Internet Engineering Task Force (IETF) G. Mirsky Request for Comments: 8169 ZTE Corp. Category: Standards Track S. Ruffini ISSN: 2070-1721 E. Gray Ericsson J. Drake Juniper Networks S. Bryant Huawei A. Vainshtein ECI Telecom May 2017
Residence Time Measurement in MPLS Networks
Abstract
This document specifies a new Generic Associated Channel (G-ACh) for Residence Time Measurement (RTM) and describes how it can be used by time synchronization protocols within an MPLS domain.
Residence time is the variable part of the propagation delay of timing and synchronization messages; knowing this delay for each message allows for a more accurate determination of the delay to be taken into account when applying the value included in a Precision Time Protocol event message.
Status of This Memo
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). Further information on Internet Standards is available in Section 2 of RFC 7841.
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc8169.
Mirsky, et al. Standards Track [Page 1]
RFC 8169 Residence Time Measurement May 2017
Copyright Notice
Copyright (c) 2017 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 (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Time synchronization protocols, e.g., the Network Time Protocol version 4 (NTPv4) [RFC5905] and the Precision Time Protocol version 2 (PTPv2) [IEEE.1588], define timing messages that can be used to synchronize clocks across a network domain. Measurement of the cumulative time that one of these timing messages spends transiting the nodes on the path from ingress node to egress node is termed "residence time" and is used to improve the accuracy of clock synchronization. Residence time is the sum of the difference between the time of receipt at an ingress interface and the time of transmission from an egress interface for each node along the network path from an ingress node to an egress node. This document defines a new Generic Associated Channel (G-ACh) value and an associated Residence Time Measurement (RTM) message that can be used in a Multiprotocol Label Switching (MPLS) network to measure residence time over a Label Switched Path (LSP).
This document describes RTM over an LSP signaled using RSVP-TE [RFC3209]. Using RSVP-TE, the LSP's path can be either explicitly specified or determined during signaling. Although it is possible to use RTM over an LSP instantiated using the Label Distribution Protocol [RFC5036], that is outside the scope of this document.
Comparison with alternative proposed solutions such as [TIMING-OVER-MPLS] is outside the scope of this document.
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.
"Packet Loss and Delay Measurement for MPLS Networks" [RFC6374] can be used to measure one-way or two-way end-to-end propagation delay over an LSP or a pseudowire (PW). But these measurements are insufficient for use in some applications, for example, time synchronization across a network as defined in the PTP. In PTPv2 [IEEE.1588], the residence time is accumulated in the correctionField of the PTP event message, which is defined in [IEEE.1588] and referred to as using a one-step clock, or in the associated follow-up message (or Delay_Resp message associated with the Delay_Req message), which is referred to as using a two-step clock (see the detailed discussion in Section 2.1).
IEEE 1588 uses this residence time to correct for the transit times of nodes on an LSP, effectively making the transit nodes transparent.
This document proposes a mechanism that can be used as one type of on-path support for a clock synchronization protocol or can be used to perform one-way measurement of residence time. The proposed mechanism accumulates residence time from all nodes that support this extension along the path of a particular LSP in the Scratch Pad field of an RTM message (Figure 1). This value can then be used by the egress node to update, for example, the correctionField of the PTP event packet carried within the RTM message prior to performing its PTP processing.
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 endpoint 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 two-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 two-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 two-step node know that there is already a follow-up message, and
* Let the endpoint 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] defines this behavior 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 [IEEE.1588], that is, the specific PTP port of the boundary clock (BC) 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 MAY be set by a two-step mode PTP device. The value MUST NOT be unset until the original and follow-up messages are combined by an endpoint (such as a BC).
For compatibility with PTP, RTM (when used for PTP packets) must behave in a similar fashion. It should be noted that the handling of Sync event messages and of Delay_Req/Delay_Resp event messages that cross a two-step RTM node is different. The following outlines the handling of a PTP Sync event message by the two-step RTM node. The details of handling Delay_Resp/Delay_Req PTP event messages by the
Mirsky, et al. Standards Track [Page 6]
RFC 8169 Residence Time Measurement May 2017
two-step RTM node are discussed in Section 2.1.1. As a summary, a two-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 the S bit 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. The wait period MUST be reasonably bounded.
Thus, an RTM packet, containing residence time information relating to an earlier packet, also contains information identifying that earlier packet.
In practice, an RTM node operating in two-step mode behaves like a two-step transparent clock.
A one-step-capable RTM node MAY elect to operate in either one-step mode (by making an update to the Scratch Pad field of the RTM message containing the PTP event message) or two-step mode (by making an update to the Scratch Pad of a follow-up message when presence of a follow-up is indicated), but it MUST NOT do both.
Two main subcases identified for an RTM node operating as a two-step clock are described in the following sub-sections.
If any of the previous RTM-capable nodes or the previous PTP clock (e.g., the BC connected to the first node) is a two-step clock and if the local RTM-capable node is also operating a two-tep clock, the residence time is added to the RTM packet that has been created to include the second PTP packet (i.e., the follow-up message in the downstream direction). This RTM packet carries the related accumulated residence time, the appropriate values of the Sequence ID and Port ID (the same identifiers carried in the original packet), and the two-step flag set to 1.
Note that the fact that an upstream RTM-capable node operating in two-step mode has created a follow-up message does not require any subsequent RTM-capable node to also operate in two-step mode, as long as that RTM-capable node 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.
Mirsky, et al. Standards Track [Page 7]
RFC 8169 Residence Time Measurement May 2017
A PTP Sync packet is carried in the RTM packet in order to indicate to the RTM node that RTM must be performed on that specific packet.
To handle the residence time of the Delay_Req message in the upstream direction, an RTM packet must be created to carry the residence time in the associated downstream Delay_Resp message.
The last RTM node of the MPLS network, in addition to updating the correctionField of the associated PTP packet, must also react properly to the two-step flag of the PTP packets.
2.1.2. Two-Step RTM with One-Step Upstream PTP Clock
When the PTP network connected to the MPLS operates in one-step clock mode and an RTM node operates in two-step mode, the follow-up RTM packet must be created by the RTM node itself. The RTM packet carrying the PTP event packet needs now to indicate that a follow-up message will be coming.
The egress RTM-capable node of the LSP will remove RTM encapsulation and, in case of two-step clock mode being indicated, will generate PTP messages to include the follow-up correction as appropriate (according to [IEEE.1588]). In this case, the common header of the PTP packet carrying the synchronization message would have to be modified by setting the twoStepFlag field indicating that there is now a follow-up message associated to the current message.
[RFC5586] and [RFC6423] define the G-ACh to extend the applicability of the Pseudowire Associated Channel Header (ACH) [RFC5085] to LSPs. G-ACh provides a mechanism to transport OAM and other control messages over an LSP. Processing of these messages by selected transit nodes is controlled by the use of the Time-to-Live (TTL) value in the MPLS header of these messages.
The message format for RTM is presented in Figure 1.
Figure 1: RTM G-ACh Message Format for Residence Time Measurement
o The first four octets are defined as a G-ACh header in [RFC5586].
o The Version field is set to 0, as defined in [RFC4385].
o The Reserved field MUST be set to 0 on transmit and ignored on receipt.
o The RTM G-ACh field (value 0x000F; see Section 7.1) identifies the packet as such.
o The Scratch Pad field is 8 octets in length. It is used to accumulate the residence time spent in each RTM-capable node transited by the packet on its path from ingress node to egress node. The first RTM-capable node MUST initialize the Scratch Pad field with its RTM. Its format is a 64-bit signed integer, and it indicates the value of the residence time measured in nanoseconds and multiplied by 2^16. Note that depending on whether the timing procedure is a one-step or two-step operation (Section 2.1), the residence time is either for the timing packet carried in the Value field of this RTM message or for an associated timing packet carried in the Value field of another RTM message.
o The Type field identifies the type and encapsulation of a timing packet carried in the Value field, e.g., NTP [RFC5905] or PTP [IEEE.1588]. Per this document, IANA has created a sub-registry called the "MPLS RTM TLV Registry" in the "Generic Associated Channel (G-ACh) Parameters" registry (see Section 7.2).
o The Length field contains the length, in octets, of any Value field defined for the Type given in the Type field.
Mirsky, et al. Standards Track [Page 9]
RFC 8169 Residence Time Measurement May 2017
o The TLV MUST be included in the RTM message, even if the length of the Value field is zero.
Figure 2 presents the format of a PTP sub-TLV that MUST be included in the Value field of an RTM message preceding the carried timing packet when the timing packet is PTP.
Figure 3: Flags Field Format of PTP Packet Sub-TLV
o The Type field identifies the PTP packet sub-TLV and is set to 1 according to Section 7.3.
o The Length field of the PTP sub-TLV contains the number of octets of the Value part of the TLV and MUST be 20.
o The Flags field currently defines one bit, the S bit, that defines whether the current message has been processed by a two-step node, where the flag is cleared if the message has been handled exclusively by one-step nodes and there is no follow-up message and is set if there has been at least one two-step node and a follow-up message is forthcoming.
Mirsky, et al. Standards Track [Page 10]
RFC 8169 Residence Time Measurement May 2017
o The PTPType field indicates the type of PTP packet to which this PTP sub-TLV applies. PTPType is the messageType field of a PTPv2 packet with possible values defined in Table 19 of [IEEE.1588].
o The 10-octet-long Port ID field contains the identity of the source port.
o The Sequence ID is the sequence ID of the PTP message to which this PTP sub-TLV applies.
A tuple of PTPType, Port ID, and Sequence ID uniquely identifies the PTP timing message included in an RTM message and is used in two-step RTM mode; see Section 2.1.1.
The Value field (see Figure 1) -- in addition to the PTP sub-TLV -- MAY carry a packet of the PTP Time synchronization protocol (as was identified by the Type field). It is important to note that the timing message packet may be authenticated or encrypted and carried over this LSP unchanged (and inaccessible to intermediate RTM capable LSRs) while the residence time is accumulated in the Scratch Pad field.
The LSP ingress RTM-capable LSR populates the identifying tuple information of the PTP sub-TLV (see section 3.1) prior to including the (possibly authenticated/encrypted) PTP message packet after the PTP sub-TLV in the Value field of the RTM message for an RTM message of the PTP Type (Type 1; see Section 7.3).
The operation of RTM depends upon TTL expiry to deliver an RTM packet from one RTM-capable interface to the next along the path from ingress node to egress node. This means that a node with RTM-capable interfaces MUST be able to compute a TTL, which will cause the expiry of an RTM packet at the next node with RTM-capable interfaces.
Note that the RTM capability of a node is with respect to the pair of interfaces that will be used to forward an RTM packet. In general, the ingress interface of this pair must be able to capture the arrival time of the packet and encode it in some way such that this information will be available to the egress interface of a node.
Mirsky, et al. Standards Track [Page 11]
RFC 8169 Residence Time Measurement May 2017
The supported mode (one-step or two-step) of any pair of interfaces is determined by the capability of the egress interface. For both modes, the egress interface implementation MUST be able to determine the precise departure time of the same packet and determine from this, and the arrival time information from the corresponding ingress interface, the difference representing the residence time for the packet.
An interface with the ability to do this and update the associated Scratch Pad in real time (i.e., while the packet is being forwarded) is said to be one-step capable.
Hence, while both ingress and egress interfaces are required to support RTM for the pair to be RTM capable, it is the egress interface that determines whether or not the node is one-step or two- step capable with respect to the interface pair.
The RTM capability used in the sub-TLV shown in Figures 4 and 5 is thus a non-routing-related capability associated with the interface being advertised based on its egress capability. The ability of any pair of interfaces on a node that includes this egress interface to support any mode of RTM depends on the ability of the ingress interface of a node to record packet arrival time and convey it to the egress interface on the node.
When a node uses an IGP to support the RTM capability advertisement, the IGP sub-TLV MUST reflect the RTM capability (one-step or two- step) associated with the advertised interface. Changes of RTM capability are unlikely to be frequent and would result, for example, from the operator's decision to include or exclude a particular port from RTM processing or switch between RTM modes.
[RFC4202] explains that the Interface Switching Capability Descriptor describes the switching capability of an interface. For bidirectional links, the switching capabilities of an interface are defined to be the same in either direction, that is, for data entering the node through that interface and for data leaving the node through that interface. That principle SHOULD be applied when a node advertises RTM capability.
A node that supports RTM MUST be able to act in two-step mode and MAY also support one-step RTM mode. A detailed discussion of one-step and two-step RTM modes is contained in Section 2.1.
Mirsky, et al. Standards Track [Page 12]
RFC 8169 Residence Time Measurement May 2017
4.3. RTM Capability Advertisement in Routing Protocols
o Type value (5) has been assigned by IANA in the "OSPFv2 Extended Link TLV Sub-TLVs" registry (see Section 7.4).
o Length value equals the number of octets of the Value field.
o Value contains a variable number of bitmap fields so that the overall number of bits in the fields equals Length * 8.
o Bits are defined/sent starting with Bit 0. Additional bitmap field definitions that may be defined in the future SHOULD be assigned in ascending bit order so as to minimize the number of bits that will need to be transmitted.
o Undefined bits MUST be transmitted as 0 and MUST be ignored on receipt.
o Bits that are NOT transmitted MUST be treated as if they are set to 0 on receipt.
o RTM (capability) is a 3-bit-long bitmap field with values defined as follows:
* 0b001 - one-step RTM supported
* 0b010 - two-step RTM supported
* 0b100 - reserved
The capability to support RTM on a particular link (interface) is advertised in the OSPFv2 Extended Link Opaque LSA as described in Section 3 of [RFC7684] via the RTM Capability sub-TLV.
The capability to support RTM on a particular link (interface) can be advertised in OSPFv3 using LSA extensions as described in [OSPFV3-EXTENDED-LSA]. The sub-TLV SHOULD use the same format as in Section 4.3.1. The type allocation and full details of exact use of OSPFv3 LSA extensions is for further study.
The capability to support RTM on a particular link (interface) is advertised in a new sub-TLV that may be included in TLVs advertising Intermediate System (IS) Reachability on a specific link (TLVs 22, 23, 222, and 223).
The format for the RTM Capability sub-TLV is presented in Figure 5.
The format for the RTM Capability TLV is presented in Figure 4.
Type value (1105) has been assigned by IANA in the "BGP-LS Node Descriptor, Link Descriptor, Prefix Descriptor, and Attribute TLVs" sub-registry (see Section 7.6).
Definitions, rules of handling, and values for fields Length, Value, and RTM are as defined in Section 4.3.1.
Mirsky, et al. Standards Track [Page 14]
RFC 8169 Residence Time Measurement May 2017
The RTM capability will be advertised in BGP-LS as a Link Attribute TLV associated with the Link NLRI as described in Section 3.3.2 of [RFC7752].
4.4. RSVP-TE Control-Plane Operation to Support RTM
Throughout this document, we refer to a node as an RTM-capable node when at least one of its interfaces is RTM capable. Figure 6 provides an example of roles a node may have with respect to RTM capability:
----- ----- ----- ----- ----- ----- ----- | A |-----| B |-----| C |-----| D |-----| E |-----| F |-----| G | ----- ----- ----- ----- ----- ----- -----
Figure 6: RTM-Capable Roles
o A is a boundary clock with its egress port in Master state. Node A transmits IP-encapsulated timing packets whose destination IP address is G.
o B is the ingress Label Edge Router (LER) for the MPLS LSP and is the first RTM-capable node. It creates RTM packets, and in each it places a timing packet, possibly encrypted, in the Value field and initializes the Scratch Pad field with its RTM.
o C is a transit node that is not RTM capable. It forwards RTM packets without modification.
o D is an RTM-capable transit node. It updates the Scratch Pad field of the RTM packet without updating the timing packet.
o E is a transit node that is not RTM capable. It forwards RTM packets without modification.
o F is the egress LER and the last RTM-capable node. It removes the RTM ACH encapsulation and processes the timing packet carried in the Value field using the value in the Scratch Pad field. In particular, the value in the Scratch Pad field of the RTM ACH is used in updating the Correction field of the PTP message(s). The LER should also include its own residence time before creating the outgoing PTP packets. The details of this process depend on whether or not the node F is itself operating as a one-step or two-step clock.
o G is a boundary clock with its ingress port in Slave state. Node G receives PTP messages.
Mirsky, et al. Standards Track [Page 15]
RFC 8169 Residence Time Measurement May 2017
An ingress node that is configured to perform RTM along a path through an MPLS network to an egress node MUST verify that the selected egress node has an interface that supports RTM via the egress node's advertisement of the RTM Capability sub-TLV, as covered in Section 4.3. In the Path message that the ingress node uses to instantiate the LSP to that egress node, it places an LSP_ATTRIBUTES object [RFC5420] with an RTM_SET Attribute Flag set, as described in Section 7.8, which indicates to the egress node that RTM is requested for this LSP. The RTM_SET Attribute Flag SHOULD NOT be set in the LSP_REQUIRED_ATTRIBUTES object [RFC5420], unless it is known that all nodes recognize the RTM attribute (but need not necessarily implement it), because a node that does not recognize the RTM_SET Attribute Flag would reject the Path message.
If an egress node receives a Path message with the RTM_SET Attribute Flag in an LSP_ATTRIBUTES object, the egress node MUST include an initialized RRO [RFC3209] and LSP_ATTRIBUTES object where the RTM_SET Attribute Flag is set and the RTM_SET TLV (Section 4.4.1) is initialized. When the Resv message is received by the ingress node, the RTM_SET TLV will contain an ordered list, from egress node to ingress node, of the RTM-capable nodes along the LSP's path.
After the ingress node receives the Resv, it MAY begin sending RTM packets on the LSP's path. Each RTM packet has its Scratch Pad field initialized and its TTL set to expire on the closest downstream RTM- capable node.
It should be noted that RTM can also be used for LSPs instantiated using [RFC3209] in an environment in which all interfaces in an IGP support RTM. In this case, the RTM_SET TLV and LSP_ATTRIBUTES object MAY be omitted.
Type value (5) has been assigned by IANA in the RSVP-TE "Attributes TLV Space" sub-registry (see Section 7.7).
The Length contains the total length of the sub-object in bytes, including the Type and Length fields.
The I bit indicates whether the downstream RTM-capable node along the LSP is present in the RRO.
The Reserved field must be zeroed on initiation and ignored on receipt.
The content of an RTM_SET TLV is a series of variable-length sub-TLVs. Only a single RTM_SET can be present in a given LSP_ATTRIBUTES object. The sub-TLVs are defined in Section 4.4.1.1.
The following processing procedures apply to every RTM-capable node along the LSP. In this paragraph, an RTM-capable node is referred to as a node for sake of brevity. Each node MUST examine the Resv message for whether the RTM_SET Attribute Flag in the LSP_ATTRIBUTES object is set. If the RTM_SET flag is set, the node MUST inspect the LSP_ATTRIBUTES object for presence of an RTM_SET TLV. If more than one is found, then the LSP setup MUST fail with generation of the ResvErr message with Error Code "Duplicate TLV" (Section 7.9) and Error Value that contains the Type value in its 8 least significant bits. If no RTM_SET TLV is found, then the LSP setup MUST fail with generation of the ResvErr message with Error Code "RTM_SET TLV Absent" (Section 7.9). If one RTM_SET TLV has been found, the node will use the ID of the first node in the RTM_SET in conjunction with the RRO to compute the hop count to its downstream node with a reachable RTM-capable interface. If the node cannot find a matching ID in the RRO, then it MUST try to use the ID of the next node in the RTM_SET until it finds the match or reaches the end of the RTM_SET TLV. If a match has been found, the calculated value is used by the node as the TTL value in the outgoing label to reach the next RTM- capable node on the LSP. Otherwise, the TTL value MUST be set to 255. The node MUST add an RTM_SET sub-TLV with the same address it used in the RRO sub-object at the beginning of the RTM_SET TLV in the associated outgoing Resv message before forwarding it upstream. If the calculated TTL value has been set to 255, as described above, then the I flag in the node's RTM_SET TLV MUST be set to 1 before the Resv message is forwarded upstream. Otherwise, the I flag MUST be cleared (0).
The ingress node MAY inspect the I bit received in each RTM_SET TLV contained in the LSP_ATTRIBUTES object of a received Resv message. The presence of the RTM_SET TLV with the I bit set to 1 indicates that some RTM nodes along the LSP could not be included in the
Mirsky, et al. Standards Track [Page 17]
RFC 8169 Residence Time Measurement May 2017
calculation of the residence time. An ingress node MAY choose to resignal the LSP to include all RTM nodes or simply notify the user via a management interface.
There are scenarios when some information is removed from an RRO due to policy processing (e.g., as may happen between providers) or the RRO is limited due to size constraints. Such changes affect the core assumption of this method and the processing of RTM packets. RTM SHOULD NOT be used if it is not guaranteed that the RRO contains complete information.
The RTM Set sub-object contains an ordered list, from egress node to ingress node, of the RTM-capable nodes along the LSP's path.
The contents of an RTM_SET sub-object are a series of variable-length sub-TLVs. Each sub-TLV has its own Length field. The Length contains the total length of the sub-TLV in bytes, including the Type and Length fields. The Length MUST always be a multiple of 4, and at least 8 (smallest IPv4 sub-object).
Sub-TLVs are organized as a last-in-first-out stack. The first-out sub-TLV relative to the beginning of RTM_SET TLV is considered the top. The last-out sub-TLV is considered the bottom. When a new sub-TLV is added, it is always added to the top.
The RTM_SET TLV is intended to include the subset of the RRO sub-TLVs that represent those egress interfaces on the LSP that are RTM capable. After a node chooses an egress interface to use in the RRO sub-TLV, that same egress interface, if RTM capable, SHOULD be placed into the RTM_SET TLV using one of the following: IPv4 sub-TLV, IPv6 sub-TLV, or Unnumbered Interface sub-TLV. The address family chosen SHOULD match that of the RESV message and that used in the RRO; the unnumbered interface sub-TLV is used when the egress interface has no assigned IP address. A node MUST NOT place more sub-TLVs in the RTM_SET TLV than the number of RTM-capable egress interfaces the LSP traverses that are under that node's control. Only a single RTM_SET sub-TLV with the given Value field MUST be present in the RTM_SET TLV. If more than one sub-TLV with the same value (e.g., a duplicated address) is found, the LSP setup MUST fail with the generation of a ResvErr message with the Error Code "Duplicate sub-TLV" (Section 7.9) and the Error Value containing a 16-bit value composed of (Type of TLV, Type of sub-TLV).
Three kinds of sub-TLVs for RTM_SET are currently defined.
After instantiating an LSP for a path using RSVP-TE [RFC3209] as described in Section 4.4, the ingress node MAY begin sending RTM packets to the first downstream RTM-capable node on that path. Each RTM packet has its Scratch Pad field initialized and its TTL set to expire on the next downstream RTM-capable node. Each RTM-capable node 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.
Mirsky, et al. Standards Track [Page 20]
RFC 8169 Residence Time Measurement May 2017
These actions should be done as close to the physical layer as possible at the same point of packet processing, striving to avoid introducing the appearance of jitter in propagation delay whereas it should be accounted as residence time. The RTM-capable node determines the difference between those two times; for one-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 a 4.6 nanosecond error for residence time on the order of 1 millisecond. This may be acceptable for applications where the target accuracy is in the order of hundreds of nanoseconds. As an example, several applications being considered in the area of wireless applications are satisfied with an accuracy of 1.5 microseconds [ITU-T.G.8271].
For two-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 2.1 for further details on the difference between one- step and two-step operation.
The last RTM-capable node 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 node 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.
This approach can be directly integrated in a PTP network based on the IEEE 1588 delay request-response mechanism. The RTM-capable nodes act as end-to-end transparent clocks, and boundary clocks, at the edges of the MPLS network, typically 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.
IANA has created a sub-registry in the "Generic Associated Channel (G-ACh) Parameters" registry called the "MPLS RTM TLV Registry". All codepoints in the range 0 through 127 in this registry shall be allocated according to the "IETF Review" procedure as specified in [RFC5226]. Codepoints in the range 128 through 191 in this registry shall be allocated according to the "First Come First Served" procedure as specified in [RFC5226]. This document defines the following new RTM TLV types:
+---------+-------------------------------+---------------+ | Value | Description | Reference | +---------+-------------------------------+---------------+ | 0 | Reserved | This document | | 1 | No payload | This document | | 2 | PTPv2, Ethernet encapsulation | This document | | 3 | PTPv2, IPv4 encapsulation | This document | | 4 | PTPv2, IPv6 encapsulation | This document | | 5 | NTP | This document | | 6-191 | Unassigned | | | 192-254 | Reserved for Private Use | This document | | 255 | Reserved | This document | +---------+-------------------------------+---------------+
IANA has created a sub-registry in the "MPLS RTM TLV Registry" (see Section 7.2) called the "MPLS RTM Sub-TLV Registry". All codepoints in the range 0 through 127 in this registry shall be allocated according to the "IETF Review" procedure as specified in [RFC5226]. Codepoints in the range 128 through 191 in this registry shall be allocated according to the "First Come First Served" procedure as specified in [RFC5226]. This document defines the following new RTM sub-TLV types:
+---------+--------------------------+---------------+ | Value | Description | Reference | +---------+--------------------------+---------------+ | 0 | Reserved | This document | | 1 | PTP | This document | | 2-191 | Unassigned | | | 192-254 | Reserved for Private Use | This document | | 255 | Reserved | This document | +---------+--------------------------+---------------+
IANA has assigned a new codepoint for the RTM Capability TLV from the "BGP-LS Node Descriptor, Link Descriptor, Prefix Descriptor, and Attribute TLVs" sub-registry in the "Border Gateway Protocol - Link State (BGP-LS) Parameters" registry as follows:
IANA has assigned a new type for the RTM_SET sub-object from the RSVP-TE "Attributes TLV Space" sub-registry as follows:
+------+------------+-----------+---------------+-----------+----------+ | Type | Name | Allowed | Allowed on | Allowed | Reference| | | | on LSP_ | LSP_REQUIRED_ | on LSP | | | | | ATTRIBUTES| ATTRIBUTES | Hop | | | | | | | Attributes| | +------+------------+-----------+---------------+-----------+----------+ | 5 | RTM_SET | Yes | No | No | This | | | sub-object | | | | document | +------+------------+-----------+---------------+-----------+----------+
Table 7: RTM_SET Sub-object Type
IANA has created a new sub-registry for sub-TLV types of the RTM_SET sub-object called the "RTM_SET Object Sub-Object Types" registry. All codepoints in the range 0 through 127 in this registry shall be allocated according to the "IETF Review" procedure as specified in [RFC5226]. Codepoints in the range 128 through 191 in this registry shall be allocated according to the "First Come First Served" procedure as specified in [RFC5226]. This document defines the following new values of RTM_SET object sub-object types:
+---------+--------------------------+---------------+ | Value | Description | Reference | +---------+--------------------------+---------------+ | 0 | Reserved | This document | | 1 | IPv4 address | This document | | 2 | IPv6 address | This document | | 3 | Unnumbered interface | This document | | 4-191 | Unassigned | | | 192-254 | Reserved for Private Use | This document | | 255 | Reserved | This document | +---------+--------------------------+---------------+
Routers that support RTM are subject to the same security considerations as defined in [RFC4385] and [RFC5085].
In addition -- particularly as applied to use related to PTP -- there is a presumed trust model that depends on the existence of a trusted relationship of at least all PTP-aware nodes on the path traversed by PTP messages. This is necessary as these nodes are expected to correctly modify specific content of the data in PTP messages, and proper operation of the protocol depends on this ability. In practice, this means that those portions of messages cannot be covered by either confidentiality or integrity protection. Though there are methods that make it possible in theory to provide either or both such protections and still allow for intermediate nodes to make detectable but authenticated modifications, such methods do not seem practical at present, particularly for timing protocols that are sensitive to latency and/or jitter.
Mirsky, et al. Standards Track [Page 26]
RFC 8169 Residence Time Measurement May 2017
The ability to potentially authenticate and/or encrypt RTM and PTP data for scenarios both with and without participation of intermediate RTM-/PTP-capable nodes is left for further study.
While it is possible for a supposed compromised node to intercept and modify the G-ACh content, this is an issue that exists for nodes in general -- for any and all data that may be carried over an LSP -- and is therefore the basis for an additional presumed trust model associated with existing LSPs and nodes.
Security requirements of time protocols are provided in RFC 7384 [RFC7384].
[IEEE.1588] IEEE, "IEEE Standard for a Precision Clock Synchronization Protocol for Networked Measurement and Control Systems", IEEE Std 1588-2008, DOI 10.1109/IEEESTD.2008.4579760.
[RFC5420] Farrel, A., Ed., Papadimitriou, D., Vasseur, JP., and A. Ayyangarps, "Encoding of Attributes for MPLS LSP Establishment Using Resource Reservation Protocol Traffic Engineering (RSVP-TE)", RFC 5420, DOI 10.17487/RFC5420, February 2009, <http://www.rfc-editor.org/info/rfc5420>.
[RFC6423] Li, H., Martini, L., He, J., and F. Huang, "Using the Generic Associated Channel Label for Pseudowire in the MPLS Transport Profile (MPLS-TP)", RFC 6423, DOI 10.17487/RFC6423, November 2011, <http://www.rfc-editor.org/info/rfc6423>.
[RFC7752] Gredler, H., Ed., Medved, J., Previdi, S., Farrel, A., and S. Ray, "North-Bound Distribution of Link-State and Traffic Engineering (TE) Information Using BGP", RFC 7752, DOI 10.17487/RFC7752, March 2016, <http://www.rfc-editor.org/info/rfc7752>.
[ITU-T.G.8271] ITU-T, "Time and phase synchronization aspects of packet networks", ITU-T Recomendation G.8271/Y.1366, July 2016.
[OSPFV3-EXTENDED-LSA] Lindem, A., Roy, A., Goethals, D., Vallem, V., and F. Baker, "OSPFv3 LSA Extendibility", Work in Progress, draft-ietf-ospf-ospfv3-lsa-extend-14, April 2017.
[TIMING-OVER-MPLS] Davari, S., Oren, A., Bhatia, M., Roberts, P., and L. Montini, "Transporting Timing messages over MPLS Networks", Work in Progress, draft-ietf-tictoc- 1588overmpls-07, October 2015.
Acknowledgments
The authors want to thank Loa Andersson, Lou Berger, Acee Lindem, Les Ginsberg, and Uma Chunduri for their thorough reviews, thoughtful comments, and, most of all, patience.