IPO Working Group D. Papadimitriou Document: draft-papadimitriou-optical-rings-02.txt Alcatel Category: Internet Draft Expires: May 2002 November 2001 Optical Rings and Hybrid Mesh-Ring Optical Networks draft-papadimitriou-optical-rings-02.txt 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 progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. Conventions used in this document: The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC-2119 [1]. Abstract The scope of this draft is to specify the IP-based signaled protection mechanisms for optical rings and hybrid optical mesh-ring topologies. With the dynamic ring configuration process, we determine the information to be distributed to dynamically emulate optical rings on top of a meshed optical network topology. This information exchange further enables the specification of Optical Ring Traffic Engineering. Within the IP Control-based signaling plane, we also specify the mechanisms to provide dynamic and fast-protection Papadimitriou D. - Internet Draft - Expires May 2002 1 draft-papadimitriou-optical-rings-02.txt November 2001 switching, and distributed LSP route computation on top of ring-based and hybrid mesh-rings optical networks. 1. Introduction The concept of optical ring emulation applies when considering the ring configuration on top of optical mesh topology including Optical Cross-Connects (OXC) and not only Optical Add-Drop-Multiplexers (O- ADM). Compared to the approach proposed in [IPO-RFR] and developed in [IPO-OMR], emulated optical rings add configuration flexibility when considering optical network protection. The mechanisms developed here are quite general and enable the deployment of hybrid mesh-rings optical network topologies. It also provides a clear separation between the dynamic ring configuration and the dynamic ring resource allocation. This allows a greater ring resource management flexibility that fits more suitably the bandwidth on demand requirements for optical networks. The complexity of ring emulation arises when defining several rings on top of a mesh topology including Optical Cross-Connects (OXC) connected through fiber trunks. This because by configuring several optical rings on top of a meshed topology, a ring cover is defined. Consequently, the shared protection of some parts of ring - specifically the links and the nodes - with other contiguous rings implies the definition of specific as described in [OPT-NET] and [OPT-RINGS]. The corresponding fast link protection mechanism is performed at the node level is completely distributed and autonomous. Note that it still requires signalling message exchange in order to recover unidirectional link failures. The approach developed here combines distributed fast link protection mechanism and signalling message exchange between nodes when a failure is detected on a link connecting to adjacent nodes belonging to the same ring. In order to specify these mechanisms, we first describe the optical ring concept currently under development at the ITU-T to determine the corresponding mechanism for all-optical rings. 2. Optical Rings - Concepts 2.1 ITU-T Optical Rings The aim of this section is neither to describe the pro and cons of optical rings with respect to optical mesh network nor to explain in detail the well known ITU-T ring technology [ITUT-G841]. Many references can be found comparing, the architecture, the protection as well as the resource use of both kinds of topologies. The objective is to determine the requirements and the functional aspects of these ring technologies in order to define the corresponding mechanisms for all-optical rings. For that purpose, we first describe the ring technology and types of protected rings currently defined (by using ITU-T terminology): Papadimitriou D. - Internet Draft - Expires May 2002 2 draft-papadimitriou-optical-rings-02.txt November 2001 - Optical Channel Dedicated Protection Ring (OCh-DPRing) or O-UPSR (Optical Unidirectional Path Switched Ring). Dedicated protection can also be provided through Single Optical Channel (1+1) Sub- network Connection Protection (OCh-SNCP). - Optical Channel Shared Protection Ring (OCh-SPRing) or O-BPSR (Optical Bi-directional Path Switched Ring) existing in two flavors: 2 Fiber O-BPSR (O-BPSR/2) and 4-Fiber O-BPSR (O-BPSR/4) and 2 switching strategies: ring switch (2- and 4-Fiber) and span switch (4-Fiber only). - Optical Multiplex-Section Dedicated Protection Ring (OMS-DPRing) or O-ULSR (Optical Unidirectional Line Switched Ring) - Optical Multiplex-Section Shared Protection Ring (OMS-SPRing) or O- BLSR (Optical Bi-directional Line Switched Ring) existing in 2 flavors: 2 Fiber O-BLSR (O-BLSR/2) and 4-Fiber O-BLSR (O-BLSR/4) and 2 switching strategies: ring switch and span switch. Details about the corresponding SDH ring architectures can be found in ITU-T G.841 (for SDH/Sonet networks). The ITU-T is currently defining Optical Transport Network (OTN) Ring Architecture by extending the G.841 specification. These ring types can be treated separately if we differentiate the unidirectional (OMS and OCh) Dedicated Protection Rings (DPRing) from the bi-directional (OMS and OCh) Shared Protection Rings (SPRing). This by taking into account that switched rings are also referred to as self-healing rings (SHR) since they protect automatically against failures. The main feature of self-healing rings is their ability to recover any LSP in case of link (i.e. fiber) failure and part of the LSP in case of node failure. SHR applied on optical rings are referred to as WDM-SHR. The corresponding protection mechanism are briefly described here: 1. OMS-DPRing (or O-ULSR): - Unidirectional OMS protection is achieved by using 1+1 fast protection switching at the OMS-layer. - One fiber is dedicated as working fiber and the other is dedicated as protection fiber. Working and protection fibers operate in opposite directions: the working ring operates on the clockwise direction on the protection ring on the counter clockwise direction. The protection ring carries the same signal than the working fiber but in the opposite direction. The same mechanism applies if we replace the term fiber by wavelength. - Since single-ended protection (or unidirectional protection switching) is used, OMS-DPRing does not require coordinated O-APS signalling at the OMS-layer. Papadimitriou D. - Internet Draft - Expires May 2002 3 draft-papadimitriou-optical-rings-02.txt November 2001 2. OMS-SPRing (or O-BLSR): - OMS switched rings are based on optical automatic protection switching (O-APS) protocol through signalling channel by using OMS overhead bytes (or equivalent). The O-APS mechanism enables coordinated actions at the OMS-layer between nodes when a failure occurs. - OMS-SPRing performs bulk OCh-layer switching based on OMS- level failure indications through OMS-layer O-APS signaling; all optical-channels are protected (i.e. switched) as a group within the OMS (so incapable of protecting optical channels independently of one another based on OCh-layer failure indications. - Ring switching protection (2-fiber and 4-fiber) is based on loopback protection switching and activated when both working and protection link failure occurs simultaneously. By using O- APS signalling, when a link failure occurs, the working lines (wavelength) from the working fiber are looped back onto the opposite direction on protection wavelength on the protection fiber in the opposite direction. Note that combination of working and protection link failure and node failure should also be considered. - Span switching protection (4-fiber only) is activated when only working link failure occurs. When a link failure occurs, the working lines (wavelengths) from the working fiber are switched to the protection fiber in the same direction. Span switching protection is also referred to as non-loopback switched protection. - In terms of SHR, OMS-SPRings are referred to as Shared Line switched WDM-SHR (SLs-WDM-SHR) since they rely on shared line switched protection. 3. OCh-DPRing (or O-UPSR) - Unidirectional path switched-protection is achieved by using 1+1 fast protection switching at the OCh-layer. - One fiber is dedicated as working fiber and the other is dedicated as protection fiber. Working and protection fibers operate in opposite directions: the working ring operates on the clockwise direction on the protection ring on the counter clockwise direction. The protection ring carries the same signal than the working fiber but in the opposite direction so that the protection ring fully protects the working ring. The mechanism applies if we replace the term fiber by wavelength. - Since single-ended protection (or unidirectional protection switching) is used, OCh-DPRing does not require coordinated O- APS OCh-layer signalling. If a working OCh failure occurs, only one direction is affected and for that direction OCh switching is performed at the far-end. In the other direction, no OCh-layer switching is performed. - OCh-layer protection is based on optical channel resilient schemes by protecting individual path; so this protection allows a selective recovery of the OMS between nodes Papadimitriou D. - Internet Draft - Expires May 2002 4 draft-papadimitriou-optical-rings-02.txt November 2001 - In terms of SHR, OCh-DPRings are referred to as Dedicated Path switched WDM-SHR (DPs-WDM-SHR) since they rely on dedicated path switched protection. 4. OCh-SPRing (or O-BPSR) - Bi-directional path switched protection requires the need for a protection architecture performing OCh switching based on independent OCh failure detection. - This implies the specification of an O-APS signalling protocol at the OCh-layer using the OCh overhead byte. - Ring switching protection can also be defined for 2- and 4- Fiber OCh-SPRings. Span switching protection can be defined for 4-Fiber OCh-SPRings. - In terms of SHR, OCh-SPRings are referred to as Shared Path switched WDM-SHR (SPs-WDM-SHR) since they rely on shared path switched protection The distinction between Working and Protected path is defined as follows: - Working path: carrying optical channel defined between the source and the destination node - Protecting path: non-carrying optical channel defined between the source and the destination node transported over the complement of the ring The distinction between Working and Protected line is defined as follows: - Working line: carrying OMS defined between the source and an intermediate node or between intermediate nodes or between an intermediate node and the destination node transported over the complement of the ring - Protection line: non-carrying OMS defined between the source and an intermediate node or between intermediate nodes or between an intermediate node and the destination node transported over the complement of the ring The following table gives the equivalence between the linear (mesh networks) and ring protection. -------------------------------------------------------------------- Linear Protection Ring Protection -------------------------------------------------------------------- Dedicated Line Protection 1+1 OMS-DPRing (O-ULSR) Shared Line Protection 1:1 - 1:N - M:N OMS-SPRing (O-BLSR) Dedicated Path Protection 1+1 OCh-DPRing (O-UPSR) Shared Path Protection 1:1 - 1:N - M:N OCh-SPRing (O-BPSR) -------------------------------------------------------------------- 2.3 Transparent and All-Optical Rings The previous section detailed the ring technologies and types for Optical Transport Networks (i.e. ITU-T G.709 based OTN). The Papadimitriou D. - Internet Draft - Expires May 2002 5 draft-papadimitriou-optical-rings-02.txt November 2001 objective here is to define the same kind of mechanisms for pure optical rings. A pure optical ring is defined as a ring including Optical Cross- Connects (OXC) and Photonic Cross-Connects (PXC). The difference between OXC and PXC is that O/E/O conversion is performed at each of its interface while PXCs do not perform O/E/O conversion at all. An all-optical ring is defined as ring including only PXC while a transparent ring is defined as including at least one OXC. In order to be independent of physical layer concerns like chromatic dispersion, PMD (affecting the edge-to-edge distance), OSNR (affecting the number of optical spans), crosstalk, etc., only transparent optical rings are considered in the remaining parts of this document. We refer to these transparent optical rings as optical rings. The fundamental problem of optical rings is to determine to which extend the IP control-based signaling plane need to be considered for defining ring protection mechanism. In the model developed in this memo, we try to extract the relevant information needed in order to provide the same level of resiliency as the one currently available with SDH/Sonet DXC-based rings [ITUT-G841] and the corresponding extension under definition for Optical Transport Networks (OTN). This implies the definition at IP control-plane level of the required O- APS signalling protocol by keeping independence from the current protection mechanisms defined for SDH/Sonet rings. The information needed to dynamically configure an optical ring on top of a mesh optical network topology and the corresponding link (and wavelengths) resources management information depends on: - the ring identification - the ring protection type - the ring directionality - the ring architecture Based on this information, the dynamic ring configuration is performed (cf. Section 3). When the optical ring has been configured (i.e. emulated on top of the mesh topology), the ring resources allocation to the Lambda-switched LSPs (i.e. the optical channels) segments or tunnels can be computed for this emulated ring. The equivalence between ITU-T rings and the corresponding protection mechanism in all-optical networks is summarized in the following table: --------------------------------------------------------- ITU-T Rings | Corresponding protection mechanism --------------------------------------------------------- OMS-DPRing | 1+1 dedicated fiber link (or wavelength) --------------------------------------------------------- Papadimitriou D. - Internet Draft - Expires May 2002 6 draft-papadimitriou-optical-rings-02.txt November 2001 OMS-SPRing | M:N shared fiber link (or wavelength) --------------------------------------------------------- OCh-DPRing | 1+1 dedicated LSP segment protection --------------------------------------------------------- OCh-SPRing | M:N shared LSP segment protection --------------------------------------------------------- The OMS-layer protection (dedicated or shared) concept corresponds to the protection of a wavelength bundle transported on a fiber link between two adjacent nodes. The OCh-layer protection (dedicated or shared) concept corresponds to the protection of LSP segment(s) transported on a wavelength switched path (i.e. partial LSP path) between the ingress and the egress node of a given ring. Knowing each of the SDH/Sonet and OTN ring functional aspects and the mapping of the underlying concepts with their counterparts in all- optical networks, the corresponding ring protection mechanism can be defined. 2.3 Meshed Optical Network Topology The basic assumptions concerning the optical network topology are to be considered within the scope of [GMPLS-ARCH]: - Mesh (and transparent) optical network including Optical Cross- Connect (OXC) connected through fiber trunks as described in [IPO- SRLG]. - OXC interfaces or ports are Lambda-Switch Capable (LSC) interfaces as described in [GMPLS-SIG] meaning that any OXC interface includes DWDM capabilities. OXC or eventually fiber switch capable (FSC) interfaces (but they are considered as a particular outside the main scope of this document). - Each of these OXCs has its own IP controller that may be or not included within the same device. Each controller is uniquely identified by its IP Address. The inter-connection of the IP controllers defines the IP based control-plane of the optical network. - Signaling channels between the IP-based controllers of OXCs is realized through either dedicated Optical Signalling Channels (out- of-band/in-fiber) or through a dedicated out-of-band/out-of-fiber network. - Signaling protocol is based on Generalized MPLS signaling as described in [GMPLS-SIG] in combination with Generalized RSVP-TE [GMPLS-RSVPTE] and Generalized CR-LDP [GMPLS-CRLDP] extensions. The basic meshed topology of the optical network can be represented as a planar graph: Papadimitriou D. - Internet Draft - Expires May 2002 7 draft-papadimitriou-optical-rings-02.txt November 2001 A ------ B ------ C | | | | | | | | | D ------ E ------ F | | | | | | | | | G ------ H ------ I Each of the nodes (vertex) of this topology is an OXC. Nodes are connected through fiber links. This topology (see above figure as example) also defines a planar graph [OPT-NET], G(V,E) where V is the vertex and E the edge of the planar graph. The considerations developed here take into account this representation of the optical network topology. The non-planar case is left for further study. 3. Optical Ring Emulation Ring emulation concept refers to the configuration of optical rings on top of a meshed optical topology including OXCs and not only O- ADMs. This approach provides the advantage of ring configuration flexibility without having to move or replace the devices already included in the optical network. The emulation concept provides also more flexibility during the (dynamic) configuration of optical ring covers and the resource allocation between the different rings included in the cover. 3.1 Optical Ring Covers A ring cover is defined as a set of closed paths that covers all links in the optical network at least once. So, a (complete) optical meshed topology ring cover of the meshed optical topology will be achieved when every node can be connected to at least one ring. Two types of ring covers can be defined: - non-overlapping ring cover: several rings covering (any node, in case of complete ring-covers) the optical network are emulated on top of a mesh topology; each of these nodes belongs to only one ring except boundary nodes; non-overlapping ring covers can be complete (covering the whole meshed topology) or incomplete (partially covering the meshed topology) - overlapping ring cover (or multi-ring cover): several rings covering (any node in case of complete ring-covers) are configured on top of an optical meshed topology; each of these nodes can be a boundary node However, in the first release of this document, we only consider non- overlapping incomplete or complete ring covers. This choice is motivated by the complexity of the multi-ring covers as described in Papadimitriou D. - Internet Draft - Expires May 2002 8 draft-papadimitriou-optical-rings-02.txt November 2001 [OPT-MWS] and [OPT-RINGS] where double covers of non-planar graphs are studied. In ring covers, fiber links shared between more than one ring are referred to as shared links. However, wavelengths are dynamically assigned to one (and defined as dedicated protection wavelengths) or more than one ring (and defined as shared protection wavelengths). Consequently, the emulation of an optical ring on top of meshed optical network topology requires that: - the configuration of the OXC optical switching matrix included within a ring is designed to provide O-ADM functions (add, drop, drop-and-continue, protection switching, etc.) - in a shared link, each of the wavelengths belongs (for a given time period) to one and only one ring precluding the usage of the dynamically assigned wavelengths by the other ring In the following mesh topology, a non-overlapping ring cover is defined for instance by the following set of rings: - Ring 1: A-B-E-D-A - Ring 2: B-C-F-E-B - Ring 3: E-F-I-H-E - Ring 4: D-E-H-G-D In this topology: - Node C is an internal node since it belongs only to Ring 2 - Node D is a boundary node since it belongs to Ring 1 and Ring4 A ------ B ------ C | | | | Ring 1 | Ring 2 | | | | D ------ E ------ F | | | | Ring 4 | Ring 3 | | | | G ------ H ------ I Node E has a particular situation in this topology since this boundary node belongs to the four rings simultaneously. So, if we consider each ring as an abstract node, the same topology can also represented as follows B,E R1 --------------- R2 | \ / | | \ / | | \ E / | E,D| =====x===== |E,F | / E \ | | / \ | Papadimitriou D. - Internet Draft - Expires May 2002 9 draft-papadimitriou-optical-rings-02.txt November 2001 | / \ | R4 --------------- R3 E,H As a result, if a node N1 connected to the ring R1 requests an LSP to a node N2 connected to ring R3, the shortest ringed path is the one going through Node E. This shows clearly that additional information is required that takes into account the working (and protection) load of the ring. This load information has to be shared between the rings and between the rings and the external nodes to the rings. In the same topology, an overlapping ring cover is defined by the following set of rings: - Ring 1: A-B-C-F-E-D-A - Ring 2: D-E-F-I-H-G-D - Ring 3: E-F-I-H-E A ------ B ------ C | | | | Ring 1 | | | | D ------ E ------ F | | | | Ring 2 | | | Ring 3 | G ------ H ------ I As mentioned previously, overlapping ring covers are not considered in the first release of this document. 3.2 Dynamic Optical Ring Configuration The emulation of a ring on top of optical meshed topology is determined by the exchange of the following information between OXC: - each ring is defined by a Ring ID (32-bit field) - each ring shares a unique Virtual IPv4 Address (32-bit field) - a Loopback IPv4 Address (32-bit field) is allocated per node belonging to a given ring and per ring contiguous to a given node such that neighboring nodes see a given ring as a single abstract node identified by its loopback IPv4 address - each ring is characterized by a Ring Protection Type (8-bit field) which enables to select a given protection-type when establishing a LSP over that ring - each ring knows the complete list of identifiers describing the SRLGs to which a given link belongs. - each ring owns an associated Ring Metric used to bootstrap the initial ring resource allocation during the ring configuration process Papadimitriou D. - Internet Draft - Expires May 2002 10 draft-papadimitriou-optical-rings-02.txt November 2001 - each ring defines a internal Ring Policy (8-bit field) including ring protection strategy, the ring protection scheme, and the priority/preemption support Consequently, the dynamic ring configuration includes the exchange of basic identification information: the Ring Identifier, the Ring Protection-Type, the Ring Virtual IP Address, the Ring Metric and the Ring Policy. 3.2.1 Ring ID Each of the node belonging to a given ring is identified by a node ID defined as a Loopback IPv4 Address (in the future we will include IPv6 node identification). So that a specific common identifier need to exchanged between nodes belonging to the same ring. For this purpose, a Ring ID defined as a 32-bit integer field that uniquely identifies a given ring within a given optical domain (i.e. administrative authority). The ring ID is exchanged between neighboring nodes until reaching a loop (on top of the meshed topology) constituted by nodes having the same Ring ID. The Ring ID verification process at each node is straightforward and does not require additional explanations. 3.2.2 Virtual IP Address Each node included within a given ring shares a unique virtual IPv4 Address with other nodes belonging to the same ring. The IPv4 address is uniquely defined per optical domain (i.e. administrative authority). By using a Virtual IPv4 Address to address a ring, a network management system can reach any node of a given ring included within a complex optical ringed topology without knowing the status of the ring node or the corresponding ring topology. Example: If we refer to the previous optical network topology (see figure), we have the following IP reachability between the multi-ring cover: - Ring 1 is identified by the Ring ID 1 and Virtual IP Address 1 - Ring 2 ... Ring ID 2 and Virtual IP Address 2 - Ring 3 ... Ring ID 3 and Virtual IP Address 3 - Ring 4 ... Ring ID 4 and Virtual IP Address 4 3.2.3 Loopback IPv4 Address In order to avoid the use of interface address and so potentially experience disrupted communication between nodes belonging to the same, a loopback IPv4 address is allocated per node. Papadimitriou D. - Internet Draft - Expires May 2002 11 draft-papadimitriou-optical-rings-02.txt November 2001 For each ring contiguous to a given node, an additional loopback IPv4 address is allocated to this node. A ring can be considered as an abstract single node for its neighboring OXC. Consequently, this IP address identifies the abstract node constituted by a ring with respect to the rings connected to this node. From the neighboring rings point-of-view, this loopback IPv4 address enables to reach a given node at any time independently of the status of its incoming links. The loopback IPv4 address is exchanged with the ring contiguous to the node owner of this address but also between nodes belonging to the same ring. These loopback IPv4 addresses are exchanged between neighboring nodes in order to avoid the use of interface address for intra-ring communication and duplicated loopback IPv4 address that are subsequently distributed to neighboring rings. Example: From a nodal point of view, the neighbor relationship is determined by local loopback IPv4 address to each of the neighboring ring from a given node: - Node B (belonging to ring 1 and ring 2) has 2 neighboring rings: - Ring 1 identified by Loopback IP address B01 - Ring 2 identified by Loopback IP address B02 - Node 1 connected has 1 neighboring ring: - Ring 1 identified by Loopback IP address A1 - Node 3 connected has 2 neighboring rings: - Ring 1 identified by Loopback IP address D1 - Ring 2 identified by Loopback IP address D2 From a ring point of view, the neighbor relationship is determined by the local loopback IPv4 address: - Ring 1 has 3 neighboring rings: - Ring 2 identified by Loopback IP address B21 and E21 - Ring 3 identified by Loopback IP address E31 - Ring 4 identified by Loopback IP address D41 and E41 Consequently, a boundary node belonging to two rings owns two virtual IP addresses. In order to pass from one to ring to the other, the boundary node performs a shortcut. Imagine that Node 1 connected to Ring 1 (through Node A) communicates with the Node 2 connected to Ring 3 (through Node I). Then, the explicit route computed and selected by Node 1 can be [A1 - E3 - Node 2]. However, the actual route followed by the signalling message is [Node 1 - A1 (tunnel1) E3 (tunnel2) I3 - Node 2] where tunnel1 could. Papadimitriou D. - Internet Draft - Expires May 2002 12 draft-papadimitriou-optical-rings-02.txt November 2001 Node 1 ------ A ------ B ------ C | | | | Ring 1 | Ring 2 | | | | Node 3 ------ D ------ E ------ F | | | | Ring 4 | Ring 3 | | | | G ------ H ------ I ------ Node 2 3.2.4 Optical Ring Protection Type The Optical Ring Protection Type defines the protection technology supported by nodes belonging to the same ring. The Optical Ring Protection Type (which can be encoded as an 8-bit field) is defined as the combination of the following parameters: - Ring technology: Optical (= 0) or OTH-based (= 1) - Ring protection type: Dedicated (= 0) or Shared (= 1) - Ring protection level: Link-level (= 0) or Path-level (= 1) - Ring protection fibers: 2-fiber (= 0) or 4-fiber (= 1) - Ring switching: ring-switch (= 0) or span-switch (= 1) - Other values (3 bits) are reserved for future use So for a instance, an OCh-DPRing is defined by a 0xA0 protection type and an 4-fiber shared path-level optical ring is defined by a 0x70 protection type. Since span-switch is only supported with 4-fiber rings, one must avoid the 2-fiber and span-switch combination. Note the ring technology could also be defined as Transparent (= 0) and All-Optical (= 1) when the optical network topology includes only OXC or PXC respectively. 3.2.5 Ring Link Diversity - SRLG The shared link risk groups (SRLGs) are exchanged between nodes belonging to the same ring. This in order to potentially avoid the use of a shared resource that can affect all links belonging to this group in case of failure of this shared resource. For instance, as described in [IPO-SRLG], two LSPs flowing through the same fiber link in the same fiber trunk. [IPO-BUNDLE] extends the SRLGs concept by demonstrating 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. 3.2.6 Ring Metric The ring metric is used to bootstrap the initial ring resource allocation during the ring configuration process. This metric is exchanged during the initial ring configuration and only considered Papadimitriou D. - Internet Draft - Expires May 2002 13 draft-papadimitriou-optical-rings-02.txt November 2001 by a node if its ring ID corresponds to the one advertised with the ring metric. The ring metric is a composed metric including the following components: the ring absolute weight, the ring capacity and the ring maximum restoration time. 1. Ring absolute weight The ring absolute weight translates the number of nodes that belongs to the ring. This value is used during the initial configuration process to determine the number of non-adjacent nodes belonging to the same ring. For instance, it the ring absolute weight metric of given ring is N then by receiving this value (or simply by consulting this value), a node belonging to a given ring knows that the number of non-adjacent nodes belonging to the same ring equals N-3. 2. Ring Capacity During the initial configuration process, the ring capacity defines the number of incoming, outgoing and dropped channels (or wavelengths) provisioned at a given node for a given ring ID. These values are defined for the working as well as the protection capacity. The number of potentially available but unused wavelengths by this ring is also announced during the initial configuration phase. These wavelengths are consequently left free for other rings that can be defined using the same node. The ring capacity (per node) is exchanged between adjacent nodes until each node belonging to the same ring has received (N-1) copy of the ring capacity. The knowledge of these values will enable to determine the initial working ring load, protection ring load and spare capacity (also defined as non-allocated ring load) for the newly configured ring. This in turn provides the required information for each node to compute the number of potential LSPs that can pass through (i.e. bar- state) or be dropped (i.e. cross-state). 3. Maximum Restoration Time Typically, this static value defines the maximum restoration time (MRT) for a given ring type at the initial configuration time. The MRT values are distributed between nodes in order to know the performance of the other nodes belonging to the same ring. With the current state of the usage of all-optical (for instance, MEMS) technology in OXC enabling less than 10 ms switching time, typical MRT values range between 10 to 50 ms for bi-directional Papadimitriou D. - Internet Draft - Expires May 2002 14 draft-papadimitriou-optical-rings-02.txt November 2001 rings. For unidirectional rings using optical splitters (as described in [IPO-MULT], typical MRT values are within 1 ms. Note that these values are statically configured and only during the initial ring configuration period. So, they are not considered during the working period of the ring. 3.2.6 Ring Protection Policy Ring protection policy refers to the different strategies applicable to optical rings. The ring protection policy includes mainly the ring protection strategy, the ring protection scheme and ring protection priority. 1. Strategy The ring protection strategy could be revertive (default for dedicated 1:1 ring protection or shared protection 1:N and M:N ring protection) or non-revertive (default for 1+1 rings). Depending on the ring protection type, the ring protection strategy is determined by the under-laying hardware (or low-layer) opto(- electro-)-mechanical technology providing the automatic protection switching function. However, these considerations are outside of the scope of this document. 2. Scheme The ring policy scheme can be configured as provisioned or non- provisioned. The non-provisioned scheme refers to restoration (or re- routing) so that this scheme is not generally used when considering ring protection in optical networks. The provisioned scheme is one used when the protection path or lines are left unused and fully dedicated to protection purposes; switching to these paths will only occur if a working path or line failure need to be recovered. 3. Priority/Preemption The prioritization capability refers to the support of initial setup and recovery priority associated with the subsequent link working and protection allocated to the LSP established on top of the emulated ring. Note that the priority support extends to the signalling support of prioritized working and protection LSPs created on a given ring. The preemption capability refers to the support of link working and protection resources pre-emptability in case of link or node failure within a given ring. 3.2.7 Summary Papadimitriou D. - Internet Draft - Expires May 2002 15 draft-papadimitriou-optical-rings-02.txt November 2001 Consequently, the exchange of information between nodes belonging to the same ring enables the common knowledge of the identification and properties that uniquely define a given optical ring. - Ring Identifier - Ring Virtual IP Address - Ring Protection Type - Ring SRLG per link - Ring Metric including the absolute ring weight, the ring capacity and the ring MRT - Ring Policy including ring protection strategy, the ring protection scheme, and the priority/preemption support 4. Dynamic Ring Resource Allocation The dynamic ring resource allocation process is closely related to intra-ring traffic engineering (intra-ring TE). The ring resource allocation process is required to dynamically setup Lambda-switched LSP segments on top of emulated rings. 4.1 Link TE Resource Link TE resources are dynamically exchanged between nodes belonging to the same ring to provide the necessary information for intra-ring TE. Dynamic Ring Resource Allocation is based on the knowledge of the following link TE resource-related information: - Link identification (or link bundle identification) - Maximum available bandwidth per link - Minimum and maximum reservable bandwidth per link - Link protection type including unprotected 0:1, dedicated 1:1, shared M:N and dedicated 1+1 protection - Link priority associated to the minimum/maximum reservable bandwidth per link - Link priority associated with the link protection type - List of SRLGs per link (in order to compute disjoint working and protection LSP, for instance) - Resource Class/Color defined per link The dynamic exchange of this information (IGP link-state advertisements) during the working period of the ring enables the real-time computation of the value(s) (i.e. the component values) defined for the ring TE metric. 4.2 Ring TE Metric The ring traffic engineering metric (ring TE metric) can be considered as a dynamic ring metric whose component value(s) are periodically re-evaluated during the working period of the emulated ring. This ring metric is a composed TE metric including the following components: - a ring weight - a ring load Papadimitriou D. - Internet Draft - Expires May 2002 16 draft-papadimitriou-optical-rings-02.txt November 2001 - a ring maximum restoration time 1. Ring weight In the previous example, the Node A does not know the number of nodes the (i.e. hop count) within a given neighboring ring. To make the distinction between rings, we introduce the concept of ring weight. Ring weight is inversely proportional to number of nodes belonging to a ring multiply by 100. The ring weight is defined as: Working ring weight = (1/[Number of nodes] x 100) x r1 where r1 is a weighted integer factor r1 is a user customizable integer whose default value equal 1 For instance, in the above example each ring have a weight of 25 since 1/[Number of nodes = 4] x 100. However, this definition does not allow discriminating between working and protection path. So we define a working ring weight and a protection ring weight (this applies only to bi-directional rings). For bi-directional rings, the ring weight is defined in two flavors: - Working ring weight = (1/[Number of nodes] x 100) x r1 - Protection ring weight = (100 - [Working ring weight]) x r2 where r1 and r2 are weighted integer factors defined by default as r1 = 1 and r2 = 1. 2. Ring Load The ring load is only defined for bi-directional rings. The working ring load quantity defines, for each direction of propagation, the largest number of working LSPs (in any fiber link) flowing in opposite direction of the protection LSPs. The working ring load is defined at each time interval as: Working ring load = [Number of working LSP] However, this definition does not give the relative working load of the ring compared to its total capacity. So, to normalize the working ring load, this value is divided by the total ring capacity (in terms of total number of LSPs). Working ring load = [Number of working LSP]/[Total number of LSPs] Correspondingly, the protection ring load will be defined as the number of protection LSPs divided by the total ring capacity (in terms of total number of LSPs). The protection ring load is defined at each time interval as: Papadimitriou D. - Internet Draft - Expires May 2002 17 draft-papadimitriou-optical-rings-02.txt November 2001 Protection ring load = [Number of protection LSP]/[Total number of LSPs] Consequently, the working ring load, the protection ring load and the non-allocated ring load are defined by the following formulas: Working Ring Load = ([Number of working LSP]/[Ring capacity] x 100) x r3 Protection Ring Load = ([Number of protection LSP]/[Ring capacity] x 100) x r4 Unallocated Ring Load = ((1 - ([Working Ring Load] + [Protection Ring Load])/[Ring Capacity]) x 100) x Mean(r3,r4) Where the r3 and r4 parameters are weighted integer factors, defined by default as r3 = 1 = r4. 3. Maximum Restoration Time Typically, this value defines the maximum restoration time (MRT) for a given ring type. Typical values are: - Bi-directional rings (shared protection): 10 to 50 ms - Unidirectional rings (dedicated protection): 1 ms So, we define the associated metric value as: MRT = (MRT[N+1] x 100) x r5 In this definition, the r5 parameter is a weighted integer factor defined by default as equal to 1. Comparing to the MRT defined in the ring metric, the values considered here are dynamic. The maximum restoration time computation per ring depends on the failure history of that ring. Each time a failure is experienced by a ring the MRT value is adapted through the use of the following computation: If MRT[N+1] > MRT[N] then MaxMRT[N+1] = MRT[N] + (k1 x MRT[N]) and MRT[N+1] = Mean(MaxMRT[N+1]; MRT[N]) If MRT[N+1] < MRT[N] then MinMRT[N] = MRT[N] - (k2 x MRT[N]) and MRT[N+1] = Mean(MRT[N]; MinMRT[N+1]) Where MRT[N+1] is defined as the Maximum Restoration Time when a given ring has already experienced N failures (so that N defines the number of ring restoration process occurrence). The k1 and k2 are defined as weighting factors (0 < k1 < 1 and 0 < k2 < 1) whose values are by default k1 = 1 and k2 = 1. Consequently, the Ring TE Metric is defined as: Papadimitriou D. - Internet Draft - Expires May 2002 18 draft-papadimitriou-optical-rings-02.txt November 2001 Ring TE Metric = (Working Ring Weight x r1 | Protection Weight x r2) + (Working Ring Load x r3 | Protection Load x r4) + (Maximum Restoration Time x r5) The corresponding advertisement will include both the working and the protection part of the metric. 5. Ring Protection in Optical Networks Within optical networks, the purpose of the ring protection mechanisms is to incorporate optical layer survivability in network- layered structures. Even if some of the higher layers have their own protection mechanisms, by incorporating survivability mechanisms at multiple layers leads to the following issues: - the protection function allocation to each of the layers - the coordination of the protection functions applied at each layer during failure recovery process Nevertheless, as described in [OPT-MRPS], providing optical layer survivability through ring protection has been strongly suggested by arguing a combination of optical ring protection with optical linear protection as well as additional restoration mechanism provides the best optical network survivability. 5.1 Escalation Strategy The escalation strategy is defined by the set of detection functions describing the originating failure, the protection functions applied within recovery process to recover the failure and the interaction between the upper and/or lower layers protection functions applied during the recovery process. The escalation strategies are governed either by arbitrarily setting failure detection and recovery completion time (using hold-off timers) or by explicit message exchange between the layers. Escalation strategies (also referred to as inter-working strategy) have been proposed in [IPO-POES] and summarized here: - Bottom-up strategy starts at closest layer to the failure (generally the bottom layer) and escalates toward the upper layer upon expiration of the recovery timer (hold timer). This timer is defined in such a way that is allocates "enough time" the lower layer(s) to detect the failure, execute the recovery process and recovery completion time before triggering recovery process defined at the higher layer(s). - Top-down strategy starts at the upper(most) layer and escalates downward to the lower layer(s); depending on the working level, the top-down strategy is suitable when each of the layers define Class- of-Services (CoS) so that the escalation strategy might take this CoS into account when executing the recovery process. 5.2 Bottom-Up Strategy Papadimitriou D. - Internet Draft - Expires May 2002 19 draft-papadimitriou-optical-rings-02.txt November 2001 We assume here a classical Bottom-up strategy since it has the shortest recovery time (with respect to the Top-down escalation strategy), and define the threshold at which the upper-layer protection needs to be "executed". In optical rings, the following escalation strategy can be defined (here we do not assume FSC capable OXC interfaces, so that GMPLS signalling is not considered at the lowest layer) Packet Layer <==> PSC interface <==> GMPLS Signalling ^ ^ | | | | | | Lambda Layer <==> LSC interface <==> GMPLS Signalling ^ ^ | | | | | | Fiber Layer <==> FSC Interface <==> Layer-1 Mechanism Compared to TDM SDH/Sonet Rings control through GMPLS signalling [GMPLS-SIG] or Optical Transport Networks as defined in [ITUT-G872] and [ITUT-G709]): Packet Layer <==> PSC interface <==> GMPLS Signalling ^ ^ | | | | | | Path (OCh) Layer <==> OCh (G.709) interface <==> GMPLS Signalling ^ ^ | | | | | | MS (OMS) Layer <==> Physical interface <==> Layer-1 Mechanism ^ ^ | | | | | | RS (OTS) Layer <==> Physical interface <==> Layer-1 Mechanism Note: Layer-1 mechanism can be used in conjunction with [LMP] and [LMP-WDM]. Or eventually when considering the G.709 ODUk Switching Layer (also referred to as Digital Path Layer): Papadimitriou D. - Internet Draft - Expires May 2002 20 draft-papadimitriou-optical-rings-02.txt November 2001 Packet Layer <==> PSC interface <==> GMPLS Signalling ^ ^ | | | | | | Packet Layer <==> TDM (G.709) interface <==> GMPLS Signalling ^ ^ | | | | | | Path (OCh) Layer <==> OCh (G.709) interface <==> GMPLS Signalling ^ ^ | | | | | | MS/RS (OMS/OTS) <==> Physical interface <==> Layer-1 Mechanism This approach is based on the common coordination achieved by resorting escalation strategies. At each level, the resiliency scheme is activated sequentially starting from the lower to the upper layer. This implies that at each level the detection of a failure condition (and restoration time) triggers either a protection/restoration mechanism and if the protection/restoration completion time is reached then triggers an upper-layer protection/restoration mechanism. 5.3 Ring Protection Mechanisms In this section the two generic ring protection mechanisms are discussed. The main difference between these mechanisms is that the dedicated protection performs a physical-layer level protection switching while the shared protection needs a dedicated O-APS signalling protocol to synchronize the execution of the protection switching mechanism. 5.3.1 Dedicated Protection Dedicated LSP protection can be easily achieved by splitting the incoming optical signal and send the one copy of the output signal into working wavelength (or fiber) and the other copy into the protection wavelength (or fiber). This results in two copies of the signal propagating the ring in opposite directions. The far-end (or receive-end) switches to the protection wavelength (or fiber) if a degradation or failure occurs. The proposed mechanism can restore a single link failure or a single node failure. Moreover, since the far-end (receiving end) needs to notify the near- end (transmitting end) of the failure, an O-APS signalling mechanism is not required. However, when a link failure is detected on the working ring, the far-end switch to the protection ring (in the Papadimitriou D. - Internet Draft - Expires May 2002 21 draft-papadimitriou-optical-rings-02.txt November 2001 opposite direction) but for any wavelength flowing through this link, so that only a bulk LSP protection switching is achieved by using this technique. By using the optical signal splitting technique (as described in [IPO-MULT]), fast protection is provided for 1+1 LSP protection. However, this technique applies only for unidirectional LSP protection rings. With this type of rings the optical signal is split on both wavelengths (fibers) at the transmission end. 5.3.2 Shared Protection Shared protection is mainly based on the use of an O-APS signalling protocol (using the signalling IP-based control-plane). The requirements as well as the specifications for the specification of such an IP-based protocol need to be seriously determined in order to achieve a fast protection switching mechanism. The specifications of the O-APS protocol are extensively detailed in [IPO-OAPS]. 6. Routing in Optical Rings and Hybrid Mesh-Ring Networks In optical networks, the following hierarchical routing model can be considered: - the intra-domain interfaces are included in an IGP (more precisely link-state routing protocols) routing protocol instance included within the same autonomous-system (AS) - the inter-domain interfaces are defined between autonomous-systems running over eBGP routing protocol. The eBGP protocol is running between border LSRs (which from the OSPF point-of-view are considered as an Autonomous System Boundary Router - ASBR). The hierarchical optical network model provides the capacity to connect several autonomous systems (i.e. optical domains) together through eBGP protocol. 6.1 Intra-Domain Routing The optical network architecture considered here is build such as the corresponding control plane defines an hierarchical optical topology as described in [IPO-FRAME]. Consequently, intra-domain routing can be considered when the IP-based control plane defines an OSPF Autonomous System (AS) on top of an hierarchical optical domain (which defines the transport plane). The same considerations can be applied when using IS-IS IGP protocol in hierarchical optical networks. With the intra-domain architecture, transparent optical network devices (i.e. OXC including O/E/O conversion) are located at the border of the corresponding areas. Consequently, OXC might be considered as internal LSR as well as border LSR. All-optical network devices (i.e. PXC without O/E/O conversion) are predominantly Papadimitriou D. - Internet Draft - Expires May 2002 22 draft-papadimitriou-optical-rings-02.txt November 2001 configured as internal LSR but might in particular cases be configured as border LSR. ++++++++++++++ ++++++++++++++ + A ------ B + + H ------ I + + | | + + | | + + | Area10 | + + | Area20 | + + | | XXXXXXXXXX | | + + D ------ C ---------- E ------ J + +++++++++X | | X+++++++++ X | | X X | Area 0 | X X | | X +++++++++X | | X+++++++++ + K ------ G ---------- F ------ N + + | | XXXXXXXXXX | | + + | Area30 | + + | Area40 | + + | | + + | | + + M ------ L + + P ------ O + ++++++++++++++ ++++++++++++++ Taking into account this optical network topology: - Ring CEFG defines backbone area 0 - Ring ABCD, HIJE, KGLM and FNOP define non-backbone area In this hierarchical network topology, each of these nodes can be considered as an abstract node so that each of them can also be considered as defining an optical ring. Moreover, each of these areas can include internal all-optical rings. However, in that case, the only restriction is related to the Area Border Routers (ABR i.e. border LSR) that can not be all-optical PXC. In order to avoid optical signal regeneration related problems, one can consider here that border LSRs are O/E/O transparent devices like OXCs. 6.2 Inter-domain Routing Inter-domain Optical Routing is considered when defining inter- optical domain routing between several carriers. In the most basic network topology, optical domains (i.e. BGP AS) are inter-connected through direct links (or redundant links) combined with linear protection mechanisms as represented in the following figure: ++++++++++++++ ++++++++++++++ + A ------ B + + H ------ I + + | | + + | | + + | AS 10 | + + | AS 20 | + + | | + + | | + + D ------ C ========= E ------ J + +++++++++++|++ ++|+++++++++++ | | Papadimitriou D. - Internet Draft - Expires May 2002 23 draft-papadimitriou-optical-rings-02.txt November 2001 direct --> | | | | +++++++++++|++ ++|+++++++++++ + K ------ G ========= F ------ N + + | | + + | | + + | AS 30 | + + | AS 40 | + + | | + + | | + + M ------ L + + P ------ O + ++++++++++++++ ++++++++++++++ In this topology, the Nodes CEFG can define an inter-domain ring. This enhancement provides higher resiliency to the classical inter- domain connections since in that case ring protection mechanism can be used. However, the most reliable topology will be achieved when the optical domains are inter-connected through a BGP Transit AS. When considering a Transit AS, the Client AS can be inter-connected by using optical rings or direct links to the Transit AS. As mentioned above the use of a meshed inter-connectivity between nodes does not preclude the use linear protection (and restoration) mechanisms such as dedicated 1+1 or shared (1:1, 1:N, M:N) LSP protection. The following figure compares optical ring from direct link inter- connection approach: + + + + |++| | + + | ---C-+--------+-E--- ---C-+--------+-E--- ++++++ ++++++ +++++++ +++++++ | | | | ++++++++++++++++ ++++++++++++++++++ +Q------------R+ + Q------------R + +||+ + | | + + | Transit AS | + + | Transit AS | + +||+ + | | + +T------------S+ + T------------S + ++++++++++++++++ ++++++++++++++++++ | | | | ++++++ ++++++ +++++++ +++++++ ---G-+--------+-F--- ---G-+--------+-F--- |++| | + + | + + + + Optical Rings Direct Links As mentioned before, a trade-off between link (or wavelength) redundancy and lost capacity need to be defined. However, these considerations are out of the scope of this document. 6.3 Inter-Ring Traffic-Engineering Papadimitriou D. - Internet Draft - Expires May 2002 24 draft-papadimitriou-optical-rings-02.txt November 2001 As described in the previous section, the network hierarchical model considered is divided into Autonomous Systems (ASs), where each AS is divided into IGP areas to allow the hiding, aggregation and summarization of routing information. The hierarchical routing model currently can be extended for traffic engineering as it hides the route taken by a LSP to the destinations in the other routing areas. Hence, from the TE perspective, requirements such as path selection and crank-back mechanism need different architectural additions to the existing link-state routing and signaling protocols for inter- area LSP setup. The protection requirements as well as crank-back mechanisms can be implemented by using the optical ring techniques described in this document. Note that in a fully distributed approach, since LSRs in different AS's have different views of the network at any given time, re- routing and crank-back mechanisms are needed. This approach requires also to clearly defining when the steady state of the optical network is reached in order to execute the protection/restoration mechanisms specified in the escalation strategy to apply when failure(s) occurs. The problems related to LSP protection/restoration are also described in [IPO-REST]. Traffic engineering (TE) practice currently involves the setup and use of Label Switched Paths (LSPs) as dedicated bandwidth pipes between two end points (i.e. between an ingress and an egress LSR). LSPs can be setup across several LSRs through the use of dynamically computed path. Since this memo focuses on fully distributed CSPF path computation, we consider here that the routes can be computed dynamically through the use of online constraint-based routing algorithms such as incremental Linear or Heap Dijkstra Algorithm (or Dijkstra-like algorithms for non-additive metrics) as described for instance in [SPT-DYN]). The online constraint-based routing model requires: (1) a set of flooding mechanism to maintain the state of TE-related information within the optical network topology (2) a constraint-based routing process implemented on certain LSRs that serve as ingress LSRs for the LSPs, (3) a topology change(s) within the optical domain does not result in a complete re-computation of the routes Most of the traffic engineering extensions for intra-area TE routing are described in [OSPF-TE] and [GMPLS-OSPF] the latter in particular when considering intra-area GMPLS TE routing. Note that new traffic engineering attributes can be defined in future versions of this document. We describe here the TE extensions for inter-ring TE routing in optical ring topologies when considering summarization of the TE values at boundaries (see also [OSPF-IRTE] concerning summarization of TE attributes). Papadimitriou D. - Internet Draft - Expires May 2002 25 draft-papadimitriou-optical-rings-02.txt November 2001 As described in [OSPF-IRTE], traffic engineering (TE) summary LSA can be originated at Area Border Router (ABR) into an area. This TE summary LSA is a type-10 opaque LSA flooded within the area. The functionality of the TE summary LSA is similar to the one of summary LSA of standard OSPF [RFC2328], but in addition it carries the traffic engineering metrics to the remote destination (IP network or ASBR). The inter-ring TE summary LSA described here has exactly the same functionality except that the corresponding LSA is generated at ring boundary nodes. 6.3.1 Ring Count The ring count is the cost in rings to reach the destination network (or the destination ring) from the source node (or the source ring). This parameter is equivalent to the hop count defined in [OSPF-IRTE] since a ring appears as an abstract node for the neighboring rings or OXCs (not belonging to the same ring). The ring count (as the hop count) is an additive metric. Note that the ring count can be generalized to an additive weighted metric. 6.3.2 Maximum Reservable Bandwidth The maximum reservable bandwidth is defined as the smallest maximum reservable bandwidth among all the rings included into the path from the source to the destination. Note the maximum reservable bandwidth should be the same for the working and the protected path. 6.3.3 Minimum Reservable Bandwidth The minimum reservable bandwidth is defined as the largest minimum reservable bandwidth among all the rings included into the path from the source to the destination. Note the minimum reservable bandwidth should be the same for the working and the protected path. 6.3.4 Delay The delay attribute is defined as the delay cost to reach the destination ring in microseconds. The delay is defined an average delay taking into account the delay introduced by the worst case protection path and best case working path since this value can not up to now be updated dynamically. The delay is an additive metric that could be generalized to an additive weighted metric. Note that the delay of the working path could be different from the delay of the protected path. 6.3.5 Resource Class/Coloring Papadimitriou D. - Internet Draft - Expires May 2002 26 draft-papadimitriou-optical-rings-02.txt November 2001 The resource class or color of the destination network is a combination of the colors for the various ringed paths to the destination network. Path coloring technique can be found in [IPO- PATH]. Path coloring can be used to prune resources as a first step of the LSP route computation and as a tie when two distinct routes having the same Inter-Ring TE Metric (see section 6.3.6) to the same destination network may be used but only differ from their color. In that case, the choice of one route or another is a local policy matter. 6.3.6 Inter-ring TE Metric The inter-ring TE metric is defined as the TE metric representing the TE cost of reaching the destination network (or the destination ring) from the advertising ring. This metric is composition of ring TE metrics (described in section 4.2) defined by the following formula: Inter-Ring TE Metric = (Working Ring Weight x r1 | Protected Weight x r2) + (Working Ring Load x r3 | Protected Load x r4) + (Maximum Restoration Time x r5) So that the inter-ring TE metric is defined as the weighted sum of the ring metrics and representing the traffic-engineering cost of reaching the destination ring (i.e. Ring N) from the advertising ring (i.e. Ring 1). Ring TE Metric = k[1] x Ring Metric 1 + k[2] x Ring Metric 2 k[n-1] Ring Metric N-1 ... k[n] x Ring Metric N 6.4 Routing Protocol Extensions for Dynamic Ring Configuration We describe here the IGP routing protocol extensions needed in order to enable dynamic ring configuration within an area (intra-area routing) between areas (inter-area routing). 6.4.1 Opaque LSA Opaque LSA are application specific link-state advertisements defined in [OSPF-OPA] used mainly for IGP traffic engineering (TE) extension purposes, see for instance [OSPF-TE] concerning intra-area TE routing and [OSPF-IRTE] concerning inter-area TE routing. The [RFC2370] describing the Opaque LSAs concepts for application specific information distributed additionally to the IGP routing information. In [OSPF-OPA], three types of Opaque LSA were defined: - Type-9 Opaque LSA: link-local scope - Type-10 Opaque LSA: area-local scope (or intra-area scope) Papadimitriou D. - Internet Draft - Expires May 2002 27 draft-papadimitriou-optical-rings-02.txt November 2001 - Type-11 Opaque LSA: Autonomous System (AS) scope (or inter-area scope) The flooding scope associated with each Opaque link-state type is defined as follows. - Type-9 Opaque LSAs are not flooded beyond the local sub-network. - Type-10 Opaque LSAs are not flooded beyond the borders of their associated area. - Type-11 Opaque LSAs are flooded throughout the Autonomous System (AS) and is equivalent to the flooding scope of AS-external (type- 5) LSAs. As a generic rule, see [RFC2370], when flooding Opaque-LSAs to adjacent neighbors, an opaque-capable router looks at the neighbor's opaque capability. Opaque LSAs are only flooded to opaque-capable neighbors. When a non-opaque-capable router inadvertently receive Opaque LSAs, it simply discard the Opaque LSA. Consequently, to achieve intra-area dynamic ring configuration we use Opaque LSAs of Type 10 area-local flooding scope. 6.4.2 Opaque LSA Header Opaque LSAs are types 9, 10 and 11 link-state advertisements. The Link-State ID [RFC2328] of the Opaque LSA is divided into an Opaque Type field (8-bit sub-field) and a Opaque Type-specific ID (24-bit sub-field). The range of topological distribution (i.e., the flooding scope) of an Opaque LSA is identified by its link-state type. In Opaque LSAs are application specific LSAs meaning that their payload has meaning only within a certain application; otherwise, they will be ignored. The Opaque Type that is contained in the Opaque Type field identifies the type of the application. The Opaque Type-specific ID (i.e. Opaque ID) is a 24-bit sub-field sub-divided in a reserved sub-field (8 MSB) and a Source specific sub-field (16 LSB). 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | LS age | Options | LS Type | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Opaque Type | Opaque ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Advertising Node ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | LS sequence number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | LS checksum | Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Papadimitriou D. - Internet Draft - Expires May 2002 28 draft-papadimitriou-optical-rings-02.txt November 2001 where LS Type value = 9, 10 or 11 For the purpose of dynamic ring configuration, only Opaque LSA Type- 10 will be considered. Moreover the dynamic ring configuration Opaque Type still remains to be determined. 6.4.3 Opaque LSA Payload As described in [OSPF-OPA], the Opaque LSA payload consists of one or more nested Type/Length/Value (TLV) structures. The format of the Opaque LSA TLV structure is defined as: 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type | Length (in bytes) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Value | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ The following additional TLV need to be defined in order to enable dynamic ring configuration through the use of application-specific Opaque LSAs. 1. Ring ID TLV The Ring ID is defined as a 32-bit integer field that uniquely identifies a given ring within a given optical domain (i.e. administrative authority). 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Ring ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ A additional field might be defined in the future to identify the Autonomous System number (16-bit) into which the ring is included. 2. Ring Virtual Address TLV Each node included within a given ring shares a unique virtual IPv4 Address with other nodes belonging to the same ring. The IPv4 address is uniquely defined per administrative authority. This information is exchanged between nodes by using the following TLV: 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | 4 | Papadimitriou D. - Internet Draft - Expires May 2002 29 draft-papadimitriou-optical-rings-02.txt November 2001 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Virtual IPv4 Ring Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 3. Loopback IP Address TLV One loopback IPv4 address is allocated per node in order to avoid communication disruption between nodes belonging to the same ring when a given node interface becomes unreachable. The local loopback IPv4 address must be the first one included in the loopback IPv4 address list. For each ring contiguous to a given node, an additional loopback IPv4 address is allocated. Since a ring is viewed as an abstract single node for its neighboring OXC, these loopback IP addresses identify the abstract node constituted by a ring for the neighboring nodes (external to this ring) and neighboring rings. 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | n x 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Loopback IPv4 Address 1 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | / ... / | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Loopback IPv4 Address n | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 4. Ring Protection Type TLV The Optical Ring Protection Type (encoded as an 8-bit flagged field) is defined as: - Ring technology: Optical (Bit0=0) or OTH-based (Bit0=1) - Ring Protection: Dedicated (Bit1=0) or Shared (Bit1=1) - Ring Protection level: Link-level (Bit2=0) or Path-level (Bit2=1) - Ring Protection fibers: 2-fiber (Bit3=0) or 4-fiber (Bit3=1) - Ring switching: ring-switch (Bit4=0) or span-switch (Bit4=1) - Other values (3 bits) are reserved for future use The corresponding TLV is defined as: 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Protection | Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Papadimitriou D. - Internet Draft - Expires May 2002 30 draft-papadimitriou-optical-rings-02.txt November 2001 5. Link TLV To encode the list of SRLG(s) to which a given link belongs, we propose to use the Link TLV by including the following sub-TLVs: - Link ID (32-bit) - Link Local Interface ID (32-bit) - Link Remote Interface ID (32-bit) - Link Encoding Type (8-bit) - Link Protection Type (8-bit) - List of SRLG(s) (n x 32-bit) The corresponding TLV are already defined in [GMPLS-OSPF] and [OSPF- TE]. Note that for optical rings, the relevant Link Protection Type are dedicated 1+1, dedicated 1:1, shared M:N and unprotected 0:1. Additional sub-TLVs can also be included within the Link TLV for intra-ring traffic engineering purposes. So, we can refer to these sub-TLVs as Link TE sub-TLVs: - Maximum Available Bandwidth per link - Minimum and Maximum Reservable Bandwidth per link - Link Priority associated to Minimum/Maximum Reservable Bandwidth - Link Priority associated with the Link Protection Type - Resource Class/Color defined per link The link resource allocation is performed dynamically here; consequently the Link TLV does not need to include the Ring ID to which this link might be allocated (or de-allocated) during the working period of the ring. A given link can include several wavelengths belonging to more than one ring if this is a shared link. This means that the per-ring resource allocation Note also that even if a fiber link can belong to more than one ring (if the link is defined as a boundary link) the wavelengths available on that link when allocated to one ring, they can not be shared by the other ring. 6. Ring Metric TLV As described in section 3.6, the ring metric is a component metric including the ring absolute weight, the ring initial MRT and the ring capacity. The corresponding TLV is defined as follows: 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | 20 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Weight | Initial MRT | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Papadimitriou D. - Internet Draft - Expires May 2002 31 draft-papadimitriou-optical-rings-02.txt November 2001 | R | Incoming | Outgoing | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | R | Dropped | Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | R | Incoming | Outgoing | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | R | Unused | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ The Ring Metric TLV includes: - Ring absolute weight (8-bit): number of nodes belonging to the same ring ID - Ring initial MRT (24-bit): in microsecond units - Ring capacity: . number of working wavelengths: incoming (15-bit), outgoing (15-bit) and dropped (15-bit) . number of protection wavelengths: incoming (15-bit) and outgoing (15-bit) . number of unused wavelengths (30-bit) for this ring 7. Ring Protection Policy TLV Ring protection policy refers to the different strategies applicable to optical rings. The ring protection policy includes mainly the ring protection strategy, the ring protection scheme and ring resource priority and preemptability support. The Ring Protection Policy TLV format is defined as: 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0|1|2|3| Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ The values are defined as follows: The ring protection strategy (bit-0): - Bit 0 set to 0 defines a non-revertive strategy - Bit 0 set to 1 defines a revertive strategy The ring protection scheme (bit-1): - Bit 1 set to 0 defines a provisioned scheme - Bit 1 set to 1 defines a non-provisioned scheme The ring resource Priority-Preemption (bit-2 and bit-3): - Bit 2 set to 0 enables setup pre-emptability of working and protection link resources during LSP setup process (meaning implicit priority support for link resources) Papadimitriou D. - Internet Draft - Expires May 2002 32 draft-papadimitriou-optical-rings-02.txt November 2001 - Bit 2 set to 1 disables setup pre-emptability of working and protection link resources during LSP setup process - Bit 3 set to 0 enables recovery pre-emptability of working and protection link resources during LSP recovery process (meaning implicit priority support for link resources) - Bit 3 set to 1 disables recovery pre-emptability of working and protection link resources during LSP recovery process Other bits (bit 4 to 31) are reserved for future use. 6.5 Routing Protocol Extensions for Dynamic Ring Resource Allocation The only additional TLVs to define as additional Opaque LSA payload is the Ring TE Metric TLV. Other traffic engineering extensions such as Link TE TLVs have already been defined in [OSPF-TE] and [GMPLS- OSPF]. 1. Ring TE Metric TLV The Ring TE Metric is a component metric including the ring weight, the ring load and the ring MRT. The Working TE Metric (16-bit) part of the Ring TE Metric includes the following components - the working ring weight - the working ring load - the computed MRT The Protection TE Metric (16-bit) part of the Ring TE Metric includes the following components: - the protection ring weight - the protection ring load - the computed MRT The Ring TE Metric TLV is encoded as: 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Working TE Metric | Protection TE Metric | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 7. Signaling Extensions in Ring Protected optical networks The signaling extensions are related to the following functions: - signalling of the working and the protected path from source ring node to the destination ring node - specific signaling extension in order to support drop-and-continue functionality; this is equivalent to define signalling extension for point-to-multipoint LSP Papadimitriou D. - Internet Draft - Expires May 2002 33 draft-papadimitriou-optical-rings-02.txt November 2001 Drop-and-continue function can be required when considering protected ring interconnection: - For dedicated-protection ring inter-connection: drop-and-continue function is used together with path selector function - For shared-protection ring inter-connection: drop-and-continue function is used together with service selector function - For both type of ring are inter-connected: drop-and-continued is used together with path selector function (dedicated-protection ring) and service selector function (shared-protection ring) 7.1 Intra-ring signalling extension When entering the ring, the signaling of the working and the protected channel need to be executed simultaneously. The principle defined for this purpose is to 'split' the signaling message (corresponding to the LSP create message) for the working and protected path through the corresponding signalling channel. For this purpose, the same technique as the one defined in [IPO-REST] can be applied. The proposed mechanism applied to emulated rings provides the suitable signalling extensions for the creation of backup LSP segments from the ingress to egress ring node when LSP protection is required. [IPO-MULT] proposes a more efficient mechanism to achieve this by considering the split of the physical optical signal. In that case, fast protection is provided for the 1+1 protected LSP. However, this technique applies only for unidirectional LSP protection rings. With this type of rings the optical signal is split on both wavelengths (fibers) at the transmission end. This results in two copies of the signal propagating the ring in opposite directions. The far-end (or receive-end) switches to the protection wavelength (or fiber) if a failure occurs. The proposed mechanism can restore a single link failure or a single node failure. When using 1+1 LSP protection, since the far-end (receiving end) does need to notify the near-end (transmitting end) of the failure, an O- APS signalling mechanism is not required. The decision is made at the far-end on an LSP-by-LSP basis. 7.2 Inter-ring signalling extension Inter-ring signalling extensions are considered for drop-and-continue ring inter-connection. In this case, the signalling message (i.e. the LSP create message) need to be continued meaning duplicated when reaching the primary destination node (far end) for that LSP. However only one LSP must be defined (in particular only one LSP segment needs to be defined from the ingress to egress ring nodes). Details concerning inter-ring signalling extensions are left for further study. Papadimitriou D. - Internet Draft - Expires May 2002 34 draft-papadimitriou-optical-rings-02.txt November 2001 8. Security Considerations Security considerations are left for further study. 9. References 1. [GMPLS-ARCH] E. Mannie et al., 'Generalized Multi-Protocol Label Switching (GMPLS) Architecture,' Internet Draft, draft-ietf-ccamp- gmpls-architecture-00.txt, July 2001. 2. [GMPLS-CRLDP] P. Ashwood-Smith, L. Berger et al., 'Generalized MPLS Signaling - CR-LDP Extensions,' Internet Draft, draft-ietf- mpls-generalized-cr-ldp-04.txt, July 2001. 3. [GMPLS-ISIS] K. Kompella et al., 'IS-IS Extensions in Support of Generalized MPLS,' Internet Draft, draft-kompella-isis-gmpls- extensions-04.txt, September 2001. 4. [GMPLS-OSPF] K. Kompella et al., 'OSPF Extensions in Support of Generalized MPLS,' Internet Draft, draft-ietf-ccamp-ospf-gmpls- extensions-00.txt, September 2001. 5. [GMPLS-RSVP] P. Ashwood-Smith, L. Berger et al., 'Generalized MPLS Signaling - RSVP-TE Extensions,' Internet Draft, draft-ietf-mpls- generalized-rsvp-te-05.txt, October 2001. 6. [GMPLS-SIG] P. Ashwood-Smith, L. Berger et al., 'Generalized MPLS Signaling - Signaling Functional Requirements,' Internet Draft, draft-ietf-mpls-generalized-signaling-06, October 2001. 7. [GMPLS-BUNDLE] K. Kompella et al., 'Link Bundling in MPLS Traffic Engineering,' Internet Draft, draft-kompella-mpls-bundle-05.txt, March 2001. 8. [IPO-OAPS] D.Guo, D. Papadimitriou et al., 'Optical Automatic Protection Switching - An IP-based Protocol', Work in Progress, draft-guo-optical-aps-00.txt, November 2001. 9. [IPO-BUNDLE] B. Rajagopalan et al., 'Link Bundling in Optical Networks,' Internet Draft, draft-rs-optical-bundling-01.txt, November 2000. 10. [IPO-FRAME] B. Rajagopalan et al., 'IP over Optical Networks: A Framework,' Internet Draft, draft-ietf-ip-optical-framework- 00.txt, July 2001. 11. [IPO-MPLS] D. Awduche et al., 'Multi-Protocol Lambda Switching: Combining MPLS Traffic Engineering Control With Optical Crossconnects,' Internet Draft, draft-awduche-mpls-te-optical- 03.txt, April 2001. Papadimitriou D. - Internet Draft - Expires May 2002 35 draft-papadimitriou-optical-rings-02.txt November 2001 12. [IPO-MULT] D. Papadimitriou et al., 'Optical Multicast - An Architectural Framework,' Internet Draft, draft-poj-optical- multicast-02.txt, November 2001. 13. [IPO-POES] D. Papadimitriou et al., 'Packet-Optical Escalation Strategies,' Internet Draft, draft-pbh-packet-optical-escalation- 01.txt, May 2001. 14. [IPO-REST] J. Hahm et al., 'Restoration Mechanisms and Signaling in Optical Networks,' Internet Draft, draft-many-optical- restoration-02.txt, February 2001. 15. [IPO-RFRM] N. Ghani et al., 'Architectural Framework for Automatic Protection Provisioning In Dynamic Optical Rings,' Internet Draft, draft-ghani-optical-rings-01.txt, March 2001. 16. [IPO-OMR] D. Guo et al, 'Hybrid Mesh-Ring Optical Networks,' Internet Draft, draft-guo-optical-mesh-rings, January 2001. 17. [IPO-SRLG] D. Papadimitriou et al., 'Inference of Shared Risk Link Groups,' Internet Draft, draft-many-inference-srlg-02.txt, November 2001. 18. [LMP] J. Lang et al., 'Link Management Protocol (LMP) v1.0', Internet Draft, Work in progress, draft-ietf-ccamp-lmp-02.txt, October 2001. 19. [LMP-WDM] A. Fredette et al., 'Link Management Protocol (LMP) for WDM Transmission Systems,' Internet Draft, draft-fredette- lmp-wdm-02.txt, July 2001. 20. [MPLS-CRLDP] B. Jamoussi et al., 'Constraint-Based LDP Setup using LDP,' Internet Draft, draft-ietf-mpls-cr-ldp-05.txt, March 2001. 21. [OPT-MWS] T. Stern and K. Bala, 'Multiwavelength Optical Network - A Layered Approach,' Addison Wesley Longman, Inc, May 1999. 22. [OPT-RINGS] G. Ellinas et al, 'Protection Cycles in Mesh WDM Networks,' IEEE Journal on Selected Areas in Communications, Volume 18, Number 10, October 2000. 23. [OSPF-TE] D. Katz and D. Yeung, "Traffic Engineering Extensions to OSPF,' Internet Draft, draft-katz-yeung-ospf-traffic-06.txt, September 2001. 24. [OSPF-IRTE] S. Venkatachalam et al, 'OSPF Extensions to Support Inter-Area Traffic Engineering,' Internet draft, draft- venkatachalam-ospf-traffic-00.txt, November 2000. 25. [RFC2328] J. Moy, 'OSPF Version 2,' Internet RFC - Standard Track, April 1998. Papadimitriou D. - Internet Draft - Expires May 2002 36 draft-papadimitriou-optical-rings-02.txt November 2001 26. [RFC2370] R. Coltun, 'The OSPF Opaque LSA Option,' Internet RFC 2370 - Standard Track, July 1998. 27. [SPT-DYN] P. Narvaez et al., 'New Dynamic Algorithms for Shortest Path Tree Computation,' IEEE/ACM Transactions on Networking, Volume 8, Number 6, December 2000. 10. Acknowledgments The authors would like to be thank Bernard Sales, Emmanuel Desmet, Hans De Neve, Fabrice Poppe, Stefan Ansorge, Katrien Skerra, Mathieu Garnot, Gert Grammel and Jim Jones for their constructive comments, suggestions and inputs. 11. Author's Addresses Dimitri Papadimitriou Alcatel Francis Wellesplein 1, B-2018 Antwerpen, Belgium Phone: +32 3 240-84-91 Email: Dimitri.Papadimitriou@alcatel.be Papadimitriou D. - Internet Draft - Expires May 2002 37 draft-papadimitriou-optical-rings-02.txt November 2001 Full Copyright Statement "Copyright (C) The Internet Society (date). All Rights Reserved. This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English. The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns. This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE." Papadimitriou D. - Internet Draft - Expires May 2002 38