Network Working Group                                 Andrew G. Malis
Internet Draft                                                Ken Hsu
Expiration Date: February 2002                  Vivace Networks, Inc.

                                                       Jeremy Brayley
                                                      Steve Vogelsang
                                                         John Shirron
                                                Laurel Networks, Inc.

                                                         Luca Martini
                                         Level 3 Communications, LLC.

                                                          Tom Johnson
                                                        Marlene Drost
                                                           Ed Hallman
                                      Litchfield Communications, Inc.

                                                            July 2001


   SONET/SDH Circuit Emulation Service Over MPLS (CEM) Encapsulation
                     draft-malis-sonet-ces-mpls-05.txt


Status of this Memo

   This document is an Internet-Draft and is in full conformance with
   all provisions of section 10 of RFC 2026 [1].

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF), its areas, and its working groups. Note that
   other groups may also distribute working documents as Internet-
   Drafts.

   Internet-Drafts are draft documents valid for a maximum of six
   months and may be updated, replaced, or obsoleted by other documents
   at any time. It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   The list of current Internet-Drafts can be accessed at
   http://www.ietf.org/ietf/1id-abstracts.txt.

   The list of Internet-Draft Shadow Directories can be accessed at
   http://www.ietf.org/shadow.html.


1. Abstract

This document describes a method for encapsulating SONET/SDH Path
signals for transport across packet-switched networks (PSNs).  The PSNs
explicitly supported by this document include MPLS and IP.




Malis, et al.           Expires February 2002                [Page 1]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



2. Conventions used in this document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC 2119 [2].



3. Introduction

   This document describes a method for encapsulating time division
   multiplexed (TDM) digital signals for transmission over packet-
   switched networks.

   The transmission system for circuit-oriented TDM signals is the
   Synchronous Optical Network (SONET) [3], [6] / Synchronous Digital
   Hierarchy (SDH) [4]. To support TDM traffic (which includes voice,
   data, and private leased line services) PSNs must emulate the
   circuit characteristics of SONET/SDH payloads.  MPLS labels and a
   new circuit emulation header are used to encapsulate TDM signals and
   provide the Circuit Emulation Service over MPLS (CEM) function.  The
   MPLS encapsulation may be further encapsulated in IP for carriage
   across IP PSNs [8].

   This document also describes an optional extension to CEM called
   Dynamic Bandwidth Allocation (DBA).  This is a method for
   dynamically reducing the bandwidth utilized by emulated SONET/SDH
   circuits in the packet network.  This bandwidth reduction is
   accomplished by not sending the SONET/SDH payload through the packet
   network under certain conditions such as AIS-P or STS SPE
   Unequipped.

   This document is closely related to references [5], which describes
   the control protocol methods used to signal the usage of CEM, [7],
   which describes a related method of encapsulating Layer 2 frames
   over MPLS and which shares the same signaling, and [11] which
   describes a MIB for controlling and observing CEM services.


4. Scope

   This document describes how to provide CEM for the following digital
   signals:

   1. SONET STS-1 synchronous payload envelope (SPE)/SDH VC-3

   2. STS-Nc SPE (N = 3, 12, or 48)/SDH VC-4, VC-4-4c, VC-4-16c

   3. Unstructured SONET Emulation, where the entire SONET bit-stream
   (including the transport overhead) is packetized and transported
   across the PSN.


Malis, et al.           Expires February 2002                [Page 2]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   For the remainder of this document, these constructs will be
   referred to as SONET/SDH channels.


   Other SONET/SDH signals, such as virtual tributary (VT) structured
   sub-rate mapping, are not explicitly discussed in this document;
   however, it can be extended in the future to support VT and lower
   speed non-SONET/SDH services. OC-192c SPE/VC-4-64c are also not
   included at this point, since most PSNs use OC-192c or slower
   trunks, and thus would not have sufficient capacity.  As trunk
   capacities increase in the future, the scope of this document can be
   accordingly extended.

5. CEM Encapsulation Format

   In order to transport SONET/SDH SPEs through a packet-oriented
   network, the SPE is broken into fragments.  A 32-bit CEM Header is
   pre-pended to each fragment.  The Basic CEM packet appears in Figure
   1.

             +-----------------------------------+
             |            CEM Header             |
             +-----------------------------------+
             |                                   |
             |                                   |
             |        SONET/SDH SPE Fragment     |
             |                                   |
             |                                   |
             +-----------------------------------+

             Figure 1. Basic CEM Packet


   The 32-bit CEM header has the following format:

      0                   1                   2                   3
      0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
      |D|R|Rvd|   Sequence Num    | Structure Pointer |N|P|   ECC-6   |
      +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                        Figure 2. CEM Header Format


   The above fields are defined as follows:

   D bit: Signals DBA Mode.  MUST be set to zero for Normal Operation.
   MUST be set to one if CEM is currently in DBA mode.  DBA is an
   optional mode during which trivial SPEs are not transmitted into the
   packet network.  See Table 1 and sections 7 and 8 for further
   details.  Note: for unstructured CEM, the D-bit MUST be set to zero.



Malis, et al.           Expires February 2002                [Page 3]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   R bit: CEM-RDI.  This bit is set to one to signal to the remote CEM
   function that a loss of packet synchronization has occurred.

   Rvd: These bits are reserved for future use, and MUST be set to
   zero.

   Sequence Number:  This is a packet sequence number, which MUST
   continuously cycle from 0 to 1023.  It SHOULD begin at zero when a
   CEM LSP is created.

   Structure Pointer: The Structure Pointer MUST contain the offset of
   the J1 byte within the CEM payload. The value is from 0 to 1,022,
   where 0 means the first byte after the CEM header. The Structure
   Pointer MUST be set to 0x3FF (1,023) if a packet does not carry the
   J1 byte.  See [3], [4] and [6] for more information on the J1 byte
   and the SONET/SDH payload pointer.  Note: for unstructured CEM, the
   Structure Pointer field MUST be set to 0x3FF.

   The N and P bits: Indicate negative and positive pointer adjustment
   events.  They are also used to relay SONET/SDH maintenance signals
   such as AIS-P.  See Table 1 and sections 7 and 8 for more details.
   Note: for unstructured CEM, the N and P bits MUST both be set to 0.

         +---+---+---+----------------------------------------------+
         | D | N | P |         Interpretation                       |
         +---+---+---+----------------------------------------------+
         | 0 | 0 | 0 | Normal Mode “ No Ptr Adjustment              |
         | 0 | 0 | 1 | Normal Mode “ Positive Ptr Adjustment        |
         | 0 | 1 | 0 | Normal Mode “ Negative Ptr Adjustment        |
         | 0 | 1 | 1 | Normal Mode “ AIS-P                          |
         |   |   |   |                                              |
         | 1 | 0 | 0 | DBA Mode    “ STS SPE Unequipped             |
         | 1 | 0 | 1 | DBA Mode    “ STS SPE Unequipped Pos Ptr Adj |
         | 1 | 1 | 0 | DBA Mode    “ STS SPE Unequipped Neg Ptr Adj |
         | 1 | 1 | 1 | DBA Mode    “ AIS-P                          |
         +---+---+---+----------------------------------------------+

         Table 1. Interpretation of D, N, and P bits

   ECC-6: An Error Correction Code to protect the CEM header.  This
   offers the ability to correct single bit errors and detect up to two
   bit errors.  The ECC algorithm is described in Appendix B.  The ECC-
   6 can be optionally disabled at provisioning time.  If the ECC-6 is
   not utilized it MUST be set to zero.

   Note: Normal CEM packets are fixed in length for all of the packets
   of a particular emulated TDM stream.  This length is signaled using
   the CEM Payload Bytes parameter defined in [5], or is statically
   provisioned for each TDM stream.  Therefore, the length of each CEM
   packet does not need to be carried in the CEM header.

5.1 Transport Encapsulation


Malis, et al.           Expires February 2002                [Page 4]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   In principle, CEM packets can be transported over any packet-
   oriented network.  The following sections describe specifically how
   CEM packets MUST be encapsulated for transport over MPLS or IP
   networks.

5.1.1 MPLS Transport

   To transport a CEM packet over an MPLS network, an MPLS label-stack
   MUST be pushed on top of the CEM packet.

   The last two labels prior to the CEM header are referred to as the
   Tunnel and Virtual Circuit (VC) labels.

   The VC label is required, and is the last label prior to the CEM
   Header.  The VC label MUST be used to identify the CEM connection
   within the MPLS tunnel.

   The optional tunnel label is immediately above the VC label on the
   label stack.  If present, the tunnel label MUST be used to identify
   the MPLS LSP used to tunnel the TDM packets through the MPLS network
   (the tunnel LSP).

   This is similar to the label stack usage defined in [5] and [7].


                 +-----------------------------------+
                 | Additional MPLS Labels (Optional) |
                 +-----------------------------------+
                 |       Tunnel Label (Optional)     |
                 +-----------------------------------+
                 |             VC Label              |
                 +-----------------------------------+
                 |            CEM Header             |
                 +-----------------------------------+
                 |                                   |
                 |                                   |
                 |       SONET/SDH SPE Fragment      |
                 |                                   |
                 |                                   |
                 +-----------------------------------+

                 Figure 3. Typical MPLS Transport Encapsulation


5.1.1 IP Transport

   It is highly desirable to define a single encapsulation format that
   will work for both IP and MPLS.  Furthermore, it is desirable that
   the encapsulation mechanism be as efficient as possible.

   One way to achieve these goals is to map CEM directly onto IP by
   mapping the previously described MPLS packets onto IP.


Malis, et al.           Expires February 2002                [Page 5]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   A mechanism for carrying MPLS over IP is described in [8].

   Using this encapsulation scheme would result in the packet format
   illustrated in figure 4.



                 +-----------------------------------+
                 |                                   |
                 |        IPv6/v4 Header [8]         |
                 |                                   |
                 +-----------------------------------+
                 |      Tunnel Label (Optional)      |
                 +-----------------------------------+
                 |             VC Label              |
                 +-----------------------------------+
                 |            CEM Header             |
                 +-----------------------------------+
                 |                                   |
                 |                                   |
                 |       SONET/SDH SPE Fragment      |
                 |                                   |
                 |                                   |
                 +-----------------------------------+

                  Figure 4. MPLS Transport Encapsulation




























Malis, et al.           Expires February 2002                [Page 6]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



6. CEM Operation

   The following sections describe CEM operation.

6.1 Introduction and Terminology

   There are two types of CEM: structured and unstructured.

   Unstructured CEM packetizes the entire SONET/SDH bit-stream
   (including transport overhead).

   Structured CEM terminates the transport overhead and packetizes
   individual channels (STS-1/Nc) within the SONET/SDH frame.  Because
   structured CEM terminates the transport overhead, structured CEM
   implementations SHOULD meet the generic requirements for SONET/SDH
   Line Terminating Equipment as defined in [3], [4], and [6].

   Implementations MUST support structured CEM and MAY support
   unstructured CEM.

   Structured CEM MUST support a normal mode of operation and MAY
   support an optional extension called Dynamic Bandwidth Allocation
   (DBA).  During normal operation, SONET/SDH payloads are fragmented,
   pre-pended with the CEM Header, the VC label, and the PSN header,
   and then transmitted into the packet network.  During DBA mode, only
   the CEM header, the VC label, and PSN header are transmitted.  This
   is done to conserve bandwidth when meaningful user data is not
   present in the SPE, such as during AIS-P or STS SPE Unequipped.

6.1.1 CEM Packetizer and De-Packetizer

   As with all adaptation functions, CEM has two distinct components:
   adapting TDM SONET/SDH into a CEM packet stream, and converting the
   CEM packet stream back into a TDM SONET/SDH.  The first function
   will be referred to as CEM Packetizer and the second as CEM De-
   Packetizer.  This terminology is illustrated in figure 5.


             +------------+              +---------------+
             |            |              |               |
   SONET --> |    CEM     | --> PSN  --> |      CEM      | --> SONET
    SDH      | Packetizer |              | De-Packetizer |      SDH
             |            |              |               |
             +------------+              +---------------+

   Figure 5. CEM Terminology

   Note: the CEM de-packetizer requires a buffering mechanism to
   account for delay variation in the CEM packet stream.  This
   buffering mechanism will be generically referred to as the CEM
   jitter buffer.


Malis, et al.           Expires February 2002                [Page 7]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


6.1.2 CEM DBA

   DBA is an optional mode of operation for structured CEM that only
   transmits the CEM Header, the VC label, and PSN Header into the
   packet network under certain circumstances such as AIS-P or STS
   Unequipped.

   If DBA is supported by a CEM implementation, the user SHOULD be able
   to configure if DBA will be triggered by AIS-P, STS Unequipped,
   both, or neither on a per channel basis.

   If DBA is supported, the determination of AIS-P and STS Unequipped
   MUST be based on the state of SONET/SDH Section, Line, and Path
   Overhead bytes.  DBA based on pattern detection within the SPE (i.e.
   all zeros, 7Es, or ATM idle cells) is for further study.

   During AIS-P, there is no valid payload pointer, so pointer
   adjustments cannot occur.  During STS Unequipped, the SONET/SDH
   payload pointer is valid, and therefore pointer adjustments MUST be
   supported even during DBA.  See Table 1 for details.

6.2 Description of Normal CEM Operation

   During normal operation, the CEM packetizer will receive a fixed
   rate byte stream from a SONET/SDH interface.  When a packets worth
   of data has been received from a SONET/SDH channel, the CEM Header,
   the VC Label, and PSN Header are pre-pended to the SPE fragment and
   the resulting CEM packet is transmitted into the packet network.
   Because all normal CEM packets associated with a specific SONET/SDH
   channel will have the same length, the transmission of CEM packets
   for that channel SHOULD occur at regular intervals.

   At the far end of the packet network, the CEM de-packetizer will
   receive packets into a jitter buffer and then play out the received
   byte stream at a fixed rate onto the corresponding SONET/SDH
   channel.  The jitter buffer SHOULD be adjustable in length to
   account for varying network delay behavior.  The receive packet rate
   from the packet network should be exactly balanced by the
   transmission rate onto the SONET/SDH channel, on average.  The time
   over which this average is taken corresponds to the depth of the
   jitter buffer for a specific CEM channel.

   The CEM sequence numbers provide a mechanism to detect lost and/or
   mis-ordered packets.  The CEM de-packetizer MUST detect lost or mis-
   ordered packets.  The CEM de-packetizer MUST play out a programmable
   byte pattern in place of any dropped packets.  The CEM de-packetizer
   MAY re-order packets received out of order.  If the CEM de-
   packetizer does not support re-ordering, it MUST drop mis-ordered
   packets.

6.3 Description of CEM Operation during DBA

   (Note: DBA is only applicable to structured CEM.)

Malis, et al.           Expires February 2002                [Page 8]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



   There are several issues that should be addressed by a workable CEM
   DBA mechanism.  First, when DBA is invoked, there should be a
   substantial savings in bandwidth utilization in the packet network.
   The second issue is that the transition in and out of DBA should be
   tightly coordinated between the local CEM packetizer and CEM de-
   packetizer at the far side of the packet network.  A third is that
   the transition in and out of DBA should be accomplished with minimal
   disruption to the adapted data stream.

   Another goal is that the reduction of CEM traffic due to DBA should
   not be mistaken for a fault in the packet network or vice-versa.
   Finally, the implementation of DBA should require minimal
   modifications beyond what is necessary for the nominal CEM case.
   The mechanism described below is a reasonable balance of these
   goals.

   During DBA, packets MUST be emitted at exactly the same rate as they
   would be during normal operation.  This SHOULD be accomplished by
   transmitting each DBA packet after a complete packet of data has
   been received from the SONET/SDH channel.  The only change from
   normal operation is that the CEM packets during DBA MUST only carry
   the CEM header, the VC Label, and the PSN Header.

   Because some links have a minimum supported packet size, the CEM
   packetizer MAY append a configurable number of bytes immediately
   after the CEM-header to pad out the CEM packet to reach the mimumum
   supported packet size.  The value of the padding bytes is
   implementation specific.  The D-bit MUST be set to one, to indicate
   that DBA is active.

   The CEM de-packetizer MUST assume that each packet received with the
   D-bit set represents a normal-sized packet containing an AIS-P or
   SPE Unequipped payload as noted by N and P.  See Table 1.  The CEM
   de-packetizer MUST accept DBA packets with or without padding.

   This allows the CEM packetization and de-packetization logic during
   DBA to be similar to the nominal case.  It insures that the correct
   SONET/SDH indication is reliably transmitted between CEM adaptation
   points.  It minimizes the risk of under or over running the jitter
   buffer during the transition in and out of DBA.  And, it guarantees
   that faults in the packet network are recognized as distinctly
   different from line conditioning on the SONET/SDH interfaces.

6.4 Packet Synchronization

   A key component in declaring the state of a CEM service is whether
   or not the CEM de-packetizer is in or out of packet synchronization.
   The following paragraphs describe how that determination is made.

6.4.1 Acquisition of Packet Synchronization



Malis, et al.           Expires February 2002                [Page 9]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   At startup, a CEM de-packetizer will be out of packet
   synchronization by default.  To declare packet synchronization at
   startup or after a loss of packet synchronization, the CEM de-
   packetizer must receive a configurable number of CEM packets with
   sequential sequence numbers.

6.4.2 Loss of Packet Synchronization

   Once a CEM de-packetizer is in packet sync, it may encounter a set
   of events that will cause it to lose packet synchronization.

   As discussed in section 6.2, a CEM de-packetizer MAY or MAY NOT
   support re-ordering of mis-ordered packets.

   If a CEM de-packetizer supports re-ordering, then the determination
   that packet synchronization has been lost cannot be made at the time
   the packets are received from the PSN.  Instead, the determination
   MUST be made as the packets are being played out onto the SONET/SDH
   interface.  This is because it is only at play-out time that the
   determination can be made as to whether the entire emulated
   SONET/SDH stream was received from the PSN.

   If a CEM de-packetizer does not support re-ordering, a number of
   approaches may be used to minimize the impact of mis-ordered or lost
   packets on the final re-assembled SONET/SDH stream.  For example,
   AAL1 [9] uses a simple state-machine to re-order packets in a sub-
   set of possible cases.  The algorithm for these state-machines is
   outside of the scope of CEM.  However, the final determination as to
   whether or not to declare loss of packet synchronization MUST be
   based on the same criteria as for implementations that do support
   re-ordering.

   Whether or not a CEM implementation supports re-ordering, the
   declaration of loss of packet synchronization MUST be based on the
   following criteria.

   As packets are played out towards the SONET/SDH interface, the CEM
   de-packetizer will encounter Ÿempty÷ packets in the place of packets
   that were dropped by the PSN, or effectively dropped due to
   limitations of the CEM implementation.  If the CEM de-packetizer
   encounters more than a configurable number of sequential dropped
   packets, the CEM de-packetizer MUST declare loss of packet
   synchronization.

7. SONET/SDH Maintenance Signals

   There are several issues that must be considered in the mapping of
   maintenance signals between SONET/SDH and a PSN.  A description of
   how these signals and conditions are mapped between the two domains
   is described below.

   For clarity, the mappings are split into two groups: SONET/SDH to
   PSN, and PSN to SONET/SDH.

Malis, et al.           Expires February 2002               [Page 10]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



7.1 SONET/SDH to PSN
   The following sections describe how SONET/SDH Maintenance Signals
   and Alarm conditions are mapped into a Packet Switched Network.


















































Malis, et al.           Expires February 2002               [Page 11]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



7.1.1 AIS-P Indication

   In a SONET/SDH network, SONET/SDH Path outages are signaled using
   maintenance alarms such as Path AIS (AIS-P).  In particular, AIS-P
   indicates that the SONET/SDH Path is not currently transmitting
   valid end-user data, and the SPE contains all ones.

   It should be noted that for structured CEM nearly every type of
   service-effecting section or line defect will result in an AIS-P
   condition.

   The SONET/SDH hierarchy is illustrated below.

                               +----------+
                               |   PATH   |
                               +----------+
                                    ^
                                    |
                                  AIS-P
                                    |
                                    |
                               +----------+
                               |   LINE   |
                               + ---------+
                                  ^     ^
                                  |     |
                                AIS-L   +------ LOP
                                  |
                                  |
                               +----------+
                               | SECTION  |
                               +----------+
                                  ^    ^
                                  |    |
                                  |    |
                                 LOS  LOF


                       Figure 6.  SONET/SDH Fault Hierarchy.

   Should the Section Layer detect a Loss of Signal (LOS) or Loss of
   Frame (LOF) condition, it sends AIS-L up to the Line Layer.  If the
   Line Layer detects AIS-L or Loss of Path (LOP), it sends AIS-P to
   the Path Layer.

   In normal mode during AIS-P, structured CEM packets are generated as
   usual.  The N and P bits MUST be set to 11 binary to signal AIS-P
   explicitly through the packet network.  The D-bit MUST be set to
   zero to indicate that the SPE is being carried through the packet
   network.  Normal CEM packets with the SPE fragment, CEM Header, the
   VC Label, and PSN Header MUST be transmitted into the packet
   network.

Malis, et al.           Expires February 2002               [Page 12]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



   However, to conserve network bandwidth during AIS-P, DBA MAY be
   employed.  If DBA has been enabled for AIS-P and AIS-P is currently
   occurring, the N and P bits MUST be set to 11 binary to signal AIS,
   and the D-bit MUST be set to one to indicate that the SPE is not
   being carried through the packet network.  Only the CEM header, the
   VC Label, and the PSN Header MUST be transmitted into the packet
   network.

   Also note that this differs from the outage mechanism in [5], which
   withdraws the VC label as a result of an endpoint outage.  TDM
   circuit emulation requires the ability to distinguish between the
   de-provisioning of a circuit (which causes the VC label to be
   withdrawn), and temporary outages (which are signaled using AIS-P).


7.1.2 STS SPE Unequipped Indication

   The STS SPE Unequipped Indication is a slightly different case than
   AIS-P.  When byte C2 of the Path Overhead (STS path signal label) is
   00h and Byte B3 (STS Path BIP-8) is valid, it indicates that the SPE
   is unequipped.  Note: this is typically signaled by setting the
   entire SPE to zeros.

   For normal structured CEM operation during SPE Unequipped, the N and
   P bits MUST be interpreted as usual.  The SPE MUST be transmitted
   into the packet network along with the CEM Header, the VC Label, and
   PSN Header, and the D-Bit MUST be set to zero.

   If DBA has been enabled for STS SPE Unequipped and the Unequipped
   condition is occurring on the SONET/SDH channel, the D-bit MUST be
   set to one to indicate DBA is active.  Only the CEM Header, the VC
   Label, and PSN Header MUST be transmitted into the packet network.
   The N and P bits MUST be used to signal pointer adjustments as
   normal.  See Table 1 and section 8 for details.

7.1.3 CEM-RDI

   The CEM function MUST send CEM-RDI towards the packet network during
   loss of packet synchronization.  This MUST be accomplished by
   setting the R bit to one in the CEM header.  This applies for both
   structured and unstructured CEM.

7.2 PSN to SONET/SDH

   The following sections discuss how the various conditions on the
   packet network are converted into SONET/SDH indications.

7.2.1 AIS-P Indication

   There are several conditions in the packet network that will cause
   the structured CEM de-packetization function to send an AIS-P
   indication onto a SONET/SDH channel.

Malis, et al.           Expires February 2002               [Page 13]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



   The first of these is the receipt of structured CEM packets with the
   N and P bits set to one, and the D-bit set to zero.  This is an
   explicit indication of AIS-P being received at the far-end of the
   packet network, with DBA disabled for AIS-P.  The CEM de-packetizer
   MUST play out the received SPE fragment (which will incidentally be
   carrying all ones), and MUST configure the SONET/SDH Overhead to
   signal AIS-P as defined in [3], [4], and [6].

   The second case is the receipt of structured CEM packets with the N
   and P bits set to one, and the D-bit set to one.  This is an
   explicit indication of AIS-P being received at the far-end of the
   packet network, with DBA enabled for AIS-P.  The CEM de-packetizer
   MUST play out one packetËs worth of all ones for each packet
   received, and MUST configure the SONET/SDH Overhead to signal AIS-P
   as defined in [3], [4], and [6].

   A third case that will cause a structured CEM de-packetization
   function to send an AIS-P indication onto a SONET/SDH channel is
   loss of packet synchronization.


7.2.2 STS SPE Unequipped Indication

   There are three conditions in the packet network that will cause the
   CEM function to transmit STS SPE Unequipped indications onto the
   SONET/SDH channel.

   The first, which is transparent to CEM, is the receipt of regular
   CEM packets that happen to be carrying an SPE that contains the
   appropriate Path overhead to signal STS SPE unequipped.  This case
   does not require any special processing on the part of the CEM de-
   packetizer.

   The second case is the receipt of structured CEM packets that have
   the D-bit set to one to indicate DBA active and the N and P bits set
   to 00 binary, 01 binary, or 10 binary to indicate SPE Unequipped
   with or without pointer adjustments.  The CEM de-packetizer MUST use
   this information to transmit a packet of all zeros onto the
   SONET/SDH interface, and adjust the payload pointer as necessary.

   The third case when a structured CEM de-packetizer MUST send an STS
   SPE Unequipped Indication towards the SONET/SDH interface is when
   the VC-label has been withdrawn due to de-provisioning of the
   circuit.

8. Clocking Modes

   It is necessary to be able to regenerate the input service clock at
   the output interface.  Two clocking modes are supported: synchronous
   and asynchronous.  Selection of the clocking mode is made as part of
   service provisioning.  Both ends of the emulated circuit must be
   configured with the same clocking mode.

Malis, et al.           Expires February 2002               [Page 14]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



8.1 Synchronous


   When synchronous SONET/SDH timing is available at both ends of the
   circuit, the issue of clock recovery becomes much simpler.

8.1.1 Synchronous Unstructured CEM

   For unstructured CEM, the external clock is used to clock each bit
   onto the optical carrier.

8.1.2 Synchronous Structured CEM

   For structured CEM, the external clock is used to clock the
   SONET/SDH carrier.  The N and P bits are used to signal negative or
   positive pointer justification events between structured CEM end-
   points.

   If there is a frequency offset between the frame rate of the
   transport overhead and that of the SONET/SDH SPE, then the alignment
   of the SPE shall periodically slip back or advance in time through
   positive or negative stuffing. The N and P bits are used to replay
   the pointer adjustment events and eliminate transport jitter.

   During a negative pointer adjustment event, the H3 byte from the
   SONET/SDH stream is incorporated into the CEM packet payload in
   order with the rest of the SPE.  During a positive pointer
   adjustment event, the stuff byte is not included in the CEM packet
   payload.

   The pointer adjustment event MUST be transmitted in three
   consecutive packets by the packetizer. The de-packetizer MUST play
   out the pointer adjustment event when the first packet of the three
   with N/P bit set is received.

   The CEM de-packetizer MUST utilize the CEM sequence numbers to
   insure that SONET/SDH pointer adjustment events are not played any
   more frequently than once per every three CEM packets transmitted by
   the remote CEM packetizer.

   References [3],[4],and [6] specify that pointer adjustment events
   MUST be separated by three SONET/SDH frames without a pointer
   adjustment event.  In order to relay all legal pointer adjustment
   events, the packet size for a specific circuit MUST be no larger
   than (783 * 4 * N)/3, where N is the STS-Nc multiplier.

   However, some SONET/SDH equipment allows pointer adjustments to
   occur in back to back SONET/SDH frames.  In order to support this
   possibility, the packet size for a particular circuit SHOULD be no
   larger than (783*N)/3.  Where N is the STS-Nc multiplier.



Malis, et al.           Expires February 2002               [Page 15]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   Since the minimum value of N is one, CEM implementations SHOULD
   support a minimum payload length of 783/3 or 261 bytes.  Smaller
   payload lengths MAY be supported as an option.


8.2 Asynchronous


   If synchronous timing is not available, other methods MAY be
   employed to regenerate the circuit timing.

   For structured CEM, the CEM packetizer SHOULD generate the N and P
   bits as usual.  However, without external synchronization, this
   information is not sufficient to reliably justify the SPE within the
   SONET/SDH transport framing at the CEM de-packetizer.  The de-
   packetizer MAY employ an adaptive algorithm to introduce pointer
   adjustment events to map the CEM SPE to the SONET/SDH transport
   framing.  Regardless of whether the N and P bits are used by the de-
   packetizer as part of the adaptive clock recovery algorithm, they
   MUST still be used in conjunction with the D-bit to signal AIS-P,
   SPE Unequipped, and DBA.

   For unstructured CEM, the CEM de-packetizer MAY use an adaptive
   clock recovery technique to regenerate the SONET/SDH transport
   clock.

   An example adaptive clock recovery method can be found in section
   3.4.2 of [10].


9. CEM LSP Signaling

   For maximum network scaling in MPLS applications, CEM LSP signaling
   may be performed using the LDP Extended Discovery mechanism as
   augmented by the VC FEC Element defined in [5].  MPLS traffic
   tunnels may be dedicated to CEM, or shared with other MPLS-based
   services.  The value 8008 is used for the VC Type in the VC FEC
   Element in order to signify that the LSP being signaled is to carry
   CEM.  Note that the generic control word defined in [6] is not used,
   as its functionality is included in the CEM encapsulation header.

   Alternatively, static label assignment may be used, or a dedicated
   traffic engineered LSP may be used for each CEM service.

   Normal CEM packets are fixed in length for all of the packets of a
   particular emulated TDM stream.  This length is signaled using the
   CEM Payload Bytes parameter defined in [5], or is statically
   provisioned for each CEM service.

   At this time, other aspects of the CEM service MUST be statically
   provisioned.  The CEM-MIB [11] provides a method for statically
   provisioning CEM services.


Malis, et al.           Expires February 2002               [Page 16]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   In [5], Parameter ID 0x05 is allocated for †CEM OptionsË.  At this
   time, the CEM Options parameter is reserved and MUST be set to zero.
   The use of the CEM Options parameter is for future consideration.

10. Open Issues

   Future revisions of this document may discuss the following items.

   Underlying MPLS QoS requirements are not covered by this revision of
   the draft.  Future revisions may discuss underlying QoS
   requirements.

   Support for VT and lower speed non-SONET/SDH services are not
   covered in this revision of the draft.  Future revisions may address
   VT and non-SONET/SDH TDM services.

   Use of the CEM Options parameter in [5] is currently undefined.
   Future revision of this draft will determine how the CEM Options
   word will be used.

   The CEM-MIB[11] calls out a number of CEM performance parameters
   such as Errored Seconds, Severely Errored Seconds, and Unavailable
   Seconds.  These terms are currently not defined in this draft.  The
   definition of these terms is for future consideration.

   The current draft only considers DBA based on SONET/SDH Overhead.
   It may be desirable to extending DBA to include pattern-based
   suppression such as long runs of HDLC flags (i.e. 0x7E).



11. Security Considerations

   As with [5], this document does not affect the underlying security
   issues of MPLS.



















Malis, et al.           Expires February 2002               [Page 17]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


12. Intellectual Property Disclaimer

   This document is being submitted for use in IETF standards
   discussions.  Vivace Networks, Inc. has filed one or more patent
   applications relating to the CEM technology outlined in this
   document.  Vivace Networks, Inc. will grant free unlimited licenses
   for use of this technology.


13. References

   [1]  Bradner, S., "The Internet Standards Process -- Revision 3",
        BCP 9, RFC 2026, October 1996.

   [2]  Bradner, S., "Key words for use in RFCs to Indicate Requirement
        Levels", BCP 14, RFC 2119, March 1997

   [3]  American National Standards Institute, "Synchronous Optical
        Network (SONET) - Basic Description including Multiplex
        Structure, Rates and Formats," ANSI T1.105-1995.

   [4]  ITU Recommendation G.707, "Network Node Interface For The
        Synchronous Digital Hierarchy", 1996.

   [5]  Martini et al, "Transport of Layer 2 Frames Over MPLS", draft-
        martini-l2circuit-trans-mpls-06.txt, work in progress, July
        2001.

   [6]  Telcordia Technologies, ŸSynchronous Optical Network (SONET)
        Transport Systems: Common Generic Criteria÷, GR-253-CORE, Issue
        3, September 2000.

   [7]  Martini et al, "Encapsulation Methods for Transport of Layer 2
        Frames Over MPLS", draft-martini-l2circuit-encap-mpls-02.txt,
        work in progress, July 2001.

   [8]  Worster, ŸMPLS Label Stack Encapsulation in IP÷, draft-worster-
        mpls-in-ip-05, work in progress, July 2001.

   [9]  ITU-T, ŸRecommendation I.363.1, B-ISDN Adaptation Layer
        Specification: Type AAL1÷, Appendix III, August 1996.

   [10]  ATM Forum, "Circuit Emulation Service Interoperability
        Specification Version 2.0", af-vtoa-0078.000, January 1997.

   [11] Danenberg et al, "SONET/SDH Circuit Emulation Service Over MPLS
        (CEM) Management Information Base Using SMIv2", draft-
        danenberg-pw-cem-mib-01.txt, work in progress, July 2001.


14. Acknowledgments



Malis, et al.           Expires February 2002               [Page 18]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   The authors would like to thank Mitri Halabi and Bob Colvin, both of
   Vivace Networks, for their comments and suggestions.


15. Authors' Addresses

   Andrew G. Malis
   Vivace Networks, Inc.
   2730 Orchard Parkway
   San Jose, CA 95134
   Email: Andy.Malis@vivacenetworks.com

   Ken Hsu
   Vivace Networks, Inc.
   2730 Orchard Parkway
   San Jose, CA 95134
   Email: Ken.Hsu@vivacenetworks.com

   Jeremy Brayley
   Laurel Networks, Inc.
   2706 Nicholson Rd.
   Sewickley, PA 15143
   Email: jbrayley@laurelnetworks.com

   Steve Vogelsang
   Laurel Networks, Inc.
   2706 Nicholson Rd.
   Sewickley, PA 15143
   Email: sjv@laurelnetworks.com

   John Shirron
   Laurel Networks, Inc.
   2607 Nicholson Rd.
   Sewickley, PA 15143
   Email: jshirron@laurelnetworks.com

   Luca Martini
   Level 3 Communications, LLC.
   1025 Eldorado Blvd.
   Broomfield, CO 80021
   Email: luca@level3.net

   Thomas K. Johnson
   Litchfield Communications, Inc.
   76 Westbury Park Rd.
   Watertown, CT 06795
   Email: tom_johnson@litchfieldcomm.com

   Ed Hallman
   Litchfield Communications, Inc.
   76 Westbury Park Rd.
   Watertown, CT 06795
   Email: ed_hallman@litchfieldcomm.com

Malis, et al.           Expires February 2002               [Page 19]


                SONET/SDH Circuit Emulation Over MPLS       July 2001



   Marlene Drost
   Litchfield Communications, Inc.
   76 Westbury Park Rd.
   Watertown, CT 06795
   Email: marlene_drost@litchfieldcomm.com
















































Malis, et al.           Expires February 2002               [Page 20]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   Appendix A. SONET/SDH Rates and Formats

   For simplicity, the discussion in this section uses SONET
   terminology, but it applies equally to SDH as well.  SDH-equivalent
   terminology is shown in the tables.

   The basic SONET modular signal is the synchronous transport signal-
   level 1 (STS-1). A number of STS-1s may be multiplexed into higher-
   level signals denoted as STS-N, with N synchronous payload envelopes
   (SPEs). The optical counterpart of the STS-N is the Optical Carrier-
   level N, or OC-N. Table 2 lists standard SONET line rates discussed
   in this document.


     OC Level          OC-1    OC-3    OC-12      OC-48     OC-192
     SDH Term             -   STM-1    STM-4     STM-16     STM-64
     Line Rate(Mb/s) 51.840 155.520  622.080  2,488.320  9,953.280

                    Table 2. Standard SONET Line Rates


   Each SONET frame is 125 ³s and consists of nine rows. An STS-N frame
   has nine rows and N*90 columns. Of the N*90 columns, the first N*3
   columns are transport overhead and the other N*87 columns are SPEs.
   A number of STS-1s may also be linked together to form a super-rate
   signal with only one SPE. The optical super-rate signal is denoted
   as OC-Nc, which has a higher payload capacity than OC-N.

   The first 9-byte column of each SPE is the path overhead (POH) and
   the remaining columns form the payload capacity with fixed stuff
   (STS-Nc only).  The fixed stuff, which is purely overhead, is N/3-1
   columns for STS-Nc.  Thus, STS-1 and STS-3c do not have any fixed
   stuff, STS-12c has three columns of fixed stuff, and so on.

   The POH of an STS-1 or STS-Nc is always nine bytes in nine rows. The
   payload capacity of an STS-1 is 86 columns (774 bytes) per frame.
   The payload capacity of an STS-Nc is (N*87)-(N/3) columns per frame.
   Thus, the payload capacity of an STS-3c is (3*87 - 1)*9 = 2,340
   bytes per frame. As another example, the payload capacity of an STS-
   192c is 149,760 bytes, which is exactly 64 times larger than the
   STS-3c.

   There are 8,000 SONET frames per second. Therefore, the SPE size,
   (POH plus payload capacity) of an STS-1 is 783*8*8,000 = 50.112
   Mb/s. The SPE size of a concatenated STS-3c is 2,349 bytes per frame
   or 150.336 Mb/s. The payload capacity of an STS-192c is 149,760
   bytes per frame, which is equivalent to 9,584.640 Mb/s. Table 2
   lists the SPE and payload rates supported.






Malis, et al.           Expires February 2002               [Page 21]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   SONET STS Level     STS-1   STS-3c  STS-12c    STS-48c   STS-192c
   SDH VC Level            -     VC-4  VC-4-4c   VC-4-16c   VC-4-64c
   Payload Size(Bytes)   774    2,340    9,360     37,440    149,760
   Payload Rate(Mb/s) 49.536  149.760  599.040  2,396.160  9,584.640
   SPE Size(Bytes)       783    2,349    9,396     37,584    150,336
   SPE Rate(Mb/s)     50.112  150.336  601.344  2,405.376  9,621.504

                      Table 2. Payload Size and Rate


   To support circuit emulation, the entire SPE of a SONET STS or SDH
   VC level is encapsulated into packets, using the encapsulation
   defined in section 5, for carriage across packet-switched networks.


Appendix B. ECC-6 Definition

   ECC-6 is an Error Correction Code to protect the CEM header.  This
   provides single bit correction and the ability to detect up to two
   bit errors.


   Error Correction Code:


   |---------------Header bits 0-25 -------------------| ECC-6 code|
   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 2
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1 1 1 1 1 0 0 0 1 0 0 0 1 1 1 1 1 0 1 0 0 0 1 0 1 1|1 0 0 0 0 0|
   |1 1 1 1 0 1 0 0 0 1 0 0 1 0 0 0 0 1 0 1 1 1 1 1 1 1|0 1 0 0 0 0|
   |1 0 0 0 1 1 1 1 0 0 1 0 1 1 1 0 0 0 1 1 1 1 0 0 1 1|0 0 1 0 0 0|
   |0 1 0 0 1 1 1 1 0 0 0 1 1 0 0 1 1 1 1 1 0 0 1 1 0 1|0 0 0 1 0 0|
   |0 0 1 0 0 0 1 0 1 1 1 1 1 1 0 0 1 1 1 1 1 0 1 0 1 0|0 0 0 0 1 0|
   |0 0 0 1 0 0 0 1 1 1 1 1 0 0 1 1 0 0 1 1 0 1 1 1 1 1|0 0 0 0 0 1|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                      Figure 7. ECC-6 Check Matrix X


   The ECC-6 code protects the 32 bit CEM header as follows:

   The encoder generates the 6 bit ECC using the matrix shown in Figure
   7.  In brief, the encoder builds another 26 column by 6 row matrix
   and calculates even parity over the rows.  The matrix columns
   represent CEM header bits 0 through 25.

   Denote each column of the ECC-6 check matrix by X[], and each column
   of the intermediate encoder matrix as Y[].  CEM[] denotes the CEM
   header and ECC[] is the error correction code that is inserted into
   CEM header bits 26 through 31.



Malis, et al.           Expires February 2002               [Page 22]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   for i = 0 to 25 {
        if CEM[i] = 0 {
                Y[i] = 0;
        } else {
                Y[i] = X[i];
        }
   }

   In other words, for each CEM header bit (i) set to 1, set the
   resulting matrix column Y[i] according to Figure 7.

   The final ECC-6 code is calculated as even parity of each row in Y
   (i.e. ECC[k]=CEM[25+k]=even parity of row k).

   The receiver also uses matrix X to calculate an intermediate matrix
   YË based on all 32 bits of the CEM header.  Therefore YË is 32
   columns wide and includes the ECC-6 code.

   for i = 0 to 31 {
        if CEM[i] = 0 {
                YË[i] = 0;
        } else {
                YË[i] = X[i];
        }
   }

   The receiver then appends the incoming ECC-6 code to Y as column 32
   (ECC[0] should align with row 0) and calculates even parity for each
   row.  The result is a single 6 bit column Z.  If all 6 bits are 0,
   there are no bit errors (or at least no detectable errors).
   Otherwise, it uses Z to perform a reverse lookup on X[] from Figure
   7.  If Z matches column X[i], then there is a single bit error.  The
   receiver should invert bit CEM[i] to correct the header.  If Z fails
   to match any column of X, then the CEM header contains more than one
   bit error and the CEM packet MUST be discarded.

   Note that the ECC-6 code provides single bit correction and 2-bit
   detection of errors within the received ECC-6 code itself


Full Copyright Statement

   Copyright (C) The Internet Society (2001). All Rights Reserved. This
   document and translations of it may be copied and furnished to
   others, and derivative works that comment on or otherwise explain it
   or assist in its implementation may be prepared, copied, published
   and distributed, in whole or in part, without restriction of any
   kind, provided that the above copyright notice and this paragraph
   are included on all such copies and derivative works. However, this
   document itself may not be modified in any way, such as by removing
   the copyright notice or references to the Internet Society or other
   Internet organizations, except as needed for the purpose of
   developing Internet standards in which case the procedures for

Malis, et al.           Expires February 2002               [Page 23]


                SONET/SDH Circuit Emulation Over MPLS       July 2001


   copyrights defined in the Internet Standards process must be
   followed, or as required to translate it into languages other than
   English.

   The limited permissions granted above are perpetual and will not be
   revoked by the Internet Society or its successors or assigns.

   This document and the information contained herein is provided on an
   "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
   TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
   BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
   HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Acknowledgement

   Funding for the RFC Editor function is currently provided by the
   Internet Society.




































Malis, et al.           Expires February 2002               [Page 24]