0                   1                   2                   3
 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 |    Type     |     Length    |             Flags             |
 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 
 |                         Router ID                           |
 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 
 |                       Interface ID                          |
 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 

Figure 9: IPv4 sub-TLV format

Type

• 0x03 Unnumbered interface

Length

• The Length contains the total length of the sub-TLV in bytes, including the Type and Length fields. The Length is always 12.

Router ID

• The Router ID interpreted as discussed in the Section 2 of RFC 3447 [RFC3477].

Interface ID

• The identifier assigned to the link by the LSR specified by the Router ID.

Flags

• TBD

5. Data Plane Theory of Operation

After instantiating an LSP for a path using RSVP-TE [RFC3209] as described in Section 4.6 or as described in the second paragraph of Section 4 and in Section 4.6, ingress LSR MAY begin sending RTM packets to the first downstream RTM capable LSR on that path. Each RTM packet has its Scratch Pad field initialized and its TTL set to expire on the next downstream RTM-capable LSR. Each RTM-capable LSR on the explicit path receives an RTM packet and records the time at which it receives that packet at its ingress interface as well as the time at which it transmits that packet from its egress interface; this should be done as close to the physical layer as possible to ensure precise accuracy in time determination. The RTM-capable LSR determines the difference between those two times; for 1-step operation, this difference is determined just prior to or while sending the packet, and the RTM-capable egress interface adds it to the value in the Scratch Pad field of the message in progress. Note, for the purpose of calculating a residence time, a common free running clock synchronizing all the involved interfaces may be sufficient, as, for example, 4.6 ppm accuracy leads to 4.6 nanosecond error for residence time on the order of 1 millisecond.

For 2-step operation, the difference between packet arrival time (at an ingress interface) and subsequent departure time (from an egress interface) is determined at some later time prior to sending a subsequent follow-up message, so that this value can be used to update the correctionField in the follow-up message.

See Section 7 for further details on the difference between 1-step and 2-step operation.

The last RTM-capable LSR on the LSP MAY then use the value in the Scratch Pad field to perform time correction, if there is no follow-up message. For example, the egress LSR may be a PTP Boundary Clock synchronized to a Master Clock and will use the value in the Scratch Pad field to update PTP's correctionField.

6. Applicable PTP Scenarios

The proposed approach can be directly integrated in a PTP network based on the IEEE 1588 delay reqest-response mechanism. The RTM capable LSR nodes act as end-to-end transparent clocks, and typically boundary clocks, at the edges of the MPLS network, use the value in the Scratch Pad field to update the correctionField of the corresponding PTP event packet prior to performing the usual PTP processing.

7. One-step Clock and Two-step Clock Modes

One-step mode refers to the mode of operation where an egress interface updates the correctionField value of an original event message. Two-step mode refers to the mode of operation where this update is made in a subsequent follow-up message.

Processing of the follow-up message, if present, requires the downstream end-point to wait for the arrival of the follow-up message in order to combine correctionField values from both the original (event) message and the subsequent (follow-up) message. In a similar fashion, each 2-step node needs to wait for the related follow-up message, if there is one, in order to update that follow-up message (as opposed to creating a new one. Hence the first node that uses 2-step mode MUST do two things:

1. Mark the original event message to indicate that a follow-up message will be forthcoming (this is necessary in order to
   • Let any subsequent 2-step node know that there is already a follow-up message, and
   • Let the end-point know to wait for a follow-up message;

2. Create a follow-up message in which to put the RTM determined as an initial correctionField value.

IEEE 1588v2 [IEEE.1588.2008] defines this behaviour for PTP messages.

Thus, for example, with reference to the PTP protocol, the PTPType field identifies whether the message is a Sync message, Follow_up message, Delay_Req message, or Delay_Resp message. The 10 octet long Port ID field contains the identity of the source port, that is, the specific PTP port of the boundary clock connected to the MPLS network. The Sequence ID is the sequence ID of the PTP message carried in the Value field of the message.

PTP messages also include a bit that indicates whether or not a follow-up message will be coming. This bit, once it is set by a 2-step mode device, MUST stay set accordingly until the original and follow-up messages are combined by an end-point (such as a Boundary Clock).

Thus, an RTM packet, containing residence time information relating to an earlier packet, also contains information identifying that earlier packet.

For compatibility with PTP, RTM (when used for PTP packets) must behave in a similar fashion. To do this, a 2-step RTM capable egress interface will need to examine the S-bit in the Flags field of the PTP sub-TLV (for RTM messages that indicate they are for PTP) and - if it is clear (set to zero), it MUST set it and create a follow-up PTP Type RTM message. If the S bit is already set, then the RTM capable node MUST wait for the RTM message with the PTP type of follow-up and matching originator and sequence number to make the corresponding residence time update to the Scratch Pad field.

In practice an RTM operating according to two-step clock behaves like a two-steps transparent clock.

A 1-step capable RTM node MAY elect to operate in either 1-step mode (by making an update to the Scratch Pad field of the RTM message containing the PTP even message), or in 2-step mode (by making an update to the Scratch Pad of a follow-up message when its presence is indicated), but MUST NOT do both.

Two main subcases can be identified for an RTM node operating as a two-step clock:

A) If any of the previous RTM capable node or the previous PTP clock (e.g. the BC connected to the first LSR), is a two-step clock, the residence time is added to the RTM packet that has been created to include the associated PTP packet (i.e. follow-up message in the downstream direction), if the local RTM-capable LSR is also operating as a two-step clock. This RTM packet carries the related accumulated residence time and the appropriate values of the Sequence Id and Port Id (the same identifiers carried in the packet processed) and the Two-step Flag set to 1.

Note that the fact that an upstream RTM-capable node operating in the two-step mode has created a follow-up message does not require any subsequent RTM capable LSR to also operate in the 2-step mode, as long as that RTM-capable LSR forwards the follow-up message on the same LSP on which it forwards the corresponding previous message.

A one-step capable RTM node MAY elect to update the RTM follow-up message as if it were operating in two-step mode, however, it MUST NOT update both messages.

A PTP event packet (sync) is carried in the RTM packet in order for an RTM node to identify that residence time measurement must be performed on that specific packet.

To handle the residence time of the Delay request message on the upstream direction, an RTM packet must be created to carry the residence time on the associated downstream Delay Resp message.

The last RTM node of the MPLS network in addition to update the correctionField of the associated PTP packet, must also properly handle the two-step flag of the PTP packets.

B) When the PTP network connected to the MPLS and RTM node, operates in one-step clock mode, the associated RTM packet must be created by the RTM node itself. The associated RTM packet including the PTP event packet needs now to indicate that a follow up message will be coming.

The last RTM node of the LSP, modeif it receives an RTM message with a PTP payload indicating a follow-up message will be forthcoming, must generate a follow-up message and properly set the two-step flag of the PTP packets.

8. IANA Considerations

8.1. New RTM G-ACh

IANA is requested to reserve a new G-ACh as follows: