3.  OAM Functions

   This subsection provides a brief summary of the common OAM functions
   used in OAM-related standards.  These functions are used as building
   blocks in the OAM standards described in this document.

   o  Connectivity Verification (CV), Path Verification, and Continuity
      Check (CC):
      As defined in Section 2.2.7.

   o  Path Discovery / Fault Localization:
      This function can be used to trace the route to a destination,
      i.e., to identify the nodes along the route to the destination.
      When more than one route is available to a specific destination,
      this function traces one of the available routes.  When a failure
      occurs, this function attempts to detect the location of the
      failure.
      Note that the term "route tracing" (or "Traceroute"), which is
      used in the context of IP and MPLS, is sometimes referred to as
      "path tracing" in the context of other protocols, such as TRILL.

   o  Performance Monitoring:
      Typically refers to:

      *  Loss Measurement (LM) - monitors the packet loss rate.

      *  Delay Measurement (DM) - monitors the delay and delay variation
         (jitter).

4.  OAM Tools in the IETF - A Detailed Description

   This section presents a detailed description of the sets of OAM-
   related tools in each of the toolsets in Table 1.

4.1.  IP Ping

   Ping is a common network diagnostic application for IP networks that
   use ICMP.  According to [NetTerms], 'Ping' is an abbreviation for
   Packet internet groper, although the term has been so commonly used
   that it stands on its own.  As defined in [NetTerms], it is a program
   used to test reachability of destinations by sending them an ICMP
   Echo request and waiting for a reply.

   The ICMP Echo request/reply exchange in Ping is used as a Continuity
   Check function for the Internet Protocol.  The originator transmits
   an ICMP Echo request packet, and the receiver replies with an Echo
   reply.  ICMP Ping is defined in two variants: [ICMPv4] is used for
   IPv4, and [ICMPv6] is used for IPv6.

   Ping can be invoked to either a unicast destination or a multicast
   destination.  In the latter case, all members of the multicast group
   send an Echo reply back to the originator.

   Ping implementations typically use ICMP messages.  UDP Ping is a
   variant that uses UDP messages instead of ICMP Echo messages.

   Ping is a single-ended Continuity Check, i.e., it allows the
   *initiator* of the Echo request to test the reachability.  If it is
   desirable for both ends to test the reachability, both ends have to
   invoke Ping independently.

   Note that since ICMP filtering is deployed in some routers and
   firewalls, the usefulness of Ping is sometimes limited in the wider
   Internet.  This limitation is equally relevant to Traceroute.

4.2.  IP Traceroute

   Traceroute ([TCPIP-Tools], [NetTools]) is an application that allows
   users to discover a path between an IP source and an IP destination.

   The most common way to implement Traceroute [TCPIP-Tools] is
   described as follows.  Traceroute sends a sequence of UDP packets to
   UDP port 33434 at the destination.  By default, Traceroute begins by
   sending three packets (the number of packets is configurable in most
   Traceroute implementations), each with an IP Time-To-Live (or Hop
   Limit in IPv6) value of one, to the destination.  These packets
   expire as soon as they reach the first router in the path.
   Consequently, that router sends three ICMP Time Exceeded Messages
   back to the Traceroute application.  Traceroute now sends another
   three UDP packets, each with the TTL value of 2.  These messages
   cause the second router to return ICMP messages.  This process
   continues, with ever-increasing values for the TTL field, until the
   packets actually reach the destination.  Because no application
   listens to port 33434 at the destination, the destination returns
   ICMP Destination Unreachable Messages indicating an unreachable port.
   This event indicates to the Traceroute application that it is
   finished.  The Traceroute program displays the round-trip delay
   associated with each of the attempts.

   While Traceroute is a tool that finds *a* path from A to B, it should
   be noted that traffic from A to B is often forwarded through Equal-
   Cost Multipaths (ECMPs).  Paris Traceroute [PARIS] is an extension to
   Traceroute that attempts to discovers all the available paths from A
   to B by scanning different values of header fields (such as UDP
   ports) in the probe packets.

   It is noted that Traceroute is an application, and not a protocol.
   As such, it has various different implementations.  One of the most
   common ones uses UDP probe packets, as described above.  Other
   implementations exist that use other types of probe messages, such as
   ICMP or TCP.

   Note that IP routing may be asymmetric.  While Traceroute discovers a
   path between a source and destination, it does not reveal the reverse
   path.

   A few ICMP extensions ([ICMP-MP], [ICMP-Int]) have been defined in
   the context of Traceroute.  These documents define several
   extensions, including extensions to the ICMP Destination Unreachable
   message, that can be used by Traceroute applications.

   Traceroute allows path discovery to *unicast* destination addresses.
   A similar tool [mtrace] was defined for multicast destination
   addresses; it allows tracing the route that a multicast IP packet
   takes from a source to a particular receiver.

4.3.  Bidirectional Forwarding Detection (BFD)

4.3.1.  Overview

   While multiple OAM tools have been defined for various protocols in
   the protocol stack, Bidirectional Forwarding Detection [BFD], defined
   by the IETF BFD working group, is a generic OAM tool that can be
   deployed over various encapsulating protocols, and in various medium
   types.  The IETF has defined variants of the protocol for IP
   ([BFD-IP], [BFD-Multi]), for MPLS LSPs [BFD-LSP], and for pseudowires
   [BFD-VCCV].  The usage of BFD in MPLS-TP is defined in [TP-CC-CV].

   BFD includes two main OAM functions, using two types of BFD packets:
   BFD Control packets and BFD Echo packets.

4.3.2.  Terminology

   BFD operates between *systems*.  The BFD protocol is run between two
   or more systems after establishing a *session*.

4.3.3.  BFD Control

   BFD supports a bidirectional Continuity Check, using BFD Control
   packets that are exchanged within a BFD session.  BFD sessions
   operate in one of two modes:

   o  Asynchronous mode (i.e., proactive): in this mode, BFD Control
      packets are sent periodically.  When the receiver detects that no
      BFD Control packets have been received during a predetermined
      period of time, a failure is reported.

   o  Demand mode: in this mode, BFD Control packets are sent on demand.
      Upon need, a system initiates a series of BFD Control packets to
      check the continuity of the session.  BFD Control packets are sent
      independently in each direction.

   Each of the endpoints (referred to as systems) of the monitored path
   maintains its own session identification, called a Discriminator;
   both Discriminators are included in the BFD Control Packets that are
   exchanged between the endpoints.  At the time of session
   establishment, the Discriminators are exchanged between the two
   endpoints.  In addition, the transmission (and reception) rate is

   negotiated between the two endpoints, based on information included
   in the control packets.  These transmission rates may be renegotiated
   during the session.

   During normal operation of the session, i.e., when no failures have
   been detected, the BFD session is in the Up state.  If no BFD Control
   packets are received during a period of time called the Detection
   Time, the session is declared to be Down.  The detection time is a
   function of the pre-configured or negotiated transmission rate and a
   parameter called Detect Mult.  Detect Mult determines the number of
   missing BFD Control packets that cause the session to be declared as
   Down.  This parameter is included in the BFD Control packet.

4.3.4.  BFD Echo

   A BFD Echo packet is sent to a peer system and is looped back to the
   originator.  The echo function can be used proactively or on demand.

   The BFD Echo function has been defined in BFD for IPv4 and IPv6
   ([BFD-IP]), but it is not used in BFD for MPLS LSPs or PWs, or in BFD
   for MPLS-TP.

4.4.  MPLS OAM

   The IETF MPLS working group has defined OAM for MPLS LSPs.  The
   requirements and framework of this effort are defined in
   [MPLS-OAM-FW] and [MPLS-OAM], respectively.  The corresponding OAM
   tool defined, in this context, is LSP Ping [LSP-Ping].  OAM for P2MP
   services is defined in [MPLS-P2MP].

   BFD for MPLS [BFD-LSP] is an alternative means for detecting data-
   plane failures, as described below.

4.4.1.  LSP Ping

   LSP Ping is modeled after the Ping/Traceroute paradigm, and thus it
   may be used in one of two modes:

   o  "Ping" mode: In this mode, LSP Ping is used for end-to-end
      Connectivity Verification between two LERs.

   o  "Traceroute" mode: This mode is used for hop-by-hop fault
      isolation.

   LSP Ping is based on the ICMP Ping operation (of data-plane
   Connectivity Verification) with additional functionality to verify
   data-plane vs. control-plane consistency for a Forwarding Equivalence
   Class (FEC) and also to identify Maximum Transmission Unit (MTU)
   problems.

   The Traceroute functionality may be used to isolate and localize MPLS
   faults, using the Time-To-Live (TTL) indicator to incrementally
   identify the sub-path of the LSP that is successfully traversed
   before the faulty link or node.

   The challenge in MPLS networks is that the traffic of a given LSP may
   be load-balanced across Equal-Cost Multipaths (ECMPs).  LSP Ping
   monitors all the available paths of an LSP by monitoring its
   different FECs.  Note that MPLS-TP does not use ECMP, and thus does
   not require OAM over multiple paths.

   Another challenge is that an MPLS LSP does not necessarily have a
   return path; traffic that is sent back from the egress LSR to the
   ingress LSR is not necessarily sent over an MPLS LSP, but it can be
   sent through a different route, such as an IP route.  Thus,
   responding to an LSP Ping message is not necessarily as trivial as in
   IP Ping, where the responder just swaps the source and destination IP
   addresses.  Note that this challenge is not applicable to MPLS-TP,
   where a return path is always available.

   It should be noted that LSP Ping supports unique identification of
   the LSP within an addressing domain.  The identification is checked
   using the full FEC identification.  LSP Ping is extensible to include
   additional information needed to support new functionality, by use of
   Type-Length-Value (TLV) constructs.  The usage of TLVs is typically
   handled by the control plane, as it is not easy to implement in
   hardware.

   LSP Ping supports both asynchronous and on-demand activation.

4.4.2.  BFD for MPLS

   BFD [BFD-LSP] can be used to detect MPLS LSP data-plane failures.

   A BFD session is established for each MPLS LSP that is being
   monitored.  BFD Control packets must be sent along the same path as
   the monitored LSP.  If the LSP is associated with multiple FECs, a
   BFD session is established for each FEC.

   While LSP Ping can be used for detecting MPLS data-plane failures and
   for verifying the MPLS LSP data plane against the control plane, BFD
   can only be used for the former.  BFD can be used in conjunction with
   LSP Ping, as is the case in MPLS-TP (see Section 4.5.4).

4.4.3.  OAM for Virtual Private Networks (VPNs) over MPLS

   The IETF has defined two classes of VPNs: Layer 2 VPNs (L2VPNs) and
   Layer 3 VPNs (L3VPNs).  [L2VPN-OAM] provides the requirements and
   framework for OAM in the context of L2VPNs, and specifically it also
   defines the OAM layering of L2VPNs over MPLS.  [L3VPN-OAM] provides a
   framework for the operation and management of L3VPNs.

4.5.  MPLS-TP OAM

4.5.1.  Overview

   The MPLS working group has defined the OAM toolset that fulfills the
   requirements for MPLS-TP OAM.  The full set of requirements for
   MPLS-TP OAM are defined in [MPLS-TP-OAM] and include both general
   requirements for the behavior of the OAM tools and a set of
   operations that should be supported by the OAM toolset.  The set of
   mechanisms required are further elaborated in [TP-OAM-FW], which
   describes the general architecture of the OAM system and also gives
   overviews of the functionality of the OAM toolset.

   Some of the basic requirements for the OAM toolset for MPLS-TP are:

   o  MPLS-TP OAM must be able to support both an IP-based environment
      and a non-IP-based environment.  If the network is IP based, i.e.,
      IP routing and forwarding are available, then the MPLS-TP OAM
      toolset should rely on the IP routing and forwarding capabilities.
      On the other hand, in environments where IP functionality is not
      available, the OAM tools must still be able to operate without
      dependence on IP forwarding and routing.

   o  OAM packets and the user traffic are required to be congruent
      (i.e., OAM packets are transmitted in-band), and there is a need
      to differentiate OAM packets from ordinary user packets in the
      data plane.  Inherent in this requirement is the principle that
      MPLS-TP OAM be independent of any existing control plane, although
      it should not preclude use of the control-plane functionality.
      OAM packets are identified by the Generic Associated Channel Label
      (GAL), which is a reserved MPLS label value (13).

4.5.2.  Terminology

   Maintenance Entity (ME)

      The MPLS-TP OAM tools are designed to monitor and manage a
      Maintenance Entity (ME).  An ME, as defined in [TP-OAM-FW],
      defines a relationship between two points of a transport path to
      which maintenance and monitoring operations apply.

      The term "Maintenance Entity (ME)" is used in ITU-T
      Recommendations (e.g., [ITU-T-Y1731]), as well as in the MPLS-TP
      terminology ([TP-OAM-FW]).

   Maintenance Entity Group (MEG)

      The collection of one or more MEs that belong to the same
      transport path and that are maintained and monitored as a group
      are known as a Maintenance Entity Group (based on [TP-OAM-FW]).

   Maintenance Point (MP)

      A Maintenance Point (MP) is a functional entity that is defined at
      a node in the network and can initiate and/or react to OAM
      messages.  This document focuses on the data-plane functionality
      of MPs, while MPs interact with the control plane and with the
      management plane as well.

      The term "MP" is used in IEEE 802.1ag and was similarly adopted in
      MPLS-TP ([TP-OAM-FW]).

   MEG End Point (MEP)

      A MEG End Point (MEP) is one of the endpoints of an ME, and can
      initiate OAM messages and respond to them (based on [TP-OAM-FW]).

   MEG Intermediate Point (MIP)

      In between MEPs, there are zero or more intermediate points,
      called MEG Intermediate Points  (based on [TP-OAM-FW]).

      A MEG Intermediate Point (MIP) is an intermediate point that does
      not generally initiate OAM frames (one exception to this is the
      use of AIS notifications) but is able to respond to OAM frames
      that are destined to it.  A MIP in MPLS-TP identifies OAM packets
      destined to it by the expiration of the TTL field in the OAM
      packet.  The term "Maintenance Point" is a general term for MEPs
      and MIPs.

   Up and Down MEPs

      IEEE 802.1ag [IEEE802.1Q] defines a distinction between Up MEPs
      and Down MEPs.  A MEP monitors traffic in either the direction
      facing the network or the direction facing the bridge.  A Down MEP
      is a MEP that receives OAM packets from and transmits them to the
      direction of the network.  An Up MEP receives OAM packets from and
      transmits them to the direction of the bridging entity.  MPLS-TP
      ([TP-OAM-FW]) uses a similar distinction on the placement of the
      MEP -- at either the ingress, egress, or forwarding function of
      the node (Down / Up MEPs).  This placement is important for
      localization of a failure.

      Note that the terms "Up MEP" and "Down MEP" are entirely unrelated
      to the conventional "Up"/"Down" terminology, where "Down" means
      faulty and "Up" means not faulty.

      The distinction between Up and Down MEPs was defined in
      [TP-OAM-FW], but has not been used in other MPLS-TP RFCs, as of
      the writing of this document.

4.5.3.  Generic Associated Channel

   In order to address the requirement for in-band transmission of
   MPLS-TP OAM traffic, MPLS-TP uses a Generic Associated Channel
   (G-ACh), defined in [G-ACh] for LSP-based OAM traffic.  This
   mechanism is based on the same concepts as the PWE3 ACH [PW-ACH] and
   VCCV [VCCV] mechanisms.  However, to address the needs of LSPs as
   differentiated from PW, the following concepts were defined for
   [G-ACh]:

   o  An Associated Channel Header (ACH), which uses a format similar to
      the PW Control Word [PW-ACH], is a 4-byte header that is prepended
      to OAM packets.

   o  A Generic Associated Channel Label (GAL).  The GAL is a reserved
      MPLS label value (13) that indicates that the packet is an ACH
      packet and the payload follows immediately after the label stack.

   It should be noted that while the G-ACh was defined as part of the
   MPLS-TP definition effort, the G-ACh is a generic tool that can be
   used in MPLS in general, and not only in MPLS-TP.

4.5.4.  MPLS-TP OAM Toolset

   To address the functionality that is required of the OAM toolset, the
   MPLS WG conducted an analysis of the existing IETF and ITU-T OAM
   tools and their ability to fulfill the required functionality.  The

   conclusions of this analysis are documented in [OAM-Analys].  MPLS-TP
   uses a mixture of OAM tools that are based on previous standards and
   adapted to the requirements of [MPLS-TP-OAM].  Some of the main
   building blocks of this solution are based on:

   o  Bidirectional Forwarding Detection ([BFD], [BFD-LSP]) for
      proactive Continuity Check and Connectivity Verification.

   o  LSP Ping as defined in [LSP-Ping] for on-demand Connectivity
      Verification.

   o  New protocol packets, using G-ACH, to address different
      functionality.

   o  Performance measurement protocols.

   The following subsections describe the OAM tools defined for MPLS-TP
   as described in [TP-OAM-FW].

4.5.4.1.  Continuity Check and Connectivity Verification

   Continuity Checks and Connectivity Verification are presented in
   Section 2.2.7 of this document.  As presented there, these tools may
   be used either proactively or on demand.  When using these tools
   proactively, they are generally used in tandem.

   For MPLS-TP there are two distinct tools: the proactive tool is
   defined in [TP-CC-CV], while the on-demand tool is defined in
   [OnDemand-CV].  In on-demand mode, this function should support
   monitoring between the MEPs and, in addition, between a MEP and MIP.
   [TP-OAM-FW] highlights, when performing Connectivity Verification,
   the need for the CC-V messages to include unique identification of
   the MEG that is being monitored and the MEP that originated the
   message.

   The proactive tool [TP-CC-CV] is based on extensions to BFD (see
   Section 4.3) with the additional limitation that the transmission and
   receiving rates are based on configuration by the operator.  The
   on-demand tool [OnDemand-CV] is an adaptation of LSP Ping (see
   Section 4.4.1) for the required behavior of MPLS-TP.

4.5.4.2.  Route Tracing

   [MPLS-TP-OAM] defines that there is a need for functionality that
   would allow a path endpoint to identify the intermediate and
   endpoints of the path.  This function would be used in on-demand
   mode.  Normally, this path will be used for bidirectional PW, LSP,

   and Sections; however, unidirectional paths may be supported only if
   a return path exists.  The tool for this is based on the LSP Ping
   (see Section 4.4.1) functionality and is described in [OnDemand-CV].

4.5.4.3.  Lock Instruct

   The Lock Instruct function [Lock-Loop] is used to notify a transport-
   path endpoint of an administrative need to disable the transport
   path.  This functionality will generally be used in conjunction with
   some intrusive OAM function, e.g., performance measurement or
   diagnostic testing, to minimize the side-effect on user data traffic.

4.5.4.4.  Lock Reporting

   Lock Reporting is a function used by an endpoint of a path to report
   to its far-end endpoint that a lock condition has been affected on
   the path.

4.5.4.5.  Alarm Reporting

   Alarm reporting [TP-Fault] provides the means to suppress alarms
   following detection of defect conditions at the server sub-layer.
   Alarm reporting is used by an intermediate point of a path, that
   becomes aware of a fault on the path, to report to the endpoints of
   the path.  [TP-OAM-FW] states that this may occur as a result of a
   defect condition discovered at a server sub-layer.  This generates an
   Alarm Indication Signal (AIS) that continues until the fault is
   cleared.  The consequent action of this function is detailed in
   [TP-OAM-FW].

4.5.4.6.  Remote Defect Indication

   Remote Defect Indication (RDI) is used proactively by a path endpoint
   to report to its peer endpoint that a defect is detected on a
   bidirectional connection between them.  [MPLS-TP-OAM] points out that
   this function may be applied to a unidirectional LSP only if a return
   path exists.  [TP-OAM-FW] points out that this function is associated
   with the proactive CC-V function.

4.5.4.7.  Client Failure Indication

   Client Failure Indication (CFI) is defined in [MPLS-TP-OAM] to allow
   the propagation information from one edge of the network to the
   other.  The information concerns a defect to a client, in the case
   that the client does not support alarm notification.

4.5.4.8.  Performance Monitoring

   The definition of MPLS performance monitoring was motivated by the
   MPLS-TP requirements [MPLS-TP-OAM] but was defined generically for
   MPLS in [MPLS-LM-DM].  An additional document [TP-LM-DM] defines a
   performance monitoring profile for MPLS-TP.

4.5.4.8.1.  Packet Loss Measurement (LM)

   Packet Loss Measurement is a function used to verify the quality of
   the service.  Packet loss, as defined in [IPPM-1LM] and
   [MPLS-TP-OAM], indicates the ratio of the number of user packets lost
   to the total number of user packets sent during a defined time
   interval.

   There are two possible ways of determining this measurement:

   o  Using OAM packets, it is possible to compute the statistics based
      on a series of OAM packets.  This, however, has the disadvantage
      of being artificial and may not be representative since part of
      the packet loss may be dependent upon packet sizes and upon the
      implementation of the MEPs that take part in the protocol.

   o  Delimiting messages can be sent at the start and end of a
      measurement period during which the source and sink of the path
      count the packets transmitted and received.  After the end
      delimiter, the ratio would be calculated by the path OAM entity.

4.5.4.8.2.  Packet Delay Measurement (DM)

   Packet Delay Measurement is a function that is used to measure one-
   way or two-way delay of a packet transmission between a pair of the
   endpoints of a path (PW, LSP, or Section).  Where:

   o  One-way packet delay, as defined in [IPPM-1DM], is the time
      elapsed from the start of transmission of the first bit of the
      packet by a source node until the reception of the last bit of
      that packet by the destination node.  Note that one-way delay
      measurement requires the clocks of the two endpoints to be
      synchronized.

   o  Two-way packet delay, as defined in [IPPM-2DM], is the time
      elapsed from the start of transmission of the first bit of the
      packet by a source node until the reception of the last bit of the
      looped-back packet by the same source node, when the loopback is
      performed at the packet's destination node.  Note that due to
      possible path asymmetry, the one-way packet delay from one
      endpoint to another is not necessarily equal to half of the

      two-way packet delay.  As opposed to one-way delay measurement,
      two-way delay measurement does not require the two endpoints to be
      synchronized.

      For each of these two metrics, the DM function allows the MEP to
      measure the delay, as well as the delay variation.  Delay
      measurement is performed by exchanging timestamped OAM packets
      between the participating MEPs.

4.6.  Pseudowire OAM

4.6.1.  Pseudowire OAM Using Virtual Circuit Connectivity Verification
        (VCCV)

   VCCV, as defined in [VCCV], provides a means for end-to-end fault
   detection and diagnostic tools to be used for PWs (regardless of the
   underlying tunneling technology).  The VCCV switching function
   provides a Control Channel associated with each PW.  [VCCV] defines
   three Control Channel (CC) types, i.e., three possible methods for
   transmitting and identifying OAM messages:

   o  Control Channel Type 1: In-band VCCV, as described in [VCCV], is
      also referred to as "PWE3 Control Word with 0001b as first
      nibble".  It uses the PW Associated Channel Header [PW-ACH].

   o  Control Channel Type 2: Out-of-band VCCV, as described in [VCCV],
      is also referred to as "MPLS Router Alert Label".  In this case,
      the Control Channel is created by using the MPLS router alert
      label [MPLS-ENCAPS] immediately above the PW label.

   o  Control Channel Type 3: TTL expiry VCCV, as described in [VCCV],
      is also referred to as "MPLS PW Label with TTL == 1", i.e., the
      Control Channel is identified when the value of the TTL field in
      the PW label is set to 1.

   VCCV currently supports the following OAM tools: ICMP Ping, LSP Ping,
   and BFD.  ICMP and LSP Ping are IP encapsulated before being sent
   over the PW ACH.  BFD for VCCV [BFD-VCCV] supports two modes of
   encapsulation -- either IP/UDP encapsulated (with IP/UDP header) or
   PW-ACH encapsulated (with no IP/UDP header) -- and provides support
   to signal the AC status.  The use of the VCCV Control Channel
   provides the context, based on the MPLS-PW label, required to bind
   and bootstrap the BFD session to a particular pseudowire (FEC),
   eliminating the need to exchange Discriminator values.

   VCCV consists of two components: (1) the signaled component to
   communicate VCCV capabilities as part of the VC label, and (2) the
   switching component to cause the PW payload to be treated as a
   control packet.

   VCCV is not directly dependent upon the presence of a control plane.
   The VCCV capability advertisement may be performed as part of the PW
   signaling when LDP is used.  In case of manual configuration of the
   PW, it is the responsibility of the operator to set consistent
   options at both ends.  The manual option was created specifically to
   handle MPLS-TP use cases where no control plane was a requirement.
   However, new use cases such as pure mobile backhaul find this
   functionality useful too.

   The PWE3 working group has conducted an implementation survey of VCCV
   [VCCV-SURVEY] that analyzes which VCCV mechanisms are used in
   practice.

4.6.2.  Pseudowire OAM Using G-ACh

   As mentioned above, VCCV enables OAM for PWs by using a Control
   Channel for OAM packets.  When PWs are used in MPLS-TP networks,
   rather than the Control Channels defined in VCCV, the G-ACh can be
   used as an alternative Control Channel.  The usage of the G-ACh for
   PWs is defined in [PW-G-ACh].

4.6.3.  Attachment Circuit - Pseudowire Mapping

   The PWE3 working group has defined a mapping and notification of
   defect states between a pseudowire (PW) and the Attachment Circuits
   (ACs) of the end-to-end emulated service.  This mapping is of key
   importance to the end-to-end functionality.  Specifically, the
   mapping is provided by [PW-MAP], by [L2TP-EC] for L2TPv3 pseudowires,
   and by Section 5.3 of [ATM-L2] for ATM.

   [L2VPN-OAM] provides the requirements and framework for OAM in the
   context of Layer 2 Virtual Private Networks (L2VPNs), and
   specifically it also defines the OAM layering of L2VPNs over
   pseudowires.

   The mapping defined in [Eth-Int] allows an end-to-end emulated
   Ethernet service over pseudowires.

4.7.  OWAMP and TWAMP

4.7.1.  Overview

   The IPPM working group in the IETF defines common criteria and
   metrics for measuring performance of IP traffic ([IPPM-FW]).  Some of
   the key RFCs published by this working group have defined metrics for
   measuring connectivity [IPPM-Con], delay ([IPPM-1DM], [IPPM-2DM]),
   and packet loss [IPPM-1LM].  It should be noted that the work of the
   IETF in the context of performance metrics is not limited to IP
   networks; [PM-CONS] presents general guidelines for considering new
   performance metrics.

   The IPPM working group has defined not only metrics for performance
   measurement but also protocols that define how the measurement is
   carried out.  The One-Way Active Measurement Protocol [OWAMP] and the
   Two-Way Active Measurement Protocol [TWAMP] each define a method and
   protocol for measuring performance metrics in IP networks.

   OWAMP [OWAMP] enables measurement of one-way characteristics of IP
   networks, such as one-way packet loss and one-way delay.  For its
   proper operation, OWAMP requires accurate time-of-day setting at its
   endpoints.

   TWAMP [TWAMP] is a similar protocol that enables measurement of both
   one-way and two-way (round-trip) characteristics.

   OWAMP and TWAMP are each comprised of two separate protocols:

   o  OWAMP-Control/TWAMP-Control: used to initiate, start, and stop
      test sessions and to fetch their results.  Continuity Check and
      Connectivity Verification are tested and confirmed by establishing
      the OWAMP/TWAMP Control Protocol TCP connection.

   o  OWAMP-Test/TWAMP-Test: used to exchange test packets between two
      measurement nodes.  Enables the loss and delay measurement
      functions, as well as detection of other anomalies, such as packet
      duplication and packet reordering.

   It should be noted that while [OWAMP] and [TWAMP] define tools for
   performance measurement, they do not define the accuracy of these
   tools.  The accuracy depends on scale, implementation, and network
   configurations.

   Alternative protocols for performance monitoring are defined, for
   example, in MPLS-TP OAM ([MPLS-LM-DM], [TP-LM-DM]) and in Ethernet
   OAM [ITU-T-Y1731].

4.7.2.  Control and Test Protocols

   OWAMP and TWAMP control protocols run over TCP, while the test
   protocols run over UDP.  The purpose of the control protocols is to
   initiate, start, and stop test sessions, and for OWAMP to fetch
   results.  The test protocols introduce test packets (which contain
   sequence numbers and timestamps) along the IP path under test
   according to a schedule, and they record statistics of packet
   arrival.  Multiple sessions may be simultaneously defined, each with
   a session identifier, and defining the number of packets to be sent,
   the amount of padding to be added (and thus the packet size), the
   start time, and the send schedule (which can be either a constant
   time between test packets or exponentially distributed
   pseudorandomly).  Statistics recorded conform to the relevant IPPM
   RFCs.

   From a security perspective, OWAMP and TWAMP test packets are hard to
   detect because they are simply UDP streams between negotiated port
   numbers, with potentially nothing static in the packets.  OWAMP and
   TWAMP also include optional authentication and encryption for both
   control and test packets.

4.7.3.  OWAMP

   OWAMP defines the following logical roles: Session-Sender,
   Session-Receiver, Server, Control-Client, and Fetch-Client.  The
   Session-Sender originates test traffic that is received by the
   Session-Receiver.  The Server configures and manages the session, as
   well as returning the results.  The Control-Client initiates requests
   for test sessions, triggers their start, and may trigger their
   termination.  The Fetch-Client requests the results of a completed
   session.  Multiple roles may be combined in a single host -- for
   example, one host may play the roles of Control-Client, Fetch-Client,
   and Session-Sender, and a second may play the roles of Server and
   Session-Receiver.

   In a typical OWAMP session, the Control-Client establishes a TCP
   connection to port 861 of the Server, which responds with a Server
   greeting message indicating supported security/integrity modes.  The
   Control-Client responds with the chosen communications mode, and the
   Server accepts the mode.  The Control-Client then requests and fully
   describes a test session to which the Server responds with its
   acceptance and supporting information.  More than one test session
   may be requested with additional messages.  The Control-Client then
   starts a test session; the Server acknowledges and then instructs the
   Session-Sender to start the test.  The Session-Sender then sends test
   packets with pseudorandom padding to the Session-Receiver until the
   session is complete or until the Control-Client stops the session.

   Once finished, the Session-Sender reports to the Server, which
   recovers data from the Session-Receiver.  The Fetch-Client can then
   send a fetch request to the Server, which responds with an
   acknowledgement and, immediately thereafter, the result data.

4.7.4.  TWAMP

   TWAMP defines the following logical roles: Session-Sender,
   Session-Reflector, Server, and Control-Client.  These are similar to
   the OWAMP roles, except that the Session-Reflector does not collect
   any packet information, and there is no need for a Fetch-Client.

   In a typical TWAMP session, the Control-Client establishes a TCP
   connection to port 862 of the Server, and the mode is negotiated as
   in OWAMP.  The Control-Client then requests sessions and starts them.
   The Session-Sender sends test packets with pseudorandom padding to
   the Session-Reflector, which returns them with timestamps inserted.

4.8.  TRILL

   The requirements of OAM in TRILL are defined in [TRILL-OAM].  The
   challenge in TRILL OAM, much like in MPLS networks, is that traffic
   between RBridges RB1 and RB2 may be forwarded through more than one
   path.  Thus, an OAM protocol between RBridges RB1 and RB2 must be
   able to monitor all the available paths between the two RBridges.

   During the writing of this document, the detailed definition of the
   TRILL OAM tools is still work in progress.  This subsection presents
   the main requirements of TRILL OAM.

   The main requirements defined in [TRILL-OAM] are:

   o  Continuity Checking (CC) - the TRILL OAM protocol must support a
      function for CC between any two RBridges RB1 and RB2.

   o  Connectivity Verification (CV) - connectivity between two RBridges
      RB1 and RB2 can be verified on a per-flow basis.

   o  Path Tracing - allows an RBridge to trace all the available paths
      to a peer RBridge.

   o  Performance monitoring - allows an RBridge to monitor the packet
      loss and packet delay to a peer RBridge.