[Search] [txt|pdf|bibtex] [Tracker] [Email] [Diff1] [Diff2] [Nits]

Versions: 01 02 03                                                      
                                             Bala Rajagopalan
Internet Draft                                 Tellium, Inc.
draft-many-ip-optical-framework-02.txt       James Luciani
Expires on: 5/14/2001                          Tollbridge Technologies
                                             Daniel Awduche
                                               UUNET (MCI Worldcom)
                                             Brad Cain, Bilel Jamoussi
                                               Nortel Networks
                                             Debanjan Saha
                                               Tellium, Inc.

                   IP over Optical Networks: A Framework

Status of this Memo

   This document is an Internet-Draft and is in full conformance with
   all provisions of Section 10 of RFC2026 except that the right to
   produce derivative works is not granted.

   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

   The list of current Internet-Drafts can be accessed at

   The list of Internet-Draft Shadow Directories can be accessed at

1. Abstract

   The Internet transport infrastructure is moving towards a model of
   high-speed routers interconnected by optical core networks. A
   consensus has emerged in the industry on utilizing IP-based
   protocols for the optical control plane. At the same time, there is
   ongoing activity in defining architectural models for IP transport
   over optical networks, specifically, the routing and signaling
   aspects. This draft defines a framework for IP over Optical
   networks, considering both the IP-based control plane for optical
   networks as well as IP transport over optical networks (together
   referred to as "IP over optical networks").

                         Expires on 5/24/2001             Page 1 of 36


                             Table of Contents

   1.  Abstract  ------------------------------------------------- 1
   2.  Conventions Used in this Document ------------------------- 3
   3.  Introduction ---------------------------------------------- 3
   4.  Terminology and Concepts ---------------------------------- 4
   5.  The Network Model ----------------------------------------- 7
     5.1  Network Interconnection -------------------------------- 7
     5.2  Control Structure -------------------------------------- 9
   6.  IP over Optical Service Models ---------------------------- 9
     6.1 Domain Services Model ---------------------------------- 10
     6.2 Unified Services Model --------------------------------- 12
     6.3 Which Service Model? ----------------------------------- 13
     6.4 What are the Possible Services? ------------------------ 13
   7.  IP Transport over Optical Networks ----------------------- 14
     7.1 Interconnection Models --------------------------------- 14
     7.2 Routing Approaches ------------------------------------- 15
     7.3 Signaling Related  ------------------------------------- 18
     7.4 End-to-End Protection Models --------------------------- 19
   8.  IP-Based Optical Control Plane Issues -------------------- 20
     8.1 Addressing  -------------------------------------------- 21
     8.2 Neighbor Discovery ------------------------------------- 22
     8.3 Topology Discovery --------------------------------------23
     8.4 Restoration Models ------------------------------------- 24
     8.5 Route Computation  ------------------------------------- 25
     8.6 Signaling Issues  -------------------------------------- 27
     8.7 Optical Internetworking -------------------------------  29
   9.  Other Issues --------------------------------------------- 30
     9.1 WDM and TDM in the Same Network ------------------------ 30
     9.2 Wavelength Conversion ---------------------------------- 30
     9.3 Service Provider Peering Points ------------------------ 31
     9.4 Rate of Lightpath Set-Up ------------------------------- 31
     9.5 Distributed vs Centralized Provisioning ---------------  32
  10.  Evolution Path for IP over Optical Architecture ---------  33
  11.  Security Considerations ---------------------------------  33
  12.  Summary and Conclusions ---------------------------------  34
  13.  References ----------------------------------------------  34
  14.  Acknowledgements ----------------------------------------  35

                          Expires on 5/24/01              Page 2 of 36


2. Conventions used in this document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   this document are to be interpreted as described in RFC 2119.

3. Introduction

   Optical network technologies are evolving rapidly in terms of
   functions and capabilities. The increasing importance of optical
   networks is evidenced by the copious amount of attention focused on
   IP over optical networks and related photonic and electronic
   interworking issues by all the major network service providers,
   telecommunications equipment vendors, and standards organizations.

   It has been realized that optical networks must be survivable,
   flexible, and controllable. There is, therefore, an ongoing trend to
   introduce intelligence in the control plane of optical networks to
   make them more versatile [1]. An essential attribute of intelligent
   optical networks is the capability to instantiate and route optical
   layer connections in real-time or near real-time, and to provide
   capabilities that enhance network survivability. Furthermore, there
   is a need for multi-vendor optical network interoperability, when an
   optical network may consist of interconnected vendor-specific
   optical sub-networks.

   The OTN must also be versatile because some service providers may
   offer generic optical layer services that may not be client-
   specific. It would therefore be necessary to have an optical network
   control layer that can handle such generic optical services.

   There is general consensus in the industry that the optical network
   control plane should utilize IP-based protocols for dynamic
   provisioning and restoration of lightpaths within and across optical
   sub-networks. This is based on the practical view that signaling and
   routing mechanisms developed for IP traffic engineering applications
   could be re-used in optical networks. Nevertheless, the issues and
   requirements that are specific to optical networking must be
   understood to suitably adopt the IP-based protocols. This is
   especially the case for restoration.  Also, there are different
   views on the model for interaction between the optical network and
   client networks, such as IP networks. Reasonable architectural
   alternatives in this regard must be supported, with an understanding
   of their pros and cons.

   Thus, there are two fundamental issues related to IP over optical
   networks. The first is the adaptation and reuse of IP control plane
   protocols within the optical network control plane, irrespective of
   the types of digital clients that utilize the optical network. The
   second is the transport of IP traffic through an optical network
   together with the control and coordination issues that arise

                          Expires on 5/24/01              Page 3 of 36


   This draft defines a framework for IP over optical networks covering
   the requirements and mechanisms for establishing an IP-centric
   optical control plane, and the architectural aspects of IP transport
   over optical networks. In this regard, it is recognized that the
   specific capabilities required for IP over optical networks would
   depend on the services expected at the IP-optical interface as well
   as the optical sub-network interfaces.  Depending on the specific
   operational requirements, a progression of capabilities is possible,
   reflecting increasingly sophisticated interactions at these
   interfaces. This draft therefore advocates the definition of
   "capability sets" that define the evolution of functionality at the
   interfaces as more sophisticated operational requirements arise.

   This draft is organized as follows. In the next section, terminology
   covering certain concepts related to this framework are described.
   In Section 5, the network model pertinent to this framework is
   described. The service model and requirements for IP-optical, and
   multi-vendor optical internetworking are described in Section 6.
   This section presently considers certain general requirements.
   Specific operational requirements may be accommodated in this
   section as they arise.  Section 7 considers the architectural models
   for IP-optical interworking, describing the pros and cons of each
   model. It should be noted that it is not the intent of this draft to
   promote any particular model over the others. However, particular
   aspects of the models that may make one approach more appropriate
   than another in certain circumstances are described. Section 8
   describes IP-centric control plane mechanisms for optical networks,
   covering signaling and routing issues in support of provisioning and
   restoration. Section 9 describes certain specialized issues in
   relation to IP over optical networks.  The approaches described in
   Section 7 and 8 range from the relatively simple to the
   sophisticated. Section 10 describes a possible evolution path for IP
   over optical networking capabilities in terms of increasingly
   sophisticated functionality supported. Section 11 considers security
   aspects. Finally, summary and conclusion are presented in Section

4. Terminology and Concepts

   This section introduces some terminology for describing common
   concepts in IP over optical networks pertinent to this framework.


   Wavelength Division Multiplexing (WDM) is a technology that allows
   multiple optical signals operating at different wavelengths to be
   multiplexed onto a single fiber so that they can be transported in
   parallel through the fiber. In general, each optical wavelength may
   carry digital client payloads at a different data rate (e.g., OC-3c,

                          Expires on 5/24/01              Page 4 of 36


   OC-12c, OC- 48c, OC-192c, etc.) and in a different format (SONET,
   Ethernet, ATM, etc.)

   Optical cross-connect (OXC)

   An OXC is a space-division switch that can switch an optical data
   stream on an input port to a output port. Such a switch may have
   optical-electrical conversion at the input port and electrical-
   optical conversion at the output port, or it can be all-optical. An
   OXC is assumed to have a control-plane processor that implements
   signaling and routing protocols necessary for realizing an optical

   Optical channel trail or Lightpath

   An optical channel trail is a point-to-point optical layer
   connection between two access points in an optical network. In this
   draft, the term "lightpath" is used interchangeably with optical
   channel trail.

   Optical mesh sub-network

   An optical sub-network, as considered in this framework, is a
   network of OXCs that supports end-to-end networking of optical
   channel trails  providing functionality like routing, monitoring,
   grooming, and protection and restoration of optical channels. The
   interconnection of OXCs in this network can be based on a general
   topology (also called "mesh" topology) Underlying this network could
   be the following:

   (a) An optical multiplex section (OMS) layer network : The optical
       multiplex section layer provides the transport of the optical
       channels.  The information contained in this layer is a data
       stream comprising a set of n optical channels, which have a
       defined aggregate bandwidth.

   (b) An optical transmission section (OTS) layer network : This
       provides functionality for transmission of the optical signal on
       optical media of different types.

   This framework does not address the interaction between the optical
   sub-network and the OMS, or between the OMS and OTS layer networks.

   Optical mesh network

   An optical mesh network, as considered in draft, is a mesh-connected
   collection of optical sub-networks.

                          Expires on 5/24/01              Page 5 of 36


   Wavelength continuity property

   A lightpath is said to satisfy the wavelength continuity property if
   it is transported over the same wavelength end-to-end. Wavelength
   continuity is required in optical  networks with no wavelength
   conversion feature.

   Trust domain

   A trust domain is a network under a single technical administration
   in which most nodes in the network are considered to be secure or
   trusted in some fashion.  An example of a trust domain is a campus
   or corporate network in which all routing protocol packets are
   considered to be authentic, without the need for additional security
   schemes to prevent unauthorized access to the network
   infrastructure.  Generally, the "single" administrative control rule
   may be relaxed in practice if a set of administrative entities agree
   to trust one another to form an enlarged heterogeneous trust domain.
   However, to simplify the discussions in this draft, it will assumed,
   without loss of generality, that the term trust domain applies to a
   single administrative entity.


   For the purpose of this document, the term flow will be used to
   represent the smallest separable stream of data, as seen by an
   endpoint (source or destination node).  It is to be noted that the
   term flow is heavily overloaded in the networking literature. Within
   the context of this document, it may be convenient to consider a
   wavelength as a flow under certain circumstances. However, if there
   is a method to partition the bandwidth of the wavelength, then each
   partition may be considered a flow, for example by quantizing time
   into some nicely manageable intervals, it may be feasible to
   consider    each quanta of time within a given wavelength as a flow.

   Traffic Trunk

   A abstraction of traffic flow that follows the same path between two
   access points which allows some characteristics and attributes of
   the traffic to be parameterized.

                          Expires on 5/24/01              Page 6 of 36


5. The Network Model

5.1  Network Interconnection

   The network model considered in this draft consists of IP routers
   attached to an optical core network, and connected to their peers
   over dynamically established switched lightpaths. The optical core
   itself is assumed to be incapable of processing individual IP

   The optical network is assumed to consist of multiple optical
   sub-networks interconnected by optical links in a general topology
   (referred to as an optical mesh network). This network may be multi-
   vendor. In the near term, it may be expected that each sub-network
   will consist of a single vendor switches. In the future, as
   standardization efforts mature, each optical sub-network may in fact
   contain optical switches from different vendors. In any case, each
   sub-network itself is assumed to be mesh-connected. In general, it
   can be expected that topologically adjacent OXCs in an optical mesh
   network will be connected via multiple, parallel (bi-directional)
   optical links. This network model is shown in Figure 1.

   Here, an optical sub-network may consist entirely of all-optical
   OXCs or OXCs with optical-electrical-optical (OEO) conversion.
   Interconnection between sub-networks is assumed to be through
   compatible physical interfaces, with suitable optical-electrical
   conversions where necessary. The routers that have direct physical
   connectivity with the optical network are referred to as "edge
   routers". As shown in the figure, other client networks (e.g., ATM)
   may connect to the optical network.

   The switching function in an OXC is controlled by appropriately
   configuring the cross-connect fabric. Conceptually, this may be
   viewed as setting up a cross-connect table whose entries are of the
   form <input port i, output port j>, indicating that the data stream
   entering input port i will be switched to output port j.  A
   lightpath from an ingress port in an OXC to an egress port in a
   remote OXC is established by setting up suitable cross-connects in
   the ingress, the egress and a set of intermediate OXCs such that a
   continuous physical path exists from the ingress to the egress port.
   Optical paths are assumed to be bi-directional, i.e., the return
   path from the egress port to the ingress port is routed along the
   same set of intermediate ports as the forward path.

   Multiple data streams output from an OXC may be multiplexed onto an
   optical link using WDM technology. The WDM functionality may exist
   outside of the OXC, and be transparent to the OXC. Or, this function
   may be built into the OXC. In the latter case, the cross-connect
   table   (conceptually) consists of pairs of the form, <{input port
   i, Lambda(j)}, {output port k, Lambda(l)}>. This indicates that the
   data stream received on wavelength Lambda(j) over input port i is
   switched to output port k on Lambda(l). Automated establishment of

                          Expires on 5/24/01              Page 7 of 36


   lightpaths involves setting up the cross-connect table entries in
   the appropriate OXCs in a coordinated manner such that the desired
   physical path is realized.

                              Optical Network
                           |                                       |
      +--------------+     |                                       |
      |              |     | +------------+        +------------+  |
      |   IP         |     | |            |        |            |  |
      |   Network    +--UNI--+   Optical  +---NNI--+   Optical  |  |
      |                    | | Subnetwork |        | Subnetwork |  |
      +--------------+     | |            |  +-----+            |  |
                           | +------+-----+  |     +------+-----+  |
                           |        |        |            |        |
                           |       NNI      NNI          NNI       |
      +--------------+     |        |        |            |        |
      |              |     | +------+-----+  |     +------+-----+  |
      |   IP         +--UNI--|            +--+     |            |  |
      |   Network    |     | |   Optical  |        |   Optical  |  |
      |              |     | | Subnetwork +---NNI--+ Subnetwork |  |
      +--------------+     | |            |        |            |  |
                           | +------+-----+        +------+-----+  |
                           |        |                     |        |
                                    |                     |
                                    |                     |
                             +------+-------+     +------------+
                             |              |     |            |
                             | Other Client |     |Other Client|
                             |   Network    |     |   Network  |
                             | (e.g., ATM)  |     |            |
                             +--------------+     +------------+

                          Figure 1: Optical Network Model

   Under this network model, a switched lightpath must be established
   between a pair of IP routers before they can communicate. This
   lightpath might traverse multiple optical sub-networks and be
   subject to different provisioning and restoration procedures in each
   sub-network.  The IP-based control plane issue is that of designing
   standard signaling and routing protocols for coherent end-to-end
   provisioning and restoration of lightpaths across multiple optical
   sub-networks. Similarly, IP transport over such an optical network
   involves determining IP reachability and seamless establishment of
   paths from one IP endpoint to another over an optical core.

                          Expires on 5/24/01              Page 8 of 36


5.2  Control Structure

   There are two logical control interfaces identified in Figure 1.
   These are the client-optical network interface, and the optical sub-
   network interface. These interfaces are also referred to as the
   User-Network Interface (UNI) and the Network-Network Interface(NNI).
   The distinction between these interfaces arises out of the type and
   amount of control flow across them. The UNI represents a technology
   boundary between the client and optical networks. Thus, the control
   flow across the UNI is dependent of the set of services defined
   across it and the manner in which the services may be accessed. The
   service models are described in Section 7. Here, we merely note that
   since the optical network implements an IP-based control plane, it
   is possible in principle to harmonize the control flow across the
   UNI and the NNI and eliminate the distinction between them. On the
   other hand, it may be required to minimize control flow information,
   especially routing-related information, over UNI.  In this case, UNI
   and NNI may look different in some respects. In this draft, these
   interfaces are treated as distinct.

   Each of these interfaces can also be categorized as public or
   private depending upon context and service models. If UNI (or NNI)
   is private, then routing information (ie, topology state
   information) can be exchanged across it. If UNI (or NNI) is public,
   then routing information is not exchanged across it, or such
   information may be exchanged across it with very explicit
   restrictions (including for example abstraction, filtration, etc).
   Thus, different relationships (e.g., peer or over-lay, Section 7)
   may occur across private and public logical interfaces.

   The physical control structure used to realize these logical
   interfaces may vary. For instance, for the UNI, some of the
   possibilities are:

   1.Direct interface: An in-band or out-of-band IP control channel
     (IPCC) may be implemented between an edge router and each OXC
     that it connects to. This control channel is used for exchanging
     signaling and routing messages between the router and the OXC.
     With a direct interface, the edge router and the OXC it connects
     to are peers in the control plane. This is shown in Figure 2. The
     type of routing and signaling information exchanged across  the
     direct interface would vary depending on the service definition.
     This issue is dealt with in the next section. Some choices for
     the routing protocol are OSPF/ISIS (with traffic engineering
      extensions) or BGP. Other directory-based routing information
     exchanges are also possible. Some of the signaling protocol
     choices are adaptations of RSVP-TE or CR-LDP. The details of how
     the IP control channel is realized is outside the scope of this

   2.Indirect interface: An out-of-band IP control channel may be
     implemented between the client and a device in the optical network
     to signal service requests and responses. For instance, a

                          Expires on 5/24/01              Page 9 of 36


     management system or a server in the optical network may receive
     service requests from clients. Similarly, out-of-band signaling
     may be used between management systems in client and optical
     networks to signal service requests. In these cases, there is no
     direct control interaction between clients and respective
      OXCs. One reason to have an indirect interface would be that the
      OXCs and/or clients do not support a direct signaling interface.

   3. Provisioned interface: In this case, the optical network services
      are manually provisioned and there is no control interactions
      between the client and the optical network.

   +-----------------------------+      +-----------------------------+
   |                             |      |                             |
   |  +---------+   +---------+  |      |  +---------+   +---------+  |
   |  |         |   |         |  |      |  |         |   |         |  |
   |  | Routing |   |Signaling|  |      |  | Routing |   |Signaling|  |
   |  | Protocol|   |Protocol |  |      |  | Protocol|   |Protocol |  |
   |  |         |   |         |  |      |  |         |   |         |  |
   |  +-----+---+   +---+-----+  |      |  +-----+---+   +---+-----+  |
   |        |           |        |      |        |           |        |
   |        |           |        |      |        |           |        |
   |     +--+-----------+---+    |      |     +--+-----------+---+    |
   |     |                  |    |      |     |                  |    |
   |     |     IP Layer     +......IPCC.......+     IP Layer     |    |
   |     |                  |    |      |     |                  |    |
   |     +------------------+    |      |     +------------------+    |
   |                             |      |                             |
   |         Edge Router         |      |             OXC             |
   +-----------------------------+      +-----------------------------+

                            Figure 2: Direct Interface

   Although different control structures are possible, further
   descriptions in this framework assume direct interfaces for IP-
   optical and optical sub-network control interactions.

6. IP over Optical Service Models and Requirements

   In this section, the service models and requirements at the IP-
   optical UNI and the optical sub-network NNI are considered. Two
   general models have emerged for the services at the IP-optical
   interface (which can also be applied at the optical sub-network
   interface). These models are as follows.

6.1  Domain Services Model

   Under this model, the optical network primarily offers high
   bandwidth connectivity in the form of lightpaths [2]. Standardized
   signaling across the UNI (Figure 1) is used to invoke the following

                          Expires on 5/24/01             Page 10 of 36


   1. Lightpath creation: This service allows a lightpath with the
      specified attributes to be created between a pair of termination
      points in the optical network. Lightpath creation may be subject
      to network-defined policies (e.g., connectivity restrictions) and
      security procedures.

   2. Lightpath deletion: This service allows an existing lightpath to
      be deleted.

   3. Lightpath modification: This service allows certain parameters of
      the lightpath to be modified.

   4. Lightpath status enquiry: This service allows the status of
      certain parameters of the lightpath (referenced by its ID) to be
      queried by the router that created the lightpath.

   Additionally, the following address resolution procedures may be
   made available over the UNI (more sophisticated routing information
   exchange over the UNI is also possible, as described later and
   covered in more detail in [3]):

   1. Client Registration: This allows a client to register its
      address(es) and user group identifier(s) with the optical
      network. The registered address may be of different types, IP,
      ATM NSAP, E.164, etc. The optical network associates the client
      address and user group ID with an optical-network-administered

   2. Client De-Registration: This allows a client to withdraw its
      address(es) and user group identifier(s) from the optical

   3. Query: This allows a client to supply another clientÆs native
      address (e.g., ATM) and user group ID, and get back an optical-
      network-administered address that can be used in lightpath create

   An end-system discovery procedure may be used over the UNI to verify
   local port connectivity between the optical and client devices, and
   allows each device to bootstrap the UNI control channel. Finally, a
   "service discovery" procedure may be employed as a precursor to
   obtaining UNI services. Service discovery allows a client to
   determine the static parameters of the interconnection with the
   optical network, including the UNI signaling protocols supported.
   The protocols for neighbor and service discovery are different from
   the UNI signaling protocol itself (for example, see LMP [4]).

   With regard to address resolution, the registration and de-
   registration procedures may be implemented using service discovery
   mechanisms. The query mechanism may be implemented as an additional
   UNI signaling procedure.

                          Expires on 5/24/01             Page 11 of 36


   Because a small set of well-defined services is offered across UNI,
   the signaling protocol requirements are minimal. Specifically, the
   signaling protocol is required to convey a few messages with certain
   attributes point-to-point between the router and the optical
   network. Such a protocol may be based on RSVP-TE or LDP, or even a
   messaging application over a TCP connection.

   The optical domain services model does not deal with the type and
   nature of routing protocols within the optical network. Furthermore,
   the integration of multiple, optical sub-networks across NNI will
   require the specification of a standard routing protocol, say, BGP.

   The optical domain services model would result in the establishment
   of a lightpath topology between routers at the edge of the optical
   network. The resulting overlay model for IP over optical networks
   is discussed in Section 7.

6.2  Unified Service Model

   Under this model, the IP and optical networks are treated together
   as a single integrated network that is managed and traffic
   engineered in a unified manner. In this regard, the OXCs are treated
   just like any other router as far as the control plane is
   considered. Thus, from a routing and signaling point of view, there
   is no distinction between UNI, NNI and any other router-to-router
   interface. It is assumed that this control plane is MPLS-based, as
   described in [1].

   Under the unified service model, optical network services are
   obtained implicitly during end-to-end MPLS signaling. Specifically,
   an edge router can create a lightpath with specified attributes, or
   delete and modify lightpaths as it creates label-switched paths
   (LSPs). In this regard, the services obtained from the optical
   network are similar to the domain services model. These services,
   however, may be invoked in a more seamless manner as compared to the
   domain services model. For instance, in principle, a remote router
   could compute an end-to-end path across the optical network
   utilizing, say, OSPF with traffic engineering extensions [5]. It can
   then establish an LSP across the optical network. But the edge
   routers must still recognize that an LSP  across the optical network
   is a lightpath, or a conduit for multiple LSPs. The concept of
   "forwarding adjacency" can be used to specify virtual links across
   optical networks in routing protocols such as OSPF [6]. In essence,
   once a lightpath is established across an optical network between
   two edge routers, it can be advertised as a forwarding adjacency (a
   virtual link) between these routers.  Thus, from a data plane point
   of view, the lightpaths result in a overlay between edge routers.
   The decisions as to when to create such lightpaths, and the
   bandwidth management for these lightpaths is identical in both the
   domain services model and the unified service model. The routing and
   signaling models for unified services is described in Section 7.

                          Expires on 5/24/01             Page 12 of 36


6.3  Which Service Model?

   The pros and cons of the above service models can be debated at
   length, but the approach recommended in this framework is to define
   routing and signaling mechanisms in support of both. As pointed out
   above, signaling for service request can be unified to cover both
   models. The developments in GMPLS signaling [7] for the unified
   service model and its adoption for UNI signaling under the domain
   services model [8] essentially supports this view. The significant
   difference between the service models, however, is in routing
   protocols, as described in Section 7.

6.4 What are the Possible Services?

   Specialized services may be built atop the point-to-point
   connectivity service offered by the optical network. For example,

6.4.1  Virtual Private Networks (VPNs)

   Given that the data plane between IP routers over an optical network
   is an overlay, it is easy to imagine a virtual private network of
   lightpaths that interconnect routers (or any other set of clients).
   Indeed, in the case where the optical network provides connectivity
   for multiple sets of external client networks, there has to be a
   way to enforce routing policies that ensure routing separation
   between different sets of clients (i.e., VPN service).

7. IP transport over Optical Networks

   To examine the architectural alternatives for IP over optical
   networks, it is important to distinguish between the data and
   control planes over the UNI. As described in Section 6, the optical
   network provides a service to external entities in the form of fixed
   bandwidth transport pipes (optical paths). IP routers at the edge of
   the optical networks must necessarily establish such paths before
   communication at the IP layer can begin. Thus, the IP data plane
   over optical networks is realized over an overlay network of optical
   paths. On the other hand, IP routers and OXCs can have a peer
   relation on the control plane, especially for the implementation of
   a routing protocol that allows dynamic discovery of IP endpoints
   attached to the optical network. The IP over optical network
   architecture is defined essentially by the organization of the
   control plane. The assumption in this framework is that an MPLS-
   based control plane [1] is used. Depending on the service
   model(Section 6), however, the control planes in the IP and optical
   networks can be loosely or tightly coupled. This coupling determines

   o the details of the topology and routing information advertised by
     the optical network across UNI;

   o Level of control that IP routers can exercise in selecting
     specific paths for connections across the optical network; and

                          Expires on 5/24/01             Page 13 of 36


   o Policies regarding the dynamic provisioning of optical paths
     between routers. This includes access control, accounting and
     security issues.

   The following interconnection models are then possible:

7.1 Interconnection Models

7.1.1 The Peer Model

   Under the peer model, the IP/MPLS layers act as peers of the optical
   transport network, such that a single routing protocol instance runs
   over both the IP/MPLS and optical domains. Presumably a common IGP
   such as OSPF or IS-IS, with appropriate extensions, can be used to
   distribute topology information [3]. In the case of OSPF, opaque
   LSAs can be used to advertise topology state information. In the
   case of IS-IS, extended TLVs will have to be defined to propagate
   topology state information. One tacit assumption here is that a
   common addressing scheme will also be used for the optical and IP
   networks. A common address space can be trivially realized by using
   IP addresses in both IP and optical domains. Thus, the optical
   network elements become IP addressable entities as noted in [1].

7.1.2 The Overlay Model

   Under the overlay model, the IP/MPLS routing, topology distribution,
   and signaling protocols are independent of the routing, topology
   distribution, and signaling protocols at the optical layer. This
   model is conceptually similar to the classical IP over ATM or MPOA
   models, but applied to a optical sub-network directly. In the
   overlay model, topology distribution, path computation and signaling
   protocols would have to be defined for the optical domain. In
   certain circumstances, it may also be feasible to statically
   configure the optical channels that provide connectivity in the
   overlay model through network management. Static configuration is,
   however, unlikely to scale in very large networks.

7.1.3  The Augmented Model

   Under the augmented model, there are actually separate routing
   instances in the IP and optical domains, but information from one
   routing instance is passed through the other routing instance. For
   example, external IP addresses could be carried within the optical
   routing protocols to allow reachability information to be passed to
   IP clients.

   The routing approaches corresponding to these interconnection models
   are described below.

                          Expires on 5/24/01             Page 14 of 36


7.2 Routing Approaches

7.2.1 Integrated Routing

   This routing approach supports the peer model described above. Under
   this approach, the IP and optical networks are assumed to run the
   same instance of an IP routing protocol, e.g., OSPF with suitable
   "optical" extensions.  These extensions must capture optical link
   parameters, and any constraints that are specific to optical
   networks. The topology and link state information maintained by all
   nodes (OXCs and routers) is identical. This permits a router to
   compute an end-to-end path to another router across the optical
   network. Suppose the path computation is triggered by the need to
   route a label switched path (LSP). Such an LSP can be established
   using MPLS signaling, e.g., RSVP-TE or CR-LDP. When the LSP is
   routed over the optical network, a lightpath must be established
   between two edge routers. This lightpath is in essence a tunnel
   across the optical network, and may have capacity much larger than
   that required to route the first LSP. Thus, it is essential that
   other routers in the network realize the availability of resources
   in the lightpath for other LSPs to be routed over it. The lightpath
   must therefore be advertised as a virtual link in the topology.

   The notion of "forwarding adjacency" (FA) described in [6] is
   essential in propagating lightpath information to routers. An FA is
   essentially a virtual link advertised into a link state routing
   protocol. Thus, an FA could be described by the same parameters that
   define resources in any regular link. While it is necessary to
   specify the mechanism for creating an FA, it is not necessary to
   specify how an FA is used by the routing scheme. Once an FA is
   advertised in a link state protocol, its usage for routing LSPs is
   defined by the route computation and traffic engineering algorithms

   It should be noted that at the IP-optical interface, the physical
   ports over which routers are connected to OXCs define the
   connectivity and resource availability. Suppose a router R1 is
   connected to OXC O1 over two ports, P1 and P2. Under integrated
   routing, the connectivity between R1 and O1 over the two ports would
   have been captured in the link state representation of the network.
   Now, suppose an FA at full port bandwidth is created from R1 to
   another router R2 over port P1. While this FA is advertised as a
   virtual link between R1 and R2, it is also necessary to remove the
   link R1-O1 (over P1) from the link state representation since that
   port is no longer available for creating a lightpath. Thus, as FAs
   are created, an overlaid set of virtual links is introduced into the
   link state representation, replacing the links previously advertised
   at the IP-Optical interface. Finally, the details of the optical
   network captured in the link state representation is replaced by a
   network of FAs. In this regard, there is a great deal of similarity
   between integrated routing   and domain-specific routing (described
   next). Both ultimately deal with the creation of the overlaid
   lightpath topology to meet the traffic engineering objectives.

                          Expires on 5/24/01             Page 15 of 36


7.2.2 Domain-Specific Routing

   This routing approach supports the augmented interconnection model.
   Under this approach, routing within the optical and IP domains are
   separated, with a standard routing protocol running between domains.
   This is similar to the IP inter-domain routing model. Two choices
   for this are considered.  Domain-Specific Routing using BGP

   The inter-domain IP routing protocol, BGP [9], may be adapted for
   exchanging routing information between IP and optical domains. This
   would allow the routers to advertise IP address prefixes within
   their    network to the optical network and to receive external IP
   address prefixes from the optical network. The optical network
   transports the reachability information from one IP network to
   others. For instance, edge routers and OXCs can run exterior BGP
   (EBGP).  Within the optical network, interior BGP (IBGP) is used
   between border OXCs within the same sub-network, and EBGP is used
   between sub-networks (over NNI, Figure 1).

   Under this scheme, it is necessary to identify the egress points in
   the optical network corresponding to externally reachable IP
   addresses. This is due to the following. Suppose an edge router
   desires to establish an LSP to a destination reachable across the
   optical network. It could directly request a lightpath to that
   destination, without explicitly specifying the egress optical port
   for   the lightpath as the optical network has knowledge of
   externally reachable IP addresses. However, if the same edge router
   has to establish another LSP to a different external destination, it
   must first determine whether there is a lightpath already available
   (with sufficient residual capacity) that leads to that destination.
   To identify this, it is necessary for edge routers to keep track of
   which egress ports in the optical network lead to which external
   destinations. Thus, a border OXC receiving external IP prefixes from
   an edge router through EBGP must include its own IP address as the
   egress point before propagating these prefixes to other border OXCs
   or   edge routers. An edge router receiving this information need
   not propagate the egress address further, but it must keep the
   association   between external IP addresses and egress OXC
   addresses. Specific BGP mechanisms for propagating egress OXC
   addresses are to be determined,  considering prior examples
   described in [10]. When VPNs are implemented, the address prefixes
   advertised by the border OXCs must be accompanied by some VPN
   identification (for example, VPN IPv4 addresses, as defined in [10],
   may be used).  Domain Specific Routing using OSPF/ISIS

   The routing information exchanged across the IP-optical UNI could be
   summarized using a hierarchical routing protocol such as OSPF/ISIS.
   The following description is based on OSPF.

                          Expires on 5/24/01             Page 16 of 36


   OSPF supports a two-level hierarchical routing scheme through the
   use of OSPF areas. Routing within each area is flat, while detailed
   knowledge of an areaÆs topology is hidden from all other areas.
   Routers attached to two or more areas are called Area Border Routers
   (ABRs). ABRs propagate IP addressing information from one area to
   another using summary LSAs. Within an OSPF routing domain, all areas
   are attached directly to a special area called the OSPF backbone
   area.   The exchange of information between areas can be controlled
   to implement domain specific routing in each area. For instance, the
   optical network can be a collection of one or more areas in which
   certain link parameters and information specific to optical networks
   is incorporated into a version of OSPF. The client network (e.g.,
   IP) could be separate OSPF area(s), running OSPF with TE extensions.
   The summary LSAs exchanged between the optical and client areas can
   be designed such that optical domain specific information is hidden
   from    client networks while providing adequate routing information
   for end-to-end routing of lightpaths.

   While the use of BGP or OSPF/ISIS allows edge routers to learn about
   reachability of destinations across the optical network, the
   determination of how many lightpaths to establish and to what egress
   points are traffic engineering decisions.

7.2.3  Overlay Routing

   This routing approach supports the overlay interconnection model.
   Under this approach, overlay mechanism that allows edge routers to
   register and query for external addresses is implemented. This is
   similar to address resolution for IP over ATM. Under this approach,
   the optical network could implement a registry that allows edge
   routers to register IP addresses and VPN identifiers. An edge router
   may be allowed to query for external addresses belonging to the same
   set of VPNs it belongs to. A successful query would return the
   address of the egress optical port through which the external
   destination can be reached.

   Because IP-optical interface connectivity is limited, the
   determination of how many lightpaths must be established and to what
   endpoints are traffic engineering decisions. Furthermore, after an
   initial set of such lightpaths are established, these may be used as
   adjacencies within VPNs for a VPN-wide routing scheme, for example,
   OSPF. With this approach, an edge router could first determine other
   edge routers of interest by querying the registry. After it obtains
   the appropriate addresses, an initial overlay lightpath topology may
   be formed. Routing adjacencies may then be established across the
   lightpaths and further routing information may be exchanged to
   establish VPN-wide routing.

   Routing approaches in optical networks are further described in [3].

                          Expires on 5/24/01             Page 17 of 36


7.3 Signaling-Related

7.3.1 The Role of MPLS

   It is possible to model wavelengths, and potentially TDM channels
   within a wavelength as "labels". This concept was proposed in [1],
   and ôgeneralizedö MPLS (GMPLS) mechanisms for realizing this are
   described in [7]. MPLS signaling protocols with traffic engineering
   extensions, such as RSVP-TE and CR-LDP can be used for signaling
   lightpath requests. In the case of the domain services model, these
   protocols can be adapted for UNI signaling [11, 12]. In the case of
   the unified services model, lightpath establishment occurs to
   support end-to-end LSP establishment using these protocols (with
   suitable GMPLS enhancements [13, 14]).

7.3.2 Signaling Models

   With the domain-services model, the signaling control plane in the
   IP and optical network are completely separate as shown in Figure 3
   below. This separation also implies the separation of IP and optical
   address spaces (even though the optical network would be using
   internal IP addressing). While RSVP-TE and LDP can be adapted for
   UNI signaling, the full functionality of these protocols will not be
   used. For example, UNI signaling does not require the specification
   of explicit routes [8]. On the other hand, based on the service
   attributes, new objects need to be signaled using these protocols as
   described in [11, 12].

        MPLS Signaling      UNI Signaling     MPLS or other signaling
   +-----------------------------+  |   +-----------------------------+
   |         IP Network          |  |   |       Optical Network       |
   |  +---------+   +---------+  |  |   |  +---------+   +---------+  |
   |  |         |   |         |  |  |   |  |         |   |         |  |
   |  | Router  +---+ Router  +-----+------+  OXC    +---+   OXC   |  |
   |  |         |   |         |  |  |   |  |         |   |         |  |
   |  +-----+---+   +---+-----+  |  |   |  +-----+---+   +---+-----+  |
   +-----------------------------+  |   +-----------------------------+
              Completely Separated Addressing and Control Planes

                 Figure 3: Domain Services Signaling Model

   With the unified services model, the addressing is common in the IP
   and optical networks and the respective signaling control are
   related, as shown in Figure 4. It is understood that GMPLS signaling
   is implemented in the IP and optical networks, using suitably
   enhanced RSVP-TE and CR-LDP protocols. But the semantics of services
   within the optical network may be different from that in the IP
   network. As an example, the protection services offered in the
   optical network may be different from that end-to-end protection

                          Expires on 5/24/01             Page 18 of 36


   services offered by the IP network. Another example is with regard
   to bandwidth. While the IP network may offer a continuum of
   bandwidths, the optical network will offer only discrete bandwidths.
   Thus, the signaling attributes and services are defined
   independently for IP and optical networks. The routers at the edge
   of the optical network must therefore identify service boundaries
   and perform suitable translations in the signaling messages crossing
   the IP-optical boundary. This must occur even though the signaling
   control plane in both networks are GMPLS-based and there is tighter
   coupling of the control plane as compared to the domain services

                        Service Boundary         Service Boundary
                              |                       |
   IP Layer GMPLS Signaling   | Optical Layer GMPLS   | IP Layer GMPLS
                              |                       |
      +--------+  +--------+  |  +-------+  +-------+ |  +--------+
      |        |  |        |  |  |       |  |       | |  |        |
      | IP LSR +--+ IP LSR +--+--+Optical+--+Optical+-+--+ IP LSR +---
      |        |  |        |  |  |  LSR  |  |  LSR  | |  |        |
      +-----+--+  +---+----+  |  +-----+-+  +---+---+ |  +--------+

                     Common Address Space, Service Translation

               Figure 4: Unified Services Signaling Model

   Thus, as illustrated in Figure 4, the signaling in the case of
   unified services is actually multi-layered. The layering is based on
   the technology and functionality. As an example, the specific
   adaptations of GMPLS signaling for SONET layer (whose functionality
   is transport) are described in [15].

7.4  End-to-End Protection Models

   Suppose an LSP is established from an ingress IP router to an egress
   router across an ingress IP network, a transit optical network and
   an egress IP network. If this LSP is to be afforded protection in
   the IP layer, how is the service coordinated between the IP and
   optical layers?

   Under this scenario, there are two approaches to end-to-end

7.4.1 Segment-Wise Protection

   The protection services in the IP layer could utilize optical layer
   protection services for the LSP segment that traverses the optical
   network. Thus, the end-to-end LSP would be treated as a
   concatenation of three LSP segments from the protection point of
   view: a segment in the ingress IP network, a segment in the optical

                          Expires on 5/24/01             Page 19 of 36


   network and a segment in the egress IP network. The protection
   services at the IP layer for an end-to-end LSP must be mapped onto
   suitable protection services offered by the optical network. Suppose
   that 1+1 protection is offered to LSPs at the IP layer, i.e., each
   protected LSP has a pre-established hot stand-by. In case of a
   failure of the primary LSP, traffic can be immediately switched to
   the stand-by. This type of protection can be realized end-to-end as
   follows. With reference to Figure 5, let an LSP originate at
   (ingress) router interface A and terminate at (egress) router
   interface F. Under the first protection option, a primary path for
   the LSP must be established first. Let this path be as shown in the
   figure, traversing router interface B in the ingress network,
   optical ports C (ingress) and D (egress), and router interface E in
   the egress network. Next, 1+1 protection is realized separately in
   each network by establishing a protection path between points A and
   B, C and D and E and F. Furthermore, the segments B-C and D-E must
   themselves be 1+1 protected, using drop-side protection. For the
   segment between C and D, the optical network must offer a 1+1
   service similar to that offered in the IP networks.

      +----------------+    +-----------------+    +---------------+
      |                |    |                 |    |               |
      A Ingress IP Net B----C Optical Network D----E Egress IP Net F
      |                |    |                 |    |               |
      +----------------+    +-----------------+    +---------------+

                  Figure 5: End-to-End Protection Example

7.4.2 Single-Layer Protection

   The protection services in the IP layer do not rely on any
   protection services offered in the optical network. Thus, with
   reference to Figure 5, two SRLG-disjoint LSPs are established
   between A and F. The corresponding segments in the optical network
   are treated as independent lightpaths in the optical network. These
   lightpaths may be unprotected in the optical network.

7.4.3 Differences

   A distinction between these two choices is as follows. Under the
   first choice, the optical network is actively involved in end-to-end
   protection, whereas under the second choice, any protection service
   offered in the optical network is not utilized. Also, under the
   first choice, the protection in the optical network may apply
   collectively to a number of IP LSPs. That is, with reference to
   Figure 5, many LSPs may be aggregated into a single lightpath
   between C and D. The optical network protection may then be applied
   to all of them at once leading to some scalability. Under the second
   choice, each IP LSP must be separately protected. Finally, the first
   choice allows different restoration signaling to be implemented in
   the IP and optical network. These restoration protocols are "patched

                          Expires on 5/24/01             Page 20 of 36


   up" at the service boundaries to realize end-to-end protection. A
   further advantage of this is that restoration is entirely contained
   within the network where the failure occurs, thereby improving the
   restoration latency. For instance, if there is a failure in the
   optical network, optical network protocols restore the segment
   within. Under the second choice, restoration signaling is always
   end-to-end between IP routers, essentially by-passing the optical
   network. A result of this is that restoration latency could be
   higher.  In addition, restoration protocol in the IP layer must run
   transparently over the optical network in the overlay mode.

8. IP-based Optical Control Plane Issues

   Provisioning and restoring lightpaths end-to-end between IP networks
   requires protocol and signaling support within optical sub-networks
   and across the interface NNI. In this regard, a distinction is made
   between control procedures within an optical sub-network (Figure 1)
   and those between sub-networks. The general guideline followed in
   this   framework is to separate these two cases, and allow the
   possibility that different control procedures are followed inside
   different sub-networks, while a common set of procedures are
   followed across sub-networks (over interface NNI). Clearly, it is
   possible to follow the same control procedures inside a sub-network
   as defined for control across sub-networks. But this is left as a
   choice as per this framework, rather than a mandate. In the
   following, signaling and routing within and across sub-networks are

8.1  Addressing

   For interoperability across optical sub-networks using an IP-centric
   control plane, the fundamental issue is that of addressing. What
   entities should be identifiable from a signaling and routing point
   of    view? How should they be addressed? This section presents some
   guidelines on this.

   Identifiable entities in optical networks includes OXCs, optical
   links, optical channels and sub-channels, Shared Risk Link Groups
   (SRLGs), etc. An issue here is how granular the identification
   should    be as far as the establishment of optical trails are
   concerned. The scheme for identification must accommodate the
   specification of the termination points in the optical network with
   adequate granularity when establishing optical trails. For instance,
   an OXC could have many ports, each of which may in turn terminate
   many optical channels, each   of which contain many subchannels etc.
   It is perhaps not reasonable to assume that every sub-channel or
   channel termination, or even OXC ports could be assigned a unique IP
   address. Also, the routing of an optical trail within the network
   does not depend on the precise termination point information, but
   rather only on the terminating OXC.   Thus, finer granularity
   identification of termination points is of relevance only to the
   terminating OXC and not to intermediate OXCs (of course, resource

                          Expires on 5/24/01             Page 21 of 36


   allocation at each intermediate point would depend on the
   granularity of resources requested). This suggests an identification
   scheme whereby OXCs are identified by a unique IP address and a
   "selector" identifies further fine-grain information of relevance at
   an OXC. This, of course, does not preclude the identification of
   these termination points directly with IP addresses(with a null
   selector). The selector can be formatted to have adequate number of
   bits and a structure that expresses port, channel, sub-channel, etc,

   Within the optical network, the establishment of trail segments
   between adjacent OXCs require the identification of specific port,
   channel, sub-channel, etc. With an MPLS-based control plane, a label
   serves this function. The structure of the "optical label" must be
   such that it can encode the required information [7].

   Another entity that must be identified is the SRLG [16]. An
   SRLG is an identifier assigned to a group of optical links that
   share a physical resource. For instance, all optical channels routed
   over the same fiber could belong to the same SRLG. Similarly, all
   fibers routed over a conduit could belong to the same SRLG. The
   notable characteristic of SRLGs is that a given link could belong to
   more than   one SRLG, and two links belonging to a given SRLG may
   individually belong to two other SRLGs. This is illustrated in
   Figure 6. Here, the   links 1,2,3 and 4 may belong to SRLG 1, links
   1,2 and 3 could belong to SRLG 2 and link 4 could belong to SRLG 3.
   Similarly, links 5 and 6 could belong to SRLG 1, and links 7 and 8
   could belong to SRLG 4. (In this example, the same SRLG, i.e., 1,
   contains links from two different adjacencies).

   While the classification of physical resources into SRLGs is a
   manual operation, the assignment of unique identifiers to these
   SRLGs    within an optical network is essential to ensure correct
   SRLG-disjoint path computation for protection. SRLGs could be
   identified with a flat identifier (e.g., 32 bit integer).

   Finally, optical links between adjacent OXCs may be bundled for
   advertisementinto a link state protocol [16]. A bundled interface
   may be numbered or unnumbered. In either case, the component links
   within the bundle must be identifiable. In concert with SRLG
   identification, this information is necessary for correct path
   computation [16].

8.2  Neighbor Discovery

   Routing within the optical network relies on knowledge of network
   topology and resource availability. This information may be gathered
   and used by a centralized system, or by a distributed link state
   routing protocol. In either case, the first step towards network-
   wide link state determination is the discovery of the status of
   local links to all neighbors by each OXC.  Specifically, each OXC
   must determine the up/down status of each optical link, the

                          Expires on 5/24/01             Page 22 of 36


   bandwidth and other parameters of the link, and the identity of the
   remote end of the link (e.g., remote port number). The last piece of
   information is used to specify an appropriate label when signaling
   for lightpath provisioning. The determination of these parameters
   could be based on a combination of manual configuration and an
   automated protocol running between adjacent OXCs. The
   characteristics of such a protocol would depend on the type of OXCs
   that are adjacent (e.g., transparent or opaque). Generically, the
   protocol may be refered to as the "Neighbor Discovery Protocol
   (NDP)" although other functions such as link management and fault
   isolation may be performed as part of the protocol (e.g., LMP [4]).

   NDP would typically require in-band communication on the bearer
   channels to determine local connectivity and link status. In the
   case of opaque OXCs with SONET termination, one instance of NDP
   would run on each OXC port, communicating with the corresponding NDP
   instance at the neighboring OXC. The protocol would utilize the
   SONET overhead bytes to transmit the (configured) local attributes
   periodically to the neighbor. Thus, two neighboring switches can
   automatically determine the identities of each other and the local
   connectivity,and also keep track of the up/down status of local
   links. Neighbor discovery with transparent OXCs is described in [4].

       +--------------+          +------------+         +------------+
       |              +-1:OC48---+            +-5:OC192-+            |
       |              +-2:OC48---+            +-6:OC192-+            |
       |    OXC1      +-3:OC48---+     OXC2   +-7:OC48--+     OXC3   |
       |              +-4:OC192--+            +-8:OC48--+            |
       |              |          |            |  +------+            |
       +--------------+          +----+-+-----+  | +----+------+-----+
                                      | |        | |          |
                                      | |        | |          |
       +--------------+               | |        | |          |
       |              |          +----+-+-----+  | |   +------+-----+
       |              +----------+            +--+ |   |            |
       |     OXC4     +----------+            +----+   |            |
       |              +----------+    OXC5    +--------+     OXC6   |
       |              |          |            +--------+            |
       +--------------+          |            |        |            |
                                 +------+-----+        +------+-----+

                Figure 6: Mesh Optical Network with SRLGs

8.3  Topology Discovery

   Topology discovery is the procedure by which the topology and
   resource state of all the links in a sub-network are determined.
   Topology discovery may be done using a link state routing protocol
   (e.g., OSPF, ISIS), or it can be through a management protocol (in

                          Expires on 5/24/01             Page 23 of 36


   the case of centralized path computation). The focus in this
   framework is on fully distributed route computation using an IP link
   state protocol.

   In general, most of the link state routing functionality is
   maintained when applied to optical networks. However, the
   representation of optical links, as well as some link parameters,
   are changed in this setting. Specifically,

   o The link state information may consist of link bundles [16].
     Each link bundle is represented as an abstract link in the network
     topology. Different bundling representations are possible. For
     instance, the parameters of the abstract link may include the
     number, bandwidth and the type of optical links contained in the
     underlying link bundle [16]. Also, the SRLGs corresponding to each
     optical link in the bundle may be included as a parameter.

   o The link state information should capture restoration-related
     parameters for optical links. Specifically, with shared protection
     (Section 8.5), the link state updates must have information that
     allows the computation of shared protection paths.

   o A single routing adjacency could be maintained between neighbors
     which may have multiple optical links (or even multiple link
     bundles) between them. This reduces the protocol messaging

   o Since link availability information changes dynamically, a
     flexible policy for triggering link state updates based on
     availability thresholds may be implemented. For instance, changes
     in availability of links of a given bandwidth (e.g., OC-48) may
     trigger updates only after the availability figure changes by a
     certain percentage.

   These concepts are relatively well-understood. On the other hand,
   the resource representation models and the topology discovery
   process for hierarchical routing (e.g., OSPF with multiple areas)
   are areas that need further work.

8.4  Restoration Models

   Automatic restoration of lightpaths is a service offered by optical
   networks. There could be local and end-to-end mechanisms for
   restoration of lightpaths within a sub-network. Local mechanisms are
   used to select an alternate link between two adjacent OXCs when a
   failure affects the primary link over which the (protected)
   lightpath is being routed. Local restoration does not affect the
   end-to-end route of the lightpath. When local restoration is not
   possible (e.g., no alternate link is available between the adjacent
   OXCs in question), end-to-end restoration may be performed. With
   this, the affected lightpath may be rerouted over an alternate path
   that completely avoids the OXCs or the link segment where the

                          Expires on 5/24/01             Page 24 of 36


   failure occurred. For end-to-end restoration, alternate paths are
   typically pre-computed. Such back-up paths may have to be physically
   diverse from the corresponding primary paths.

   End-to-end restoration may be based on two types of protection
   schemes; "1 + 1" protection or shared protection. Under 1 + 1
   protection, a back-up path is established for the protected primary
   path along a physically diverse route. Both paths are active and the
   failure along the primary path results in an immediate switch-over
   to the back-up path. Under shared protection, back-up paths
   corresponding to physically diverse primary paths may share the same
   network resources. When a failure affects a primary path, it is
   assumed that the same failure will not affect the other primary
   paths whose back-ups share resources.

8.5  Route Computation

   The computation of a primary route for a lightpath within an optical
   sub-network is essentially a constraint-based routing problem. The
   constraint is typically the bandwidth required for the lightpath,
   perhaps along with administrative and policy constraints. The
   objective of path computation could be to minimize the total
   capacity required for routing lightpaths [17].

   Route computation with constraints may be accomplished using a
   number of algorithms [18]. When 1+1 protection is used, a back-up
   path that does not traverse on any link which is part of the same
   SRLG as links in the primary path must be computed. Thus, it is
   essential that the SRLGs in the primary path be known during
   alternate path computation,    along with the availability of
   resources in links that belong to other SRLGs. This requirement has
   certain implications on optical link bundling. Specifically, a
   bundled LSA must include adequate information such that a remote OXC
   can determine the resource availability under each SRLG that the
   bundled link refers to, and the    relationship between links
   belonging to different SRLGs in the bundle.   For example,
   considering Figure 3, if links 1,2,3 and 4 are bundled
   together in an LSA, the bundled LSA must indicate that there are
   three   SRLGs which are part of the bundle (i.e., 1, 2 and 3), and
   that links    in SRLGs 2 and 3 are also part of SRLG 1.

   It is somewhat complex to encode the SRLG relationships in a link
   bundle LSA. This information, however, is naturally captured if the
   link bundle is encoded as a set of "link groups", each specifying
   the links that belong to exactly the same set of SRLGs. Within the
   link group, it is possible to specify the number of links of a
   particular type, for example, OC-48. With reference to Figure 3,
   for example, a bundle LSA can be advertised for the entire set of
   links between OXC1 and OXC2, with the following information:

                          Expires on 5/24/01             Page 25 of 36


   Link Group ID     SRLGs    Link Type   Number   Other Info
   -------------     -----    ---------   ------   ----------
       1             1,2       OC-48       3          ---
       2             1,3       OC-192      1          ---

   Assuming that the above information is available for each bundle at
   every node, there are several approaches possible for path
   computation.  For instance,

   1. The primary path can be computed first, and the (exclusive
      or shared) back-up is computed next based on the SRLGs chosen
      for the primary path.  In this regard,

      o The primary path itself can be computed by taking into account
        specific link groups in a bundle. That is, the primary path
        computation procedure can output a series of link groups the
        path  is routed over. Since a link group is uniquely identified
        with a set of SRLGs, the alternate path can be computed right
        away based on this knowledge. In this case, if the primary path
        set up does not succeed for lack of resources in a chosen link
        group, the primary and backup paths muse be recomputed.

      o It might be desirable to compute primary paths using bundle-
        level information (i.e., resource availability in all link
        groups in a bundle) rather than specific link group level
        information. In this case, the primary path computation
        procedure would output  a series of bundles the path traverses.
        Each OXC in the path would have the freedom to choose the
        particular link group to route that segment of the primary
        path. This procedure would increase the chances of successfully
        setting up the primary path when link state information is not
        up to date everywhere. But the specific link group chosen, and
        hence the SRLGs in the primary path, must be captured during
        primary path set-up, for example, using the RSVP-TE Route
        Record Object [19].  This SRLG information is then used for
        computing the back-up path. The back-up path may also be
        established specifying only which SRLGs to AVOID in a given
        segment, rather than which link groups to use. This would
        maximize the chances of establishing the back-up.

    2. The primary path and the back-up path are comptuted together in
       one step, for example, using Suurbaale's algorithm [20]. In this
       case, the paths must be computed using specific link group

    To summarize, it is essential to capture sufficient information in
    link bundle LSAs to accommodate different path computation
    procedures    and to maximize the chances of successful path
    establishment. Depending on the path computation procedure used,
    the type of support needed during path establishment (e.g., the
    recording of link group or SRLG information during path
    establishment) may differ.

                          Expires on 5/24/01             Page 26 of 36


   When shared protection is used, the route computation algorithm must
   take into account the possibility of sharing links among multiple
   back-up paths. Under shared protection, the back-up paths
   corresponding to SRLG-disjoint primary paths can be assigned the
   same    links. The assumption here is that since the primary paths
   are not routed over links that have the same SRLG, a given failure
   will affect   only one of them. Furthermore, it is assumed that
   multiple failure events affecting links belonging to more than one
   SRLG will not occur    concurrently. Unlike the case of 1+1
   protection, the back-up paths are not established apriori. Rather, a
   failure event triggers the establishment of a single back-up path
   corresponding to the affected primary path.

   The distributed implementation of route computation for shared back-
   up paths require knowledge about the routing of all primary and
   back-up paths at every node. This raises scalability concerns. For
   this reason, it may be practical to consider the centralization of
   the route computation algorithm in a route server that has complete
   knowledge of the link state and path routes. Heuristics for fully
   distributed route computation without complete knowledge of path
   routes are to be determined. Path computation for restoration is
   further described in [17, 21].

8.6  Signaling Issues

   Signaling within an optical sub-network for lightpath provisioning
   is a relatively simple operation. After a route is determined for a
   lightpath, each OXC in the path must establish appropriate cross-
   connects in a coordinated fashion. This coordination is akin to
   selecting incoming and outgoing labels in a label-switched
   environment. Thus, protocols like RSVP-TE [14] and CR-LDP [13] can
   be used for this. A few new concerns, however, must be addressed.

8.6.1 Bi-Directional Lightpath Establishment

   Lightpaths are typically bi-directional. That is, the output port
   selected at an OXC for the forward direction is also the input port
   for the reverse direction of the path. Since signaling for optical
   paths may be autonomously initiated by different nodes, it is
   possible   that two path set-up attempts are in progress at the same
   time. Specifically, while setting up an optical path, an OXC A may
   select output port i which is connected to input port j of the
   "next" OXC B.    Concurrently, OXC B may select output port j for
   setting up a different optical path, where the "next" OXC is A. This
   results in a "collision". Similarly, when WDM functionality is built
   into OXCs, a collision occurs when adjacent OXCs choose directly
   connected output ports and the same wavelength for two different
   optical paths. There are two ways to deal with such collisions.
   First, collisions may be detected and the involved paths may be torn
   down and re-established. Or, collisions may be avoided altogether.

                          Expires on 5/24/01             Page 27 of 36


8.6.2  Failure Recovery

   The impact of transient partial failures must be minimized in an
   optical network. Specifically, optical paths that are not directly
   affected by a failure must not be torn down due to the failure. For
   example, the control processor in an OXC may fail, affecting
   signaling   and other internodal control communication. Similarly,
   the control channel between OXCs may be affected temporarily by a
   failure. These failure may not affect already established optical
   paths passing through the OXC fabric. The detection of such failures
   by adjacent nodes, for example, through a keepalive mechanism
   between signaling peers, must not result in these optical paths
   being torn down.

   It is likely that when the above failures occur, a backup processor
   or a backup control channel will be activated. The signaling
   protocol must be designed such that it is resilient to transient
   failures. During failure recovery, it is desirable to recover local
   state at the concerned OXC with least disruption to existing optical

8.6.3 Restoration

   Signaling for restoration has two distict phases. There is a
   reservation phase in which capacity for the protection path is
   established. Then, there is an activation phase in which the
   back-up path is actually put in service. The former phase typically
   is not subject to strict time constraints, while the latter is.

   Signaling to establish a "1+1" back-up path is relatively straight-
   forward. This signaling is very similar to signaling used for
   establishing the primary path. Signaling to establish a shared back-
   up   path is a little bit different. Here, each OXC must understand
   which back-up paths can share resources. The signaling message must
   itself indicate shared reservation. The sharing rule is as described
   in Section 8.4: back-up paths corresponding to physically diverse
   primary   paths may share the same network resources. It is
   therefore necessary    for the signaling message to carry adequate
   information that allows an   OXC to verify that back-up paths that
   share a certain resources are allowed to do so.

   Under both 1+1 and shared protection, the activation phase has two
   parts: propagation of failure information to the source OXC from the
   point of failure, and activation of the back-up path. The signaling
   for these two phases must be very fast in order to realize response
   times in the order of tens of milliseconds. When optical links are
   SONET-based, in-band signals may be used, resulting in quick
   response.   With out-of-band control, it is necessary to consider
   fast signaling over the control channel using very short IP packets
   and prioritized processing. While it is possible to use RSVP or CR-
   LDP for activating protection paths, these protocols do not provide
   any means to give priority to restoration signaling as opposed to
   signaling for provisioning. For instance, it is possible for a

                          Expires on 5/24/01             Page 28 of 36


   restoration-related RSVP message to be queued behind a number of
   provisioning messages thereby delaying restoration. It is therefore
   necessary to develop a definition of QoS for restoration signaling
   and incorporate mechanisms   in existing signaling protocols to
   achieve this. Or, a new signaling protocol may be developed
   exclusively for activating protection paths during restoration.

8.7   Optical Internetworking

   Ideally, a set of interconnected optical sub-networks must be
   functionally similar to a single optical sub-network. Thus, it must
   be possible to dynamically provision and restore lightpaths across
   optical sub-networks. Therefore:

   o A standard scheme for uniquely identifying lightpath end-points in
     different sub-networks is required.

   o A protocol is required for determining reachability of end-points
     across sub-networks.

   o A standard signaling protocol is required for provisioning
     lightpaths across sub-networks.

   o A standard procedure is required for the restoration of lightpaths
     across sub-networks.

   o It should be possible to apply proprietary provisioning and
     restoration procedures for the segment of a lightpath passing
     through a given sub-net.

   The IP-centric control architecture for optical sub-networks can be
   extended to satisfy the functional requirements of optical
   internetworking. Routing and signaling interaction between optical
   sub-networks can be standardized across the interface NNI (Figure
   1). For the joint control and management of the network, an
   integration of the sub-network management systems is required. The
   functionality provided across NNI is as follows.

8.7.1 Neighbor Discovery

   NDP, as described in Section 8.2, can be used for this. Indeed, a
   single protocol should be standardized for neighbor discovery within
   and across sub-networks.

8.7.2 Addressing and Routing Model

   The addressing mechanisms described in Section 8.1 can be used to
   identify OXCs, ports, channels and sub-channels in each sub-network.
   It is essential that the OXC IP addresses are unique network-wide.

   Provisioning an end-to-end lightpath across multiple sub-networks
   involves the establishment of path segments in each sub-network

                          Expires on 5/24/01             Page 29 of 36


   sequentially. Thus, a path segment is established from the source
   OXC to a border OXC in the source sub-network. From this border OXC,
   signaling across NNI is used to establish a path segment to a border
   OXC in the next sub-network. Provisioning then continues in the next
   sub-network and so on until the destination OXC is reached.

   A version of BGP may be used to determine the routes to destinations
   across sub-networks. Using exterior BGP, adjacent border OXCs in
   different sub-networks can exchange reachability of OXCs and other
   external IP endpoints (border routers). Using interior BGP, the same
   information is propagated from one border OXC to others in the same
   sub-network. Thus, every border OXC eventually learns of all IP
   addresses reachable across different neighboring sub-networks. These
   addresses may be propagated to other OXCs within the sub-network
   thereby allowing them to select appropriate border OXCs as exit
   points for external destinations. To support VPNs, the external
   reachability information should include VPN identifiers.

8.7.3 Restoration

   It is likely that proprietary restoration schemes may be implemented
   within optical sub-networks. It is therefore necessary to consider a
   two-level restoration mechanism. Path failures within an optical
   sub-network should be handled using procedures specific to the
   sub-network. If this fails, end-to-end restoration across sub-
   networks should be invoked. The border OXC that is the ingress to a
   sub-network can act as the source for restoration procedures within
   a sub-network. The signaling for invoking end-to-end restoration
   across NNI is similar to the signaling described in Section 8.6.3.
   The computation of the back-up path for end-to-end restoration may
   be based on various criteria. It is assumed that the back-up path is
   computed by the source OXC, and signaled using standard methods.

9. Other Issues

9.1   WDM and TDM in the Same Network

   A practical assumption would be that if SONET (or some other TDM
   mechanism that is capable partitioning the bandwidth of a
   wavelength) is used, then TDM is leveraged as an additional method
   to differentiate between "flows."  In such cases, wavelengths and
   time intervals (sub-channels) within a wavelength become analogous
   to labels (as noted in [1]) which can be used to make switching
   decisions. This would be somewhat akin to using VPI (e.g.,
   wavelength) and VCI (e.g., TDM sub-channel) in ATM networks. More
   generally, this will be akin to label stacking and to LSP nesting
   within the context of Multi-Protocol Lambda Switching [1]. GMPLS
   signaling [7] supports this type of multiplexing.

9.2   Wavelength Conversion

   Some form of wavelength conversion may exist at some switching
   elements. This however may not be case in some pure optical

                          Expires on 5/24/01             Page 30 of 36


   switching elements.  A switching element is essentially anything
   more sophisticated than a simple repeater, that is capable of
   switching and converting a wavelength Lambda(k) from an input port
   to a wavelength  Lambda(l) on an output port.  In this display, it
   is not necessarily the case that Lambda(k) = Lambda(l), nor is it
   necessarily the case that the data carried on Lambda(k) is switched
   through the device without being examined or modified.

   It is not necessary to have a wavelength converter at every
   switching element.  A number of studies have attempted to address
   the issue of the value of wavelength conversion in an optical
   network. Such studies typically use the blocking probability (the
   probability that a lightpath cannot be established because the
   requisite wavelengths are not available) as a metric to adjudicate
   the effectiveness of wavelength conversion.  The IP over optical
   architecture must take into account hybrid networks with some OXCs
   capable of wavelength conversion and others incapable of this. The
   GMPLS "label set" mechanism [7] supports the selection of the same
   label (i.e., wavelength) across an optical sub-network.

9.3   Service Provider Peering Points

   There are proposed inter-network interconnect models which allow
   certain types of peering relationships to occur at the optical
   layer. This is consistent with the need to support optical layer
   services independent of higher layers payloads. In the context of IP
   over optical networks, peering relationships between different trust
   domains will eventually have to occur at the IP layer, on IP routing
   elements, even though non-IP paths may exist between the peering

9.4   Rate of Lightpath Set-Up

   Dynamic establishment of optical channel trails and lightpaths is
   quite desirable in IP over optical networks, especially when such
   instantiations are driven by a stable traffic engineering control
   system, or in response to authenticated and authorized requests from

   However, there are many proposals suggesting the use of dynamic,
   data-driven shortcut-lightpath setups in IP over optical networks.
   The arguments put forth in such proposals are quite reminiscent of
   similar discussions regarding ATM deployment in the core of IP
   networks.  Deployment of highly dynamic data driven shortcuts within
   core networks has not been widely adopted by carriers and ISPs for a
   number   of reasons: possible CPU overhead in core network elements,
   complexity   of proposed solutions, stability concerns, and lack of
   true economic drivers for this type of service.  This draft assumes
   that this paradigm will not change and that highly dynamic, data-
   driven shortcut lightpath setups are for future investigation.
   Instead, the optical channel trails and lightpaths that are expected
   to be widely used at the initial phases in the evolution of IP over
   optical networks will include the following:

                          Expires on 5/24/01             Page 31 of 36


   o Dynamic connections for control plane traffic and default path
     routed data traffic,

   o Establishment and re-arrangement of arbitrary virtual topologies
     over rings and other physical layer topologies.

   o Use of stable traffic engineering control systems to engineer
     lightpath connections to enhance network performance, either for
     explicit demand based QoS reasons or for load balancing).

   Other issues surrounding dynamic connection setup within the core
   center around  resource usage at the edge of the optical domain.
   One potential issue pertains to the number of flows that can be
   processed by an ingress or egress network element either because of
   aggregate bandwidth limitations or because of a limitation on the
   number of flows (e.g., lightpaths) that can be processed

   Another possible short term reason for dynamic shortcut lightpath
   setup would be to quickly pre-provisioned paths based on some
   criteria (TOD, CEO wants a high BW reliable connection, etc.).  In
   this scenario, a set of paths is pre-provisioned, but not actually
   instantiated until the customer initiates an authenticated and
   authorized setup requests, which is consistent with existing
   agreements between the provider and the customer.   In a sense, the
   provider may have already agreed to supply this service, but will
   only instantiate it by setting up a lightpath when the customer
   submits an explicit request.

9.5   Distributed vs. Centralized Provisioning

   This draft has mainly dealt with a distributed model for lightpath
   provisioning, in which all nodes maintain a synchronized topology
   database, and advertise topology state information to maintain and
   refresh the database. A constraint-based routing entity in each node
   then uses the information in the topology database and other
   relevant details to compute appropriate paths through the optical
   domain. Once a path is computed, a signaling protocol (e.g., [14])
   is used to instantiate the lightpath.

   Another provisioning model is to have a centralized server which has
   complete knowledge of the physical topology, the available
   wavelengths, and where applicable, relevant time domain information.
   A corresponding client will reside on each network element that can
   source or sink a lightpath.  The source client would query the
   server in order to set up a lightpath from the source to the
   destination.  The server would then check to see if such a lightpath
   can be established based on prevailing conditions. Furthermore,
   depending on the specifics of the model, the server may either setup
   the lightpath on behalf of the client or provide the necessary

                          Expires on 5/24/01             Page 32 of 36


   information to the client or to some other entity to allow the
   lightpath to be instantiated.

   Centralization aids in implementing complex capacity optimization
   schemes, and may be the near-term provisioning solution in optical
   networks with interconnected multi-vendor optical sub-networks. In
   the long term, however, the distributed solution with centralization
   of some control procedures (e.g., traffic engineering) is likely to
   be the approach followed.

10.  Evolution Path for IP over Optical Architecture

   The architectural models described in Section 7 imply a certain
   degree of implementation complexity. Specifically, the overlay
   model was described as the least complex for near term deployment
   and the peer model the most complex. Nevertheless, each model has
   certain advantages and this raises the question as to the evolution
   path for IP over optical network architectures.

   The evolution approach recommended in this framework is the
   definition of capability sets that start with simpler functionality
   in the beginning and include more complex functionality later. In
   this regard, it is realistic to expect that initial IP over optical
   deployments will be based on the domain services model (with overlay
   interconnection), with no routing exchange between the IP and
   optical domains. Under this model, direct signaling between IP
   routers and optical networks is likely to be triggered by offline
   traffic engineering decisions. The next step in the evolution of IP-
   optical interaction is the introduction of reachability information
   exchange between the two domains. This would potentially allow
   lightpaths to be established as part of end-to-end LSP set-up. The
   final phase is the support for the full peer model with more
   sophisticated routing interaction between IP and optical domains.

   Using a common signaling framework (based on GMPLS) from the
   beginning facilitates this type of evolution. For the domain
   services model, implementation agreement based on GMPLS UNI
   signaling is being developed in the Optical Interworking Forum (OIF)
   [8, 11, 12]. This agreement is aimed at near term deployment and
   this could be the precursor to a future peer model architecture. In
   this evolution, the signaling capability and semantics at the IP-
   optical boundary would become more sophisticated, but the basic
   structure of signaling would remain. This would allow incremental
   developments as the interconnection model becomes more
   sophisticated, rather than complete re-development of signaling

11. Security Considerations


                          Expires on 5/24/01             Page 33 of 36


12. Summary and Conclusions

   The objective of this draft was to define a framework for IP over
   optical networks, considering the service models, routing and
   signaling issues. There are a diversity of choices, as described
   in this draft, for IP-optical interconnection, service models
   and protocol mechanisms. The approach advocated in this draft
   was to allow different service models and proprietary enhancements
   in optical networks, and define complementary signaling and
   routing mechanisms that would support these. An evolution scenario,
   based on a common GMPLS-based signaling framework with increasing
   interworking functionality was suggested. Under this scenario, the
   IP-optical interaction is first based on the domain services model
   with overlay interconnection that eventually evolves to support full
   peer interaction.

13. References

   1. D. Awduche, Y. Rekhter, J. Drake, R. Coltun, "Multi-Protocol
      Lambda Switching: Combining MPLS Traffic Engineering Control
      With Optical Crossconnects," draft-awduche-mpls-te-optical-
      02.txt, Work in Progress, July, 2000.

   2. K. Arvind, et. al, "Optical Domain Services Interconnect (ODSI)
      Signaling Control Specification, Version 1.4.5" www.odsi-
      coalition.com, March, 2000.

   3. D. Pendarakis, B. Rajagopalan and D. Saha, "Routing Information
      Exchange in Optical Networks," draft-prs-optical-routing-01.ps,
      Internet Draft, Work in Progress, November, 2000.

   4. J. P. Lang, et. al., "Link Management Protocol," draft-ietf-mpls-
      lmp-01.txt, Internet Draft, Work in progress, November, 2000.

   5. K. Kompella et al, "OSPF Extensions in Support of Generalized
      MPLS," draft-kompella-ospf-gmpls-extensions-00.txt, Work in
      Progress, November, 2000.

   6. K. Kompella and Y. Rekhter, "LSP Hierarchy with MPLS TE," draft-
      ietf-mpls-lsp-hierarchy-01.txt, Work in Progress, November, 2000.

   7. P. Ashwood-Smith et. al, "Generalized MPLS - Signaling Functional
      Description", draft-ietf-mpls-generalized-signaling-00.txt,
      Internet Draft, Work in Progress, November, 2000.

   8. O. Abul-Magd, et. al., "Signaling Requirements at the Optical
      UNI," draft-bala-mpls-optical-uni-signaling-01.txt, Internet
      Draft, Work in Progress, November, 2000.

   9. Y. Rekhter and T. Li, "A Border Gateway Protocol 4 (BGP4)",RFC
      1771, March, 1995.

                          Expires on 5/24/01             Page 34 of 36


   10.  E. Rosen and Y. Rekhter, "BGP/MPLS VPNs", RFC 2547, March,

   11.  E. Gray, et. al., "RSVP Extensions in Support of OIF Optical
      UNI Signaling," draft-gray-mpls-rsvp-oif-uni-ext-00.txt, Internet
      Draft, Work in Progress, October, 2000.

   12.  O. Abul-Magd, et. al., "LDP Extensions for Optical UNI
      Signaling," draft-ietf-mpls-ldp-optical-uni-00.txt, Internet
      Draft, Work in Progress, October, 2000.

   13.  P. Ashwood-Smith, et. al., "Generalized MPLS - CR-LDP Signaling
      Functional Description," draft-ietf-mpls-generalized-cr-ldp-
      00.txt, Internet Draft, Work in Progress, November, 2000.

   14.  P. Ashwood-Smith, et. al., "Generalized MPLS - RSVP-TE
      Signaling Functional Description," draft-ietf-mpls-generalized-
      rsvp-te-00.txt, Internet Draft, Work in Progress, November, 2000.

   15.  B. Mack-Crane, et. al., "Enhancements to GMPLS Signaling for
      Optical Technologies," draft-mack-crane-gmpls-signaling-
      enchancements-00.txt, Internet Draft, Work in Progress, November,

   16.  B. Rajagopalan and D. Saha, "Link Bundling in Optical
      Networks," draft-rs-optical-bundling-01.txt, Internet Draft, Work
      in Progress, October, 2000.

   17.  B. Doshi, S. Dravida, P. Harshavardhana, et. al, "Optical
      Network Design and Restoration," Bell Labs Technical Journal,
      Jan-March, 1999.

   18.  E. Crawley, R. Nair, B. Rajagopalan and H. Sandick, "A
      Framework for QoS-based Routing in the Internet," RFC 2386,
      August, 1998.

   19.  D. Awduche, L. Berger, Der-Hwa Gan, T. Li, G. Swallow, V.
      Srinivasan, "RSVP-TE: Extensions to RSVP for LSP Tunnels,"draft-
      ietf-mpls-rsvp-lsp-tunnel-07.txt, Internet Draft, Work in
      progress, October, 2000.

   20.  J. Suurballe, "Disjoint Paths in a Network," Networks, vol. 4,

   21.  S. Ramamurthy, Z. Bogdanowicz, S. Samieian, et al., "Capacity
      Performance of Dynamic Provisioning in Optical Networks", to
      appear in J. of Lightwave Technology.

14. Acknowledgments

   We would like to thank Zouheir Mansourati and Ian Duncan of Nortel
   Networks for their comments on this draft.

                          Expires on 5/24/01             Page 35 of 36


15. Author's Addresses

      Bala Rajagopalan            James V. Luciani
      Debanjan Saha               TollBridge Technologies
      Tellium, Inc.               P,O. Box 1010
      2 Crescent Place            Concord, MA 01742
      P.O. Box 901                Email: james_luciani@mindspring.com
      Oceanport, NJ 07757-0901
      Email: {braja, dsaha}@tellium.com

      Daniel O. Awduche          Brad Cain, Bilel Jamoussi
      UUNET (MCI Worldcom)       Nortel Networks
      Loudoun County Parkway     600 Tech Park
      Ashburn, VA 20247          Billerica, MA 01821
      Phone: 703-886-5277        Phone: 978-288-4734
      Email: awduche@uu.net      Email: bcain@nortelnetworks.com

        ******** This draft expires on May, 24, 2001 ***********

                          Expires on 5/24/01             Page 36 of 36