of the MPLS echo request messages has no knowledge of this nor any way to control this traversal. In the worst case, MPLS echo request messages with specific entropies will exercise every LAG member link from R1 to S1 and can all reach R2 via the same LAG member link. Thus, it is impossible for the MPLS echo request sender to verify that packets intended to traverse a specific LAG member link from R1 to S1 did actually traverse that LAG member link and to deterministically exercise "receive" processing of every LAG member link on R2. 😭 (Note: As far as we can tell, there's not a better option than "try a bunch of entropy labels and see what responses you can get back", and that's the same remedy in all the described topologies.) 🫠

A.2. Deviating Numbers of LAG Members

              ____
   R1 ==== S1 ==== R2

There are deviating numbers of LAG members on the two sides of the L2 switch. The issue with this LAG provisioning model is the same as with the previous model: the sender of MPLS echo request messages has no knowledge of the L2 load-balancing algorithm nor entropy values to control the traversal. πŸ™ƒ

A.3. LAG Only on Right

   R1 ---- S1 ==== R2

The issue with this LAG provisioning model is that there is no way for an MPLS echo request sender to deterministically exercise both LAG member links from S1 to R2. And without such, "receive" processing of R2 on each LAG member cannot be verified. πŸ’…

A.4. LAG Only on Left

   R1 ==== S1 ---- R2

The MPLS echo request sender has knowledge of how to traverse both LAG members from R1 to S1. However, both types of packets will terminate on the non-LAG interface at R2. It becomes impossible for the MPLS echo request sender to know that MPLS echo request messages intended to traverse a specific LAG member from R1 to S1 did indeed traverse that LAG member. Ngl, that's a lot πŸ₯²


Acknowledgements 🫢

The authors would like to thank Nagendra Kumar and Sam Aldrin for providing useful comments and suggestions. The authors would like to thank Loa Andersson for performing a detailed review and providing a number of comments.

The authors also would like to extend sincere thanks to the MPLS RT review members who took the time to review and provide comments. The members are Eric Osborne, Mach Chen, and Yimin Shen. The suggestion by Mach Chen to generalize and create the LSR Capability TLV was tremendously helpful for this document and likely for future documents extending the MPLS LSP Ping and Traceroute mechanisms. The suggestion by Yimin Shen to create two separate validation procedures had a big impact on the contents of this document. Literal shoutouts all around 🌟

Authors' Addresses πŸ¦‹

  • Nobo Akiya, Big Switch Networks, Email: nobo.akiya.dev@gmail.com
  • George Swallow, Southend Technical Center, Email: swallow.ietf@gmail.com
  • Stephane Litkowski, Orange, Email: stephane.litkowski@orange.com
  • Bruno Decraene, Orange, Email: bruno.decraene@orange.com
  • John E. Drake, Juniper Networks, Email: jdrake@juniper.net
  • Mach(Guoyi) Chen, Huawei, Email: mach.chen@huawei.com
πŸ“¦ where this page comes from