RFC 9994

MPLS Network Action (MNA) Sub-Stack Specification Including In-Stack Network Actions and Data

๐Ÿ“… published June 2026, from ietf

MPLS Network Action (MNA) Sub-Stack Specification Including In-Stack Network Actions and Data โœจ

Okay bestie, let's get into RFC 9994 ๐Ÿฆ‹ It's an IETF Standards Track document, so this is a real spec and not just vibes ๐Ÿ’… It was published in June 2026 and carries ISSN 2070-1721. The authors are J. Rajamanickam (Ed.), R. Gandhi (Ed.), R. Zigler, H. Song, and K. Kompella. Rajamanickam and Gandhi are from Cisco Systems, Inc., Zigler is from Broadcom, Song is from Futurewei Technologies, and Kompella is from Juniper Networks ๐ŸŒˆ It updates RFC 9789. The header has no Obsoletes line, so it doesn't obsolete anything ๐Ÿ™ˆ

Abstract ๐ŸŒธ

This doc is the MNA Sub-Stack spec ๐ŸŽ€ It covers carrying network actions and Ancillary Data (AD) in the MPLS label stack. MNA can be used to influence packet-forwarding decisions, carry additional Operations, Administration, and Maintenance (OAM) information in the MPLS packet, or perform user-defined operations.

This document updates RFC 9789 to refine the list of pieces of information that must be included in any document that defines an MNA.

Status of This Memo ๐Ÿ“

This is an Internet Standards Track document.

This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). For more on Internet Standards, check Section 2 of RFC 7841.

For the current status, any errata, and how to send feedback, go to https://www.rfc-editor.org/info/rfc9994.

Copyright (c) 2026 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, since 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.

Table of Contents ๐Ÿ—บ๏ธ

1.  Introduction
2.  Conventions Used in This Document
  2.1.  Requirements Language
  2.2.  Abbreviations
  2.3.  Terminology
3.  Overview
4.  Label Stack Entry Formats
  4.1.  LSE Format A: The MNA Sub-Stack Indicator
  4.2.  LSE Format B: The Initial Opcode
  4.3.  LSE Format C: Subsequent Opcodes
  4.4.  LSE Format D: Additional Data
5.  The MNA Sub-Stack
  5.1.  Opcodes
  5.2.  Ancillary Data
  5.3.  Scope
  5.4.  Unknown Network Action Handling
  5.5.  Ordering
6.  Special Opcodes
  6.1.  bSPL Protection
  6.2.  Flag-Based NAIs Without AD
  6.3.  No-Operation Opcode
  6.4.  Extension Opcode
7.  NAS Placement in the Label Stack
  7.1.  Actions When Pushing Labels
8.  Node Capability Signaling
9.  Processing the Network Action Sub-Stack
  9.1.  Encapsulating Node Responsibilities
  9.2.  Transit Node Responsibilities
  9.3.  Penultimate Node Responsibilities
  9.4.  Egress Node Responsibilities
10. Network Action Indicator Opcode Definition
11. Security Considerations
12. Operational Considerations
  12.1.  Manageability Considerations
  12.2.  Performance and Scale Considerations
  12.3.  Backward Compatibility
13. IANA Considerations
  13.1.  MNA bSPL Label
  13.2.  MPLS Network Actions Parameters
    13.2.1.  Network Action Flags Without Ancillary Data
    13.2.2.  Network Action Opcodes
14. References
  14.1.  Normative References
  14.2.  Informative References
Appendix A.  Examples
  A.1.  Network Action Encoding Examples
    A.1.1.  Network Action Flags Without AD
    A.1.2.  Network Action Opcode with AD
    A.1.3.  Network Action Opcode with More AD with Format B
    A.1.4.  Network Action Opcode with More AD with Format C
  A.2.  Network Action Processing Order
    A.2.1.  Network Action Processing Order
Acknowledgments
Contributors
Authors' Addresses

1. Introduction ๐Ÿงš

[RFC3032] defines the encoding of the MPLS label stack, which is the basic structure used to define a forwarding path. Some applications need MPLS packets to perform special network actions and carry optional Ancillary Data (AD) that can affect the packet-forwarding decision or trigger Operations, Administration, and Maintenance (OAM) logging, as described in [RFC9791] for example. AD can also carry extra information, for example for network slice purposes (see [RFC9791]).

The requirements for in-stack network action and In-Stack Data (ISD) are described in [RFC9613].

This document defines the syntax and semantics of network actions and AD encoded in an MPLS label stack. In-stack actions and AD are contained in a Network Action Sub-Stack (NAS), which is recognized by a new base Special-Purpose Label (bSPL). This document follows the framework specified in [RFC9789].

Section 5 of [RFC9789] provides details about information that a document defining a network action must contain. Section 10 of this document updates [RFC9789] by providing a refined list of pieces of information that must be included in any document that defines an MNA.

2. Conventions Used in This Document ๐Ÿ’–

2.1. Requirements Language

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.

2.2. Abbreviations ๐Ÿ”ค

+==============+====================================+===============+
| Abbreviation | Meaning                            | Reference     |
+==============+====================================+===============+
| AD           | Ancillary Data                     | [RFC9613]     |
+--------------+------------------------------------+---------------+
| bSPL         | base Special-Purpose Label         | [RFC9017]     |
+--------------+------------------------------------+---------------+
| BoS          | Bottom of Stack                    | [RFC9789]     |
+--------------+------------------------------------+---------------+
| ECMP         | Equal-Cost Multipath               | [RFC6790]     |
+--------------+------------------------------------+---------------+
| HbH          | Hop-by-Hop                         | [RFC9789]     |
+--------------+------------------------------------+---------------+
| I2E          | Ingress to Egress                  | [RFC9789]     |
+--------------+------------------------------------+---------------+
| IHS          | I2E, HbH, or Select                | This document |
+--------------+------------------------------------+---------------+
| ISD          | In-Stack Data                      | [RFC9613]     |
+--------------+------------------------------------+---------------+
| LSE          | Label Stack Entry                  | [RFC9789]     |
+--------------+------------------------------------+---------------+
| LSP          | Label Switched Path                | [RFC3031]     |
+--------------+------------------------------------+---------------+
| MNA          | MPLS Network Action                | [RFC9789]     |
+--------------+------------------------------------+---------------+
| NAI          | Network Action Indicator           | [RFC9613]     |
+--------------+------------------------------------+---------------+
| NAL          | Network Action Length              | This document |
+--------------+------------------------------------+---------------+
| NAS          | Network Action Sub-Stack           | [RFC9789]     |
+--------------+------------------------------------+---------------+
| NSI          | Network Action Sub-Stack           | This document |
|              | Indicator                          |               |
+--------------+------------------------------------+---------------+
| NASL         | Network Action Sub-Stack           | This document |
|              | Length                             |               |
+--------------+------------------------------------+---------------+
| OAM          | Operations, Administration,        | [RFC6291]     |
|              | and Maintenance                    |               |
+--------------+------------------------------------+---------------+
| RLD          | Readable Label Depth               | [RFC9789]     |
+--------------+------------------------------------+---------------+
| TC           | Traffic Class                      | [RFC5462]     |
+--------------+------------------------------------+---------------+
| TTL          | Time to Live                       | [RFC3032]     |
+--------------+------------------------------------+---------------+

Table 1: Abbreviations

2.3. Terminology ๐Ÿง 

Here are the terms this doc uses:

  • MPLS Egress Node: An MPLS edge node in its role in handling traffic as it leaves an MPLS domain [RFC3031].
  • MPLS Ingress Node: An MPLS edge node in its role in handling traffic as it enters an MPLS domain [RFC3031].
  • MPLS Domain: A contiguous set of nodes that operate MPLS routing and forwarding and that are also in one Routing or Administrative Domain [RFC3031].
  • Encapsulating Node: A node that adds a NAS to the label stack.

3. Overview โœจ

okay so bestie, this is where the MPLS NAS (network action sub-stack) actually shows up ๐ŸŽ€ It's a bunch of Label Stack Entries (LSEs) sitting inside an MPLS label stack, and their whole job is to tell the network which actions to run on the packet ๐Ÿ’… You can have several NASes in one label stack too, and Section 5 explains where they go ๐Ÿฆ‹

This document says how a network action, plus its optional AD (ancillary data), gets packed into a NAS as a stack of LSEs ๐Ÿงš How you'd share AD between several actions in the same NAS? That's handled in other documents, and it doesn't need any special support from the encodings here ๐Ÿ’œ

Here's the new bit ๐Ÿ”ฅ This document defines LSE formats that go beyond the ones in [RFC3032], because these ones either do something different or get processed differently from regular MPLS labels as defined in [RFC3031] ๐Ÿ˜ญ There are three new formats, and together they carry 7 bits of network action opcodes plus however much opcode-specific AD is needed. Format B LSE holds up to 13 bits of AD ๐ŸŒธ Format C LSE holds up to 20 bits of AD โœจ And Format D LSE shows up whenever the opcodes in Format B or Format C need more AD ๐Ÿซถ

Figure 1 shows the whole vibe ๐Ÿ‘‘ The first LSE in an MNA Sub-Stack is Format A ๐Ÿ’– The second is Format B, and it's followed by a Format D LSE for extra data ๐Ÿฆ‹ After that, there might be a Format C LSE for another network action, then another Format D for its data ๐ŸŒˆ You can keep adding Format C and Format D LSEs as you need more actions and more data ๐Ÿ’…

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
   |      MNA-Label=bSPL                   | TC  |S|    TTL        |A
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
   |   Opcode    |        13-bit Data      |R|IHS|S|  NASL |U| NAL |B
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
   |1|                    22-bit Data            |S|  8-bit Data   |D*
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
   |   Opcode    |        16-bit Data            |S|4b Data|U| NAL |C
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
   |1|                    22-bit Data            |S|  8-bit Data   |D*
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
   |   Opcode    |        16-bit Data            |S|4b Data|U| NAL |C
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
   ~                                                               ~
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--
  • Format D LSE presence indicated by NAL greater than one

Figure 1: An MNA Sub-Stack Encoding Example ๐Ÿ“

4. Label Stack Entry Formats ๐ŸŒท

okay so this section is the cheat sheet for all the LSE formats ๐Ÿง The NAS uses a bunch of different formats for different jobs ๐Ÿ’œ Here we just explain the syntax of each format, and the bigger picture (how the NAS fits together and what the LSEs actually mean) comes in the sections after this one ๐Ÿซถ

4.1. LSE Format A: The MNA Sub-Stack Indicator ๐ŸŽ€

Format A is an LSE just like the one in [RFC3032] and [RFC5462] ๐Ÿ’– Its label value is 4, which is the MNA bSPL label from the "Base Special-Purpose MPLS Label Values" IANA registry (see Section 13.1) ๐Ÿฆ‹ That tells everyone there's an MNA in the packet and that an MNA Sub-Stack starts right here in the label stack โœจ

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |      MNA-Label=bSPL                   | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 2: LSE Format A: The MNA Sub-Stack Indicator ๐ŸŒธ

S (1 bit): The BoS [RFC3032]. MUST be set to 0 on transmitted packets ๐Ÿ™ˆ If a packet comes in with an LSE that has the bSPL (4) and the S bit set to 1, then the packet MUST be dropped ๐Ÿ’€

4.2. LSE Format B: The Initial Opcode ๐Ÿฆ„

Format B is the LSE that carries the first opcode in the NAS, plus a bunch of other fields that describe the NAS ๐Ÿงš Its Data field can hold up to 13 bits of AD โœจ

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Opcode    |        13-bit Data      |R|IHS|S|  NASL |U| NAL |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 3: LSE Format B: The Initial Opcode ๐Ÿ“

Opcode (7 bits): The operation code for this LSE. See Section 5.1. ๐Ÿ’…

Data (13 bits): Opcode-specific AD. ๐ŸŽ€

R (1 bit): Reserved. This bit MUST be set to zero on transmission and ignored upon receipt. ๐Ÿ™ˆ

IHS (2 bits): The scope of all the network actions in this NAS. See Section 5.3. ๐ŸŒˆ

S (1 bit): The BoS [RFC3032]. If the NASL value is non-zero, then the S bit MUST be 0. If a packet is received with the S bit set to 1 and a non-zero NASL value, then the packet MUST be dropped. The encapsulating node MUST ensure that the S bit is set to 1 only in the last LSE in the MPLS header. ๐Ÿ˜ญ

NASL (4 bits): The Network Action Sub-Stack Length. This is the number of Format C and Format D LSEs in the NAS, which means it doesn't count the leading Format A LSE or the Format B LSE ๐Ÿง

U (1 bit): Unknown Network Action Handling. See Section 5.4. ๐Ÿ”ฅ

NAL (3 bits): Network Action Length. This is the number of LSEs of additional data, encoded in Format D LSEs (Section 4.4), that follow this Format B LSE ๐Ÿ’– The NAL value MUST be less than or equal to the NASL value in the Format B LSE. If it isn't, the packet MUST be dropped ๐Ÿ’€ A Format C LSE comes next when the NAL value is less than the NASL value โœจ

4.3. LSE Format C: Subsequent Opcodes ๐Ÿฆ‹

Format C is for the subsequent opcodes in the NAS ๐ŸŒธ

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Opcode    |        16-bit Data            |S|4b Data|U| NAL |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 4: LSE Format C: Subsequent Opcodes ๐Ÿงš

Opcode (7 bits): The operation code for this LSE. See Section 5.1. ๐Ÿ’œ

Data (16 bits + 4 bits): Opcode-specific AD. ๐Ÿซถ

S (1 bit): The BoS [RFC3032]. If the NAL value is non-zero and the S bit is set to 1, then the packet MUST be dropped. If this is not the last LSE in the NAS and the S bit is set to 1, then the packet MUST be dropped. The encapsulating node MUST ensure that the S bit is set to 1 only in the last LSE. ๐Ÿ’€

U (1 bit): Unknown Network Action Handling. See Section 5.4. ๐Ÿ”ฅ

NAL (3 bits): Network Action Length. This is the number of LSEs of additional data, encoded in Format D LSEs (Section 4.4) that follow this Format C LSE. The NAL value MUST be less than or equal to the NASL value in the Format B LSE. If it isn't, the packet MUST be dropped ๐Ÿ™ˆ

A Format A and a Format B LSE MUST be present when a Format C LSE is carried in the NAS ๐Ÿ‘‘

4.4. LSE Format D: Additional Data ๐ŸŒท

Format D is used to carry additional data that didn't fit in the LSE with the preceding opcode ๐Ÿ“

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1|                    22-bit Data            |S|  8-bit Data   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 5: LSE Format D: Additional Data ๐ŸŒˆ

1 (1 bit): The most significant bit MUST be set. This stops legacy implementations from misreading this LSE as a special-purpose label when the data starts with zeros ๐ŸŽ€

S (1 bit): The BoS [RFC3032]. If this is not the last LSE for the network action based on the NAL value and the S bit is set to 1, then the packet MUST be dropped. If this is not the last LSE in the NAS and the S bit is set to 1, then the packet MUST be dropped. The encapsulating node MUST ensure that the S bit is set to 1 only in the last LSE. ๐Ÿ˜ญ

Data (22 bits + 8 bits): Opcode-specific AD. ๐Ÿฆ‹

A Format A and a Format B LSE MUST be present when a Format D LSE is carried in the NAS ๐Ÿ’…

5. The MNA Sub-Stack ๐Ÿงœ

The MNA Sub-Stack MUST begin with a Format A LSE (Section 4.1) ๐Ÿ’– Its label value contains the MNA bSPL (4) to show that the MNA Sub-Stack is there โœจ

The TC and TTL values of the Format A LSE keep their meanings as defined in [RFC3032] and [RFC5462] ๐ŸŒธ The TTL and TC values in the Format A LSE get copied from the forwarding label at the top of the label stack ๐Ÿง The penultimate node on the path copies the TTL and TC values from the preceding LSE into the next LSE on the label stack, overwriting that next LSE's TTL and TC values, as specified in Section 3.5 of [RFC3443] and Section 2.6.3 of [RFC3270] in Uniform Mode LSPs ๐Ÿ”ฅ If the node doing this copy isn't aware of MNA, it could overwrite the values in the Format A LSE of the NAS ๐Ÿ˜ญ

The second LSE in a NAS MUST be a Format B LSE (Section 4.2) ๐ŸŒท This one holds an initial opcode plus extra fields that describe the NAS ๐Ÿซถ

The Format B LSE (Section 4.2) could optionally carry additional data in Format D (Section 4.4) LSEs, up to the length encoded in the LSE's NAL value ๐Ÿฆ„

A NAS MAY contain more Format C (Section 4.3) and Format D (Section 4.4) LSEs, up to the length encoded in the NASL value ๐Ÿงš All Format D LSEs MUST follow a Format C or Format B LSE and be included in that LSE's NAL value ๐Ÿ’œ

5.1. Opcodes ๐ŸŽ€

The opcode is a 7-bit field that tells you the semantics of its LSE ๐Ÿ’– A few opcodes have special semantics assigned to them (Section 6) โœจ Everything else acts as NAIs, and those get assigned through IANA (see Sections 10 and 13.2.2) ๐ŸŒˆ

5.2. Ancillary Data ๐Ÿ’…

Okay so, the data field carries opcode-specific data that is AD for a network action ๐ŸŽ€. For opcode 1, the data field carries Flag-Based NAIs without AD โœจ.

The label value (the most significant 20 bits) in one or more consecutive LSEs is commonly used for load-balancing data flows in an ECMP environment ๐Ÿฆ‹. Changing the first 20 bits of an LSE might alter a packet's path and cause packets from the same flow to arrive out of order ๐Ÿ˜ญ. So to keep deployed services stable in ECMP environments that rely on label value information for load-balancing, you have to be careful about what you encode in that LSE ๐Ÿซถ. If the network action data might differ among packets in the same flow, or change while the packet is forwarded across the MPLS network, it MUST NOT be placed in the most significant 20 bits of a Format B LSE (Section 4.2), a Format C LSE (Section 4.3), or a Format D LSE (Section 4.4) ๐Ÿšซ. That means the bits that can safely change at a transit node, or differ between packets of the same flow, are 0 in Format A and Format B LSEs, 7 in a Format C LSE (bits 20-22 and 25-28), and 11 in a Format D LSE (bits 20-22 and 24-31) ๐Ÿ’œ.

Similarly, to preserve service stability, such data also MUST NOT be carried in the most significant 23 bits of these LSEs when the legacy implementation also uses the TC value, in addition to the label value, in all LSEs when computing ECMP decisions ๐ŸŒธ.

The available mitigations for these problems are to use additional Format D LSEs to carry the data, or to place the data in Post-Stack Data as described in [RFC9789] ๐Ÿงš.

In network deployments where it is known that load-balancing of data flows is not used, or where only the explicitly signaled entropy value is used, and it is certain that load-balancing path selection will not be based on the label value of the LSEs, then the data in the label value of the LSEs in the ISD MAY be mutable within the data flow without causing out-of-order delivery of packets ๐Ÿ™Œ.


5.3. Scope ๐ŸŽฏ

The IHS field in the Format B LSE tells you the scope of all the NAIs encoded in the NAS ๐Ÿ‘‘. Scope decides which nodes along the MPLS path should perform the network actions found within the NAS ๐Ÿ’–. The IHS field values are:

                 +======+=========================+
                 | Bits | Scope                   |
                 +======+=========================+
                 | 00   | I2E                     |
                 +------+-------------------------+
                 | 01   | HbH                     |
                 +------+-------------------------+
                 | 10   | Select                  |
                 +------+-------------------------+
                 | 11   | Reserved for future use |
                 +------+-------------------------+

                     Table 2: IHS Scope Values
  • Ingress to Egress (I2E): The network actions in this NAS MUST NOT be processed by any node except the egress node ๐Ÿ™ˆ.
  • Hop-by-Hop (HbH): All nodes along the path MUST process the NAS ๐ŸŒˆ.
  • Select: Only specific nodes along the path that bring the NAS to the top of the stack will perform the action โœจ.

A given NAS can only carry NAIs with the same scope (I2E/HbH/Select) ๐Ÿ“. If you want to support multiple scopes for a single packet, multiple NASes MAY be included in a single label stack ๐Ÿฆ‹.

The egress node is included in the HbH scope ๐Ÿ’…. This means the penultimate node MUST NOT remove a NAS with HbH scope ๐Ÿซถ. The egress node may receive a NAS at the top of the label stack, as discussed in Section 9.4 ๐Ÿ‘€.

A NAS with I2E scope, if present, MUST be encoded after any HbH or Select scope NASes ๐ŸŒŸ. This makes life easier for transit nodes that need to process a NAS with HbH or Select scope ๐Ÿงš.

If a packet is received with the IHS scope set to "Reserved for future use", the packet is processed based on the U bit in the Format B LSE in the NAS ๐Ÿค”.


5.4. Unknown Network Action Handling ๐Ÿ›ก๏ธ

The Unknown Network Action Handling (U) field in a Format B LSE (Section 4.2) and Format C LSE (Section 4.3) is a 1-bit value. It tells a node what to do when it doesn't understand an action within the NAS ๐Ÿ’œ. The possible Unknown Network Action Handling actions are below ๐Ÿ‘‡.

                  +=====+=====================+
                  | Bit | Action              |
                  +=====+=====================+
                  | 0   | Skip to the next NA |
                  +-----+---------------------+
                  | 1   | Drop the packet     |
                  +-----+---------------------+

                     Table 3: Unknown Network
                         Action Handling

When a packet with an Unknown Network Action Handling is dropped, the node should maintain a local counter for this event and may send a rate-limited notification to the operator ๐Ÿ“Š.


5.5. Ordering ๐Ÿ”ข

The network actions encoded in the NAS MUST be processed in the order they appear in the NAS, from the top of the NAS to the bottom ๐Ÿชœ. NAIs encoded as flags (see Section 6.2) MUST be processed from the most significant bit to the least significant bit ๐Ÿ’–. If a label stack contains multiple NASes, they MUST be processed in the order they appear in the label stack, subject to the restrictions in Section 7 ๐ŸŒธ.


6. Special Opcodes ๐ŸŒŸ

Here are the special opcodes defined to build a basic in-stack MNA solution. They're assigned in the "Network Action Opcodes" IANA registry (see Section 13.2.2) ๐Ÿ“‹. In the future, additional special opcodes may be defined, and their code points assigned from this registry ๐Ÿ”ฎ.


6.1. bSPL Protection ๐Ÿ”’

Opcode: 0

Purpose: Legacy implementations may scan the label stack looking for bSPL values ๐Ÿ‘€. As long as the opcode field is non-zero, an LSE cannot be misinterpreted as containing a bSPL ๐Ÿ™Œ. So opcode 0 is reserved and not to be used ๐Ÿšซ.


6.2. Flag-Based NAIs Without AD ๐Ÿšฉ

Opcode: 1

Purpose: This opcode is used for network actions that don't need AD ๐Ÿ’…. A single flag can be used to indicate each of these network actions โœจ.

LSE Formats: B, C, D

Data: The data field carries NAIs, which should be evaluated from the most significant bit to the least significant bit ๐Ÿ”ข. If this opcode is used with LSE Format B only, up to 13 flags may be carried ๐ŸŽ€. If it's used with LSE Format C only, up to 20 flags may be carried ๐Ÿฆ‹. Format D LSEs can be used with Format C LSEs to encode more than 20 flags ๐ŸŒˆ. Flags are assigned from the "Network Action Flags Without Ancillary Data" registry (Section 13.2.1) ๐Ÿ“‹. If flags need to be evaluated in a different order, multiple LSEs using this opcode may be used to specify the requested order ๐Ÿ”€. The Flag-Based NAIs MUST follow the procedure for data specified in Section 5.2 ๐Ÿ’œ.

Scope: This opcode can be used with any scope ๐ŸŒ.


6.3. No-Operation Opcode ๐Ÿซฅ

Opcode: 2

Purpose: This opcode is used to indicate that it does not perform any network action, and it MUST be skipped โญ๏ธ.

LSE Format: B

Scope: Any scope value may be set and it MUST be ignored ๐Ÿ™ˆ.


6.4. Extension Opcode ๐Ÿงฉ

Opcode: 127

Purpose: This opcode is used to extend the current opcode range beyond 127 in the future ๐Ÿš€. If this opcode is not supported, then the packet with opcode 127 MUST be dropped, regardless of the setting of the U bit โŒ. Use of this opcode is outside the scope of this document ๐Ÿ“.


7. NAS Placement in the Label Stack ๐Ÿฅž

The node adding a NAS to the label stack places a copy of the NAS where the relevant nodes can read it ๐Ÿ‘“. Each downstream node along the path has a Readable Label Depth (RLD) ๐Ÿ“. If the NAS is to be processed by a downstream MNA-capable node, then the entire NAS MUST be placed so that it is within RLD by the time the packet reaches that downstream MNA-capable node ๐ŸŽฏ. The RLD of the downstream MNA-capable node MUST be learned as described in Section 2.3.1 of [RFC9789] ๐Ÿ”.

If the label stack is deep, several copies of the NAS may need to be encoded in the label stack ๐Ÿช†.

For a NAS with HbH scope, every node will process the top copy of the NAS ๐Ÿ”. However, the NAS MUST NOT appear at the top of the stack at any MNA-incapable node on the path that is ensured by the encapsulating node using the node capability, as described in Section 8 ๐Ÿ™…โ€โ™€๏ธ.

A NAS MUST NOT appear at the top of the stack after popping the forwarding label on an MNA-incapable node on the path ๐Ÿšซ.

The behavior of a node where a NAS with I2E and HbH scopes is also removed along with popping the forwarding label on a PHP node is outside the scope of this document ๐Ÿคทโ€โ™€๏ธ.

A NAS with Select scope is processed by the node that brings the NAS to the top of the stack (for example, when using the MPLS label pop operation in Segment Routing) ๐Ÿง . Then the NAS is removed from the stack ๐Ÿ—‘๏ธ. The Select scope NAS needs to be inserted after the forwarding label and before the next forwarding label ๐Ÿ“. It could be inserted before or after a NAS with HbH scope ๐Ÿ”„. Note that the case of a NAS with Select scope with an MPLS label swap operation (for example, with RSVP-TE LSPs) is for future study ๐Ÿ”ญ.

For a NAS with I2E scope, only one copy of the NAS needs to be added at the bottom of the stack ๐Ÿ.

A transit node that is not the penultimate node, and that pops a forwarding label and exposes a copy of a NAS, MUST remove that NAS ๐Ÿงน.

An MNA-capable node performing Penultimate Hop Popping (PHP) that pops the forwarding label, with only the NAS(es) remaining on the stack, MUST NOT remove the NAS(es) ๐Ÿ™…. Instead, it forwards the packet with the NAS(es) at the top of the stack to the next node ๐Ÿ“ฆ. Note that the behavior of the PHP node, as defined in [RFC3270] for TC processing and as defined in [RFC3443] for TTL processing, is not modified, regardless of whether the PHP node supports MNA ๐Ÿ’ซ.

The node that receives the NAS at the top of the label stack MUST process it and remove it โœ….


7.1. Actions When Pushing Labels ๐Ÿท๏ธ

An MNA-capable node may need to push additional labels, as well as push new network actions, onto a received packet ๐Ÿ’ช.

While pushing additional labels onto the label stack of the received packet, the MNA-capable node MUST verify that the entire topmost NAS with HbH scope is still within the RLD of the downstream MNA-capable nodes ๐Ÿ”Ž. If required, the MNA-capable node MAY create a copy of the topmost NAS with HbH scope and insert it within the RLD of the downstream MNA-capable nodes on the label stack ๐Ÿ“‘.

When an MNA-capable node needs to push a new NAS with HbH scope onto a received packet that already has a NAS with HbH scope, it SHOULD copy (and merge) the network actions (including their AD) from the received topmost NAS with HbH scope into the new NAS with HbH scope ๐Ÿค. The new NAS MUST be placed within the RLD of the downstream MNA-capable nodes ๐Ÿ“. This behavior can be based on local policy โš–๏ธ.

The new network actions added MUST NOT conflict with the network actions in the received NAS with HbH scope ๐Ÿ™…โ€โ™€๏ธ. The mechanism to resolve such conflicts depends on the network actions and can be based on local policy ๐Ÿงฉ. The MNA-capable node that pushes entries MUST understand any network action it is pushing that may result in a conflict, and it MUST resolve any conflicts between new and received network actions ๐Ÿ› ๏ธ. In the usual case of a conflict from duplicating a network action, the definition of that network action MUST give guidance on conflict resolution ๐Ÿ“–.

8. Node Capability Signaling โœจ

okay so bestie, the encapsulating node MUST make sure the NAS can be processed by the transit and egress nodes ๐Ÿ’… and on top of that, the encapsulated packet MUST NOT exceed the path MTU as described in [RFC3032]. no cap, this is the foundation of the whole vibe ๐Ÿซถ

  • the node responsible for picking a path through the MPLS network needs to know and consider the MNA-capabilities and RLD of the transit nodes, plus the MNA-capabilities of the egress node, as described in Section 2.3 of [RFC9789]. ๐Ÿฆ‹
  • info about node capabilities may be configured, collected through management protocols, or distributed by control protocols (like advertising by routing protocols). it's giving flexibility ๐ŸŒธ
  • the node responsible for picking a path through the MPLS network learns about node capabilities using mechanisms that are out of scope for this document. not our problem here, bestie ๐Ÿ™ˆ
  • for Segment Routing over MPLS (SR-MPLS), as well as the RLD, the path computation system needs to know the Maximum SID Depth (MSD) [RFC8664] that can be imposed at the ingress node of a given SR path. this makes sure the label stack depth of a computed path does not exceed the maximum number of labels (i.e., MSD) the node can impose, AND the maximum number of labels that can be read by the MNA-processing nodes in the path. the MSD MUST include the MNA Sub-Stacks that will be added. โญ
  • the encapsulating node MUST learn about the RLD of the nodes in the path as described in Section 2.3.1 of [RFC9789]. ๐Ÿ‘‘

9. Processing the Network Action Sub-Stack ๐ŸŽ€

this section lays out the specific responsibilities for nodes along an LSP [RFC3031]. let's get into it ๐Ÿ’–

9.1. Encapsulating Node Responsibilities ๐ŸŒˆ

the encapsulating node MAY add NASes to the label stack in accordance with its policies, the placement restrictions in Section 7, and the capabilities learned from Section 8. it's got options, lowkey ๐Ÿงš

if there is an existing label stack, the encapsulating node MUST NOT modify the first 20 bits of any LSE in the label stack when the ECMP technique in the network uses hashing of the labels on the label stack. do NOT touch those bits, periodt ๐Ÿ™…โ€โ™€๏ธ๐Ÿ’œ

9.2. Transit Node Responsibilities ๐Ÿ“

the transit node is the node that processes a NAS in the label stack but does not push any new NAS. ๐Ÿ’…

the transit node MUST follow the procedure for data specified in Section 5.2. no skipping, bestie โœจ

transit nodes MUST process the NASes in the label stack according to the rules set out in Section 5.5. ๐Ÿ”ฅ

a transit node that processes a NAS and does not recognize the value of an opcode MUST follow the rules according to the setting of the Unknown Network Action Handling value in the NAS as described in Section 5.4. that's the vibe when it meets something unfamiliar ๐Ÿฆ‹

9.3. Penultimate Node Responsibilities ๐ŸŒธ

in addition to the transit node responsibilities, the penultimate node and penultimate SR-MPLS segment node MUST NOT remove the last copy of an HbH or I2E NAS when it is exposed after removing the forwarding (transport) label. this way the egress node can still process the NAS. protect that NAS, babe ๐Ÿซถ

9.4. Egress Node Responsibilities ๐Ÿ’€

the egress node MUST remove any NAS it receives. clean up after yourself, it's giving tidy ๐Ÿงนโœจ

10. Network Action Indicator Opcode Definition ๐Ÿ’–

the following information MUST be defined for a new NAI opcode request in the document that specifies the network action. this updates the list found in Section 5 of [RFC9789] and should be used instead of that list. ngl that's a big update ๐Ÿ‘‘

Format: the definition of the new network action MUST specify the LSE formats. the opcode can define the network action in Format B or C or both Formats B and C. both Format B and C LSEs MAY optionally carry Format D LSEs. ๐ŸŒˆ

Scope: the definition of the new network action MUST specify at least one scope (I2E, HbH, Select) for the network action and MAY specify more than one scope. ๐ŸŽ€

Ancillary Data: the definition of the new network action MUST specify the quantity, syntax, and semantics of any associated AD. the AD MAY be variable length, but the NAL MUST be computable based on the data added in the NAS. ๐Ÿงš

Processing: the definition of the new network action MUST specify the detailed procedure for processing the network action. ๐Ÿ’…

Interactions: the definition of the new network action MUST specify its interaction including merging with other currently defined network action if there is any. ๐Ÿฆ‹

an assignment for a NAI MAY make requests from any combination of the "Network Action Opcodes" or "Network Action Flags Without Ancillary Data" assignments (see Section 13). this decision should optimize for eventual encoding efficiency. if the NAI does not require any AD, then a flag is preferred as only one bit is used in the encoding. bit-saving is iconic ๐Ÿ“โœจ

11. Security Considerations ๐Ÿ”ฅ

this part is SERIOUS bestie, so read carefully ๐Ÿ˜ญ

the security considerations in [RFC3032] and [RFC9789] also apply to this document. ๐Ÿซถ

in addition, MNA creates a new dimension in security concerns:

  • the actions of an encapsulating node can affect any or all of the nodes along the path. in the most common and benign situations, a syntactically incorrect packet could result in packet loss or corruption, for example. ๐Ÿ’€
  • the semantics of a network action are unbounded and may be insecure. a network action could be defined that makes arbitrary changes to the memory of the forwarding router, which could then be used by the encapsulating node to compromise every MNA-capable router in the network. this is NOT it, chat ๐Ÿ˜ฑ๐Ÿ”ฅ
  • the MNA architecture supports locally defined network actions. for such actions, there will be limited oversight to ensure that the semantics do not create security issues. implementors and network operators will need to ensure that even the locally defined network actions do not compromise the security of the network by following the security considerations specified in this document. everyone needs to be careful ๐Ÿ™ˆ
  • the MPLS domain border nodes MUST ensure that the MPLS packets with MNA from any domain with a different administrative control can be filtered to prevent entering the provider MPLS domain. the filtering capability MAY be enabled on a per-network-action basis, and it can be based on a local policy. the filtering capability MUST be implemented on those nodes before deploying MNA in the provider MPLS domain. the RLD on the filtering node MUST be higher than the RLD on all other nodes in the provider MPLS domain. guard the border, no exceptions ๐Ÿ›ก๏ธ๐Ÿ’œ
  • the MNA architecture supports modifying the AD on the intermediate nodes so the critical network functions either should not rely on the data or should be aware of the risks and use other means to verify the security of the whole network. don't trust the AD blindly, bestie ๐Ÿ™…โ€โ™€๏ธ
  • system designers must be aware that information included in AD may be transmitted "in the clear". network actions that require the exchange of sensitive data MUST be defined in such a way that the data is encrypted in transit. otherwise, sensitive data MUST NOT be transmitted using these mechanisms. private stuff stays encrypted, period ๐Ÿ”๐Ÿ’–
  • mis-delivery of a packet due to malformed forwarding action data could be considered a security risk. ๐Ÿšจ

12. Operational Considerations ๐ŸŒธ

12.1. Manageability Considerations ๐ŸŽ€

an MNA implementation MAY collect the following counters:

  • Packets with MNA received
  • MNA Sub-Stacks processed
  • MNA per-network-action counters
  • Packets with MNA dropped due to unknown actions
  • Packets with MNA skipped due to unknown actions
  • Packets with MNA dropped due to malformed NAS

additionally, tracking both successful invocations and failures for each specific network action is RECOMMENDED to provide granular visibility. nodes MAY generate rate-limited notifications or alarms for significant operational events, such as sustained high rates of MNA packet drops or frequent encounters of malformed MNA Sub-Stacks, to alert operators to potential issues. comprehensive logging of MNA processing details and outcomes can aid in network diagnostics and post-mortem analysis. the data tea is always useful ๐Ÿ“Šโ˜•

12.2. Performance and Scale Considerations ๐Ÿฆ‹

performance and scale assessments are outside the scope of this document; the authors of any future MNA application documents are encouraged to address them. future authors, you've got homework ๐Ÿ“š๐Ÿงš

12.3. Backward Compatibility ๐Ÿ’œ

this section talks about interactions between MNA-capable and MNA-incapable nodes. the drama of old meets new ๐Ÿฟ

an MNA encapsulating node MUST ensure that the MPLS NAS is not at the top of the MPLS label stack when the packet arrives at an MNA-incapable node. if such a packet did arrive at an MNA-incapable node, it will most likely be dropped as described in Section 2.1.1 of [RFC7325]. top of the stack is a no-go, bestie ๐Ÿ™…โ€โ™€๏ธ

any node could scan the label stack, potentially looking for a label value containing a bSPL. to ensure that the LSE formats described herein do not appear to contain a bSPL value, the opcode value of 0 has been reserved. by ensuring that there is a non-zero value in the high-order 7 bits, we are assured that the high-order 20 bits cannot be misinterpreted as containing a bSPL value (0-15). it's a clever little disguise ๐Ÿฅธโœจ

the TC and TTL values of the Format A LSE are not repurposed for encoding, as the penultimate node on the MPLS packet path may propagate TTL from the transport (or forwarding) label to the next label on the label stack, overwriting the TTL on the next label. if the penultimate node is a legacy node, it might perform this action, potentially corrupting other values stored in the TC and TTL values. to protect against this, we retain the TC and TTL values in the Format A LSE. TC and TTL stay put, iykyk ๐Ÿซถ๐ŸŽ€

when adding the Entropy Label Indicator (ELI) (bSPL 7) and Entropy Label (EL) as defined in [RFC6790], along with an MNA NAS, the RLD MUST be considered for the placement of both, and they both can be placed in any order. if a transit LSR chooses to use as much of the whole label stack as feasible as a key for the load-balancing function, the MNA-reserved label MUST NOT be used as a key for the load-balancing function, as specified in Section 4.3 of [RFC6790]. note that the behavior of an MNA-incapable transit LSR that scans the label stack for ELI and EL but encounters a different, unrecognized reserved label first, is not modified by this document. ELI and EL have a whole friendship arc ๐ŸŒˆ๐Ÿ’–

similarly, when adding the Flow-ID Label Indicator (FLI) (including the extension label 15) and Flow-ID Label (FL) as defined in [RFC9714], along with an MNA NAS, the RLD MUST be considered for the placement of both, and they both can be placed in any order. note that the behavior of an MNA-incapable transit LSR that scans the label stack for FLI (including the extension label 15) and FL, but encounters a different, unrecognized reserved label first, is not modified by this document. same energy as ELI and EL, FLI and FL are giving siblings ๐Ÿ‘ฏโ€โ™€๏ธโœจ

however, as the existing behavior is not specified for transit LSRs, upon encountering any unrecognized bSPLs or extended SPLs (eSPLs) below the top of the label stack, some existing implementations may have chosen to implement non-standardized actions, such as discarding packets. any uses of a new bSPL or eSPL would cause issues with such existing implementations using the non-standardized actions upon encountering unrecognized bSPLs or eSPLs below the top of the label stack. since this is a generic problem, any clarifications for the treatment of unrecognized bSPL or eSPL are outside the scope of this document. this one's a bit of a mess, not gonna lie ๐Ÿ˜ฌ๐Ÿ’€

13. IANA Considerations ๐Ÿฆ‹

13.1. MNA bSPL Label ๐Ÿ’œ

IANA has allocated the value 4 for the MNA bSPL label from the "Base Special-Purpose MPLS Label Values" registry to indicate the presence of an MNA Sub-Stack in the label stack. the description of the value is "MPLS Network Actions". the registry has spoken, and label 4 is MNA's forever ๐Ÿ‘‘๐ŸŒธ

13.2. MPLS Network Actions Parameters โœจ

Okay so IANA made a registry group called "MPLS Network Actions" ๐ŸŽ€ It holds two registries: the "Network Action Flags Without Ancillary Data" registry (see Section 13.2.1) and the "Network Action Opcodes" registry (see Section 13.2.2). Let's go ๐Ÿฆ‹

13.2.1. Network Action Flags Without Ancillary Data

Bestie, for the "Network Action Flags Without Ancillary Data" registry, registration requests should comply with Section 10 ๐Ÿ’… The registration procedure depends on the range and is "IETF Review", "Experimental Use", or "Private Use" (as defined in [RFC8126]). The fields are "Bit Position" (integer), "Description" (string), and "Reference" (string) ๐Ÿ“‹

Bit Position counts from the most significant bit of the LSE Format B or C Data field and any Format D LSEs that come after it ๐ŸŒธ Bit Position 0 is the most significant bit in an LSE Format B or C Data field. Bit Position 20 is the most significant bit in the first LSE Format D Data field. Format C has 20 bits available and Format D has 30 bits available. There are at most 14 Format D LSEs per opcode, because of the NASL limit of 15 and the rule that Format D needs a Format C LSE. So there are at most 20 + 14 * 30 = 440 bit positions ๐Ÿงฎ The Bit Position value is an integer from 0-439. The initial registry has no entries ๐Ÿซถ

The registration procedures for code point allocation in this registry are in Table 4 ๐Ÿ’–

                 +========+========================+
                 | Range  | Registration Procedure |
                 +========+========================+
                 | 0-14   | IETF Review            |
                 +--------+------------------------+
                 | 15-16  | Experimental Use       |
                 +--------+------------------------+
                 | 17-19  | Private Use            |
                 +--------+------------------------+
                 | 20-439 | IETF Review            |
                 +--------+------------------------+

                        Table 4: Registration
                     Procedures for the "Network
                    Action Flags Without Ancillary
                            Data" Registry

13.2.2. Network Action Opcodes

For the "Network Action Opcodes" registry, registration requests should comply with Section 10 as well as the Security Considerations section (Section 11) ๐Ÿ” The registration procedure depends on the range and is "IETF Review", "Experimental Use", or "Private Use" (as defined in [RFC8126]). The fields are "Opcode" (integer), "Description" (string), and "Reference" (string). Opcode is an integer from 1-126 ๐Ÿ“

                +=========+========================+
                | Range   | Registration Procedure |
                +=========+========================+
                | 1-110   | IETF Review            |
                +---------+------------------------+
                | 111-114 | Experimental Use       |
                +---------+------------------------+
                | 115-126 | Private Use            |
                +---------+------------------------+
                | 127     | IETF Review            |
                +---------+------------------------+

                  Table 5: Registration Procedures
                  for the "Network Action Opcodes"
                              Registry

IANA has already allocated values for these network action opcodes from the "Network Action Opcodes" registry, as shown in Table 6 ๐ŸŒˆ

         +========+===========================+===========+
         | Opcode |        Description        | Reference |
         +========+===========================+===========+
         | 0      | Reserved                  | RFC 9994  |
         +--------+---------------------------+-----------+
         | 1      | Flag-Based Network Action | RFC 9994  |
         |        | Indicators without AD     |           |
         +--------+---------------------------+-----------+
         | 2      | No operation Opcode       | RFC 9994  |
         +--------+---------------------------+-----------+
         | 127    | Opcode Range Extension    | RFC 9994  |
         |        | Beyond 127                |           |
         +--------+---------------------------+-----------+

             Table 6: Initial Contents of the "Network
                      Action Opcodes" Registry

14. References ๐Ÿ“š

14.1. Normative References

14.2. Informative References


Appendix A. Examples ๐Ÿง

A.1. Network Action Encoding Examples

A.1.1. Network Action Flags Without AD

Here's the first figure, a NAS with Flag-Based NAIs without AD ๐Ÿ‘€

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |          MNA-Label=bSPL               | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Opcode=1   |        13-bit Flags     |R|IHS|S|NASL=0 |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                  Figure 6: NAS with Network Action Flags

This is an example of a NAS with Flag-Based NAIs without AD ๐Ÿฆ‹

Details:

  • Opcode=1: This opcode indicates that the LSE carries Flag-Based NAIs without AD.
  • Data: The data field carries the Flag-Based NAIs.
  • S: This is the bottom of the stack bit. Set if and only if this LSE is the bottom of the stack.
  • U: Action to be taken if one of the NAIs is not recognized by the processing node.
  • NASL: The NASL value is set to 0, as there are no additional LSEs.
  • NAL: The NAL value is set to 0, as there are no additional AD encoded using Format D.

Now the second one, which is a bit more of a glow-up ๐ŸŒŸ

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |      MNA-Label=bSPL                   | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Opcode=2   |        Data=0           |R|IHS|S|NASL=2 |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Opcode=1   |        Flag-Based NAIs        |S| NAIs  |U|NAL=1|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1| Additional Flag-Based NAIs                |S|Flag-Based-NAIs|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

        Figure 7: Network Action Flags Without AD Using LSE Format D

In this example, the NAS has a Format B LSE with a No-Operation Opcode value of 2 ๐Ÿ™ˆ The next LSE uses Format C, but the network action flag isn't in a bit position inside that Format C LSE. So a single Format D LSE gets added to the NAS to carry the flag ๐Ÿ’…

NAL is set to 1 to show that Flag-Based NAIs are also encoded in the next LSE โœจ NASL is set to 2 to show that two additional LSEs are used ๐ŸŽ€

A.1.2. Network Action Opcode with AD ๐Ÿ’–

Okay so this one is a single LSE with the network action's AD tucked right in โœจ Here's the picture:

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |      MNA-Label=bSPL                   | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Opcode=8   |      Ancillary Data     |R|IHS|S|NASL=0 |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 8: Network Action Opcode with Ancillary Data

In this example, the NAS is only carrying one Network Action, and that action needs 13 bits of AD ๐ŸŽ€

Details on the second LSE:

  • Opcode=8: A network action allocation is outside of this document.
  • Data: The data field contains 13 bits of AD.

A.1.3. Network Action Opcode with More AD with Format B ๐Ÿฆ‹

Sometimes a network action needs more AD than fits in one LSE, so we add a Format D LSE to carry the rest ๐ŸŒธ

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |          MNA-Label=bSPL               | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Opcode=10  |      Ancillary Data     |R|IHS|S|NASL=1 |U|NAL=1|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1|            Ancillary Data                 |S|Ancillary Data |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 9: Network Action with Additional Ancillary Data

In this example, opcode 10 is encoded in Format B, and since it needs more than one LSE's worth of AD, a Format D LSE gets added ๐Ÿงš

Details on the second LSE:

  • Opcode=10: An opcode allocation is outside of this document.
  • Ancillary Data: AD required to process the network action opcode 10.
  • NAL: Length of additional LSEs used to encode its AD.

Details on the third LSE:

  • Ancillary Data: 22 bits of additional AD.
  • Ancillary Data: 8 bits of additional AD.

A.1.4. Network Action Opcode with More AD with Format C ๐Ÿ’œ

Same vibe as before: a network action may need more AD than one LSE can hold, so a Format D LSE is added โœจ

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |          MNA-Label=bSPL               | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Opcode=2   |      Data=0             |R|IHS|S|NASL=2 |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Opcode=9   |      Ancillary Data           |S|   AD  |U|NAL=1|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1|            Ancillary Data                 |S|Ancillary Data |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 10: Network Action with Additional Ancillary Data

Here opcode 9 needs more than one LSE's worth of AD, so a Format D LSE is added ๐Ÿ”ฅ

Details on the third LSE:

  • Opcode=9: An opcode allocation is outside of this document.
  • Ancillary Data: Most significant bits of AD.
  • AD: 4 bits of additional AD.

Details on the fourth LSE:

  • Ancillary Data: 22 bits of additional AD.
  • Ancillary Data: 8 bits of additional AD.

A.2. Network Action Processing Order ๐ŸŒˆ

The semantics of a network action can vary a lot, and the results of processing one network action can change how a later one gets processed. See Section 5.5 ๐Ÿ’…

A.2.1. Network Action Processing Order โญ

Here's the first one, where the processing order is set by the stack itself ๐Ÿ‘‘

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |           MNA-Label=bSPL              | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Opcode=8    |      Ancillary Data     |R|IHS|S|NASL=2 |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Opcode=7    |      Ancillary Data7          |S|  AD7  |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Opcode=1    |      Flag-Based NAIs          |S|  NAI  |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 11: In-Stack NA Processing Order

In this example, opcode 8 goes first, then opcode 7, and then the network action flags get processed from most significant to least significant ๐Ÿง 

Now for a sneaky twist ๐Ÿ˜ˆ Some Flag-Based NAIs might need to run before opcode 7, and some might need to run after it. You can get that by placing some NAIs earlier in the NAS, like so:

    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
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |              MNA-Label=bSPL           | TC  |S|    TTL        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Opcode=8    |      Ancillary Data     |R|IHS|S|NASL=3 |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Opcode=1    |        0x01                   |S|  NAI  |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Opcode=7    |      Ancillary Data7          |S|  AD7  |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Opcode=1    |        0x02                   |S|  NAI  |U|NAL=0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Figure 12: Interleaving Network Actions

In this one, opcode 8 is processed first, then Flag-Based NAI 0x01, then opcode 7, and finally NAI 0x02 ๐Ÿ’–


Acknowledgments ๐Ÿซถ

The authors want to give a huge thank-you to the MPLS Working Group Open Design Team for the discussions and comments on this document. They also thank Amanda Baber for reviewing the IANA Considerations and sharing lots of useful suggestions. Loa Andersson, Stewart Bryant, Greg Mirsky, Joel M. Halpern, and Adrian Farrel get a big shoutout too, for reviewing the document and offering many useful suggestions. Fabian Ihle and Michael Menth, both from the University of Tuebingen, also deserve applause for reviewing the solution and implementing it in a P4 pipeline ๐Ÿ‘‘ Thank you to Tarek Saad for the Shepherd's review, Joe Clarke for the OpsDir review, Matthew Bocci for the Rtgdir review, Derrell Piper for the Secdir review, and James Guichard for the AD review. Mohamed Boucadair, ร‰ric Vyncke, Deb Cooley, Ketan Talaulikar, and Mahesh Jethanandani also helped make this document better through the IESG review ๐ŸŒธ


Contributors โœจ

The following people substantially contributed to this document ๐Ÿ’…

Jisu Bhattacharya Cisco Systems, Inc. Email: jisu@cisco.com

Bruno Decraene Orange Email: bruno.decraene@orange.com

Weiqiang Cheng China Mobile Email: chengweiqiang@chinamobile.com

Xiao Min ZTE Corp. Email: xiao.min2@zte.com.cn

Luay Jalil Verizon Email: luay.jalil@verizon.com

Jie Dong Huawei Technologies Huawei Campus, No. 156 Beiqing Rd. Beijing 100095 China Email: jie.dong@huawei.com

Tianran Zhou Huawei Technologies China Email: zhoutianran@huawei.com

Bin Wen Comcast Email: Bin_Wen@cable.comcast.com

Sami Boutros Ciena Email: sboutros@ciena.com

Tony Li Juniper Networks United States of America Email: tony.li@tony.li

John Drake Juniper Networks United States of America Email: jdrake@juniper.net


Authors' Addresses ๐Ÿฆ‹

Jaganbabu Rajamanickam (editor) Cisco Systems, Inc. Canada Email: jrajaman@cisco.com

Rakesh Gandhi (editor) Cisco Systems, Inc. Canada Email: rgandhi@cisco.com

Royi Zigler Broadcom Email: royi.zigler@broadcom.com

Haoyu Song Futurewei Technologies Email: haoyu.song@futurewei.com

Kireeti Kompella Juniper Networks United States of America Email: kireeti.ietf@gmail.com

๐Ÿ“ฆ where this page comes from