Patent Application
20070268821
At least some of the nodes of network 20 are connected by a layer 2 ring network 28. In the example of
Each ring node comprises a network interface 40 for communicating with other ring nodes over ring 28. In some embodiments, such as in nodes 24A and 24B, interface 40 is also used for communicating with network nodes outside of ring 28. Each ring node comprises a processor 44, which carries out, inter alia, methods related to establishing communication paths through network 20, as described below. Processor 44 may comprise a general-purpose computer, which is programmed in software to carry out the functions described herein. The software may be downloaded to the computer in electronic form, over a network, for example, or it may alternatively be supplied to the computer on tangible media, such as CD-ROM. Alternatively, processor 44 may comprise one or more hardware logic components (hard-wired or programmable) or a combination of hardware- and software-implemented elements.
In some embodiments, at least one of the ring nodes comprises a bandwidth broker (BWB) module 48, which carries out resource (e.g., bandwidth) reservation functions for the entire ring network. These functions are also referred to as ring-level connection admission control (CAC) functions. In some embodiments, BWB 48 provides the ring nodes with up-to-date information regarding resource allocations in the ring network, to enable distribution of the TE-related attributes. This process is described in greater detail further below.
CAC and bandwidth allocation methods in ring configurations are described, for example, in U.S. Patent Application Publication 2004/0085899 A1, whose disclosure is incorporated herein by reference. Resource reservation methods and other traffic engineering aspects in ring networks are also described in U.S. Pat. No. 6,963,537 and in U.S. Patent Application Publication 2003/0103449 A1, whose disclosures are incorporated herein by reference. Another exemplary bandwidth manager is described by Yavatkar et al. in IETF RFC 2814 entitled “SBM (Subnet Bandwidth Manager): A Protocol for RSVP-Based Admission Control over IEEE 802-Style Networks,” May 2000, which is incorporated herein by reference.
Typically, BWB 48 is implemented as a software process or thread running on processor 44. In some embodiments, the BWB functionality is active in only one of nodes 24, i.e., only one bandwidth broker is active in ring network 28 at any given time.
Some IP networks use layer 3 protocols, such as the OSPF and OSPF-TE protocols cited above, for determining the routing of a communication path between a source node and a destination node through the network. In principle, the OSPF-TE protocol distributes (“advertises”) TE-related information of communication links in the network using link state advertisement (LSA) messages. Network nodes that support OSPF-TE, typically comprising layer 3 routers, can use this information to determine an optimal routing path, taking into account the TE-related information as constraints. Typically, each node stores the advertised information in a TE database (TED) 50.
When a request is accepted to establish a new communication path from a source node to a destination node, the source node determines the optimal routing path based on the TE-related information stored in TE database 50. In some embodiments, the source node may take into account additional constraints specified in the request when determining the optimal routing path.
Once the optimal path is determined, resources (e.g., bandwidth) are reserved in the nodes and/or links along the path using a reservation protocol such as RSVP-TE. Thus, in an MPLS network, OSPF, OSPF-TE and RSVP-TE are used to set up a unidirectional label switched path (LSP), also referred to as an MPLS tunnel, from the source to the destination node.
Although the embodiments described herein mainly refer to the routing of LSP using OSPF, OSPF-TE and RSVP-TE, the disclosed methods and systems can also be used with other distribution, routing and/or reservation protocols. For example, the methods and systems described herein can be used with the Intermediate System to Intermediate System (IS-IS) link state protocol and its extension for traffic engineering (IS-IS-TE). These protocols are described by Smit and Li in “Intermediate System to Intermediate System (IS-IS) Extensions for Traffic Engineering (TE),” IETF RFC 3784, June 2004, and by Kompella and Rekhter (editors) in “Intermediate System to Intermediate System (IS-IS) Extensions in Support of Generalized Multi-Protocol Label Switching (GMPLS),” IETF RFC 4205, October 2005, which are incorporated herein by reference.
However, known layer 3 routing and reservation protocols are best suited for establishing communication paths over point-to-point links and are incapable of accounting for more complex layer 2 entities, such as ring networks. For example, as noted above, RPRs are typically represented in IP networks as multi-access interfaces. The multi-access interface representation does not preserve the topological structure of the RPR, i.e., the fact that it comprises two ringlets and the different hop distances between different ring nodes. The multi-access representation also does not address the available resources on the different ring segments and ringlets. All this information, which may significantly affect routing decisions through the ring, is lost.
The methods and systems described herein enable the use of distribution and routing protocols such as OSPF-TE and OSPF over ring networks, while preserving the ring topology and the traffic engineering (TE) related attributes of the different ring segments and ringlets. In principle, the RPR network is first represented as a plurality of point-to-point links that connect pairs of ring nodes. The point-to-point links are advertised, or distributed across the network, in accordance with the OSPF-TE protocol. Each point-to-point link is associated with one or more TE-related attributes, as will be described in detail below. The TE-related attributes are distributed as part of the link advertisement messages.
Nodes 24 of network 20 receive the OSPF-TE advertisement (LSA) messages. The nodes use the TE-related attributes carried in the messages to construct and update TE database 50. Thus, each node 24 maintains an up-to-date database of the advertised links 20 and their TE-related attributes. Generally, an arbitrary node 24 does not distinguish between point-to-point links that are part of the representation of ring network 28 and between other point-to-point links of network 20 unrelated to the ring.
The representation of ring network 28 as a plurality of point-to-point links thus provides detailed information regarding the routing options through the ring network to nodes 24. When a particular node 24 is required to make a routing decision, for example using OSPF, the node takes into account the advertised TE-related attributes, and is thus able to make better routing decisions. Note that link representation and advertisement can be carried out using standard OSPF-TE mechanisms, without modification.
Each link 52 corresponds to an alternative path, which may be chosen through the ring. Consider, for example, the alternative paths that may be chosen to connect ring nodes 24A and 24B. Traffic can be sent from node 24A to node 24B over two alternative paths: (1) directly over CW ringlet 32, or (2) via nodes 24D and 24C over CCW ringlet 36. Similarly, traffic from node 24B to node 24A can also be sent over two alternative paths: (1) directly over CCW ringlet 36, or (2) via nodes 24C and 24D over CW ringlet 32. Thus, the connectivity between ring nodes 24A and 24B can be represented using a total of four unidirectional point-to-point links 52. Note that some of links 52 represent physical paths that traverse several network segments and ring nodes. The collection of links 52 thus fully preserves the topology of RPR network 28.
Each link 52 has one or more TE-related attributes. In the context of the present patent application and in the claims, the term “TE-related attribute” is used to describe any link property that may affect a decision to route or to refrain from routing a communication path through it. For example, TE-related attributes may comprise a maximum bandwidth of the link (also referred to as the link capacity or link rate), a definition of the maximum reservable bandwidth (which may be different from the link capacity, for example when allowing a certain amount of overbooking), and/or a currently available (unreserved) bandwidth over the link.
Some TE-related attributes may be related to the ring topology. For example, an attribute may identify the ringlet (e.g., CW ringlet 32 or CCW ringlet 36) used by the link in question. Another attribute may comprise the number of hops (ring segments) traversed by the link. In some embodiments, an attribute may comprise an estimate of the round-trip time (RTT) over the link, which is conventionally measured in RPR networks. Other TE-related attributes may have an administrative nature or have to do with certain network policies. For example, attributes may define the security status of the link, a metric indicating the cost of passing traffic over the link, and/or an indication that the link belongs to a different service provider or even to a different country. Additionally or alternatively, any other suitable link property can be used as a TE-related attribute.
In some embodiments, each ring node transmits OSPF-TE advertisement messages advertising the outgoing links 52 that are directed from it to the other ring nodes. In other words, for every pair of ring nodes denoted X and Y. node X advertises two unidirectional point-to-point links directed from it to node Y. one link traversing CW ringlet 32 and one link traversing CCW ringlet 36. Although in OSPF-TE and other TE protocols the ring nodes typically advertise their outgoing links, the methods described herein are not limited to advertising outgoing links, and can be similarly used with protocols in which each ring node advertises the inbound links directed to it from the other ring nodes.
OSPF-TE advertisement messages allow for optional type length value (TLV) fields. When distributing the TE-related attributes, the messages may carry at least some of the following data is TLV fields and/or sub-fields:
An identifier marking the advertised link as a point-to-point link (as opposed to a multi-access link).
A local IP address identifying the RPR interface of the node transmitting the message. This value is typically configured manually by an administrator as part of the configuration of the ring node.
A remote IP address identifying the RPR interface of the ring node to which the advertised link is directed. This value is typically determined automatically by the advertising ring node from RPR topology messages, as specified in IEEE 802.17 cited above.
A traffic engineering (TE) metric. This field gives a quantitative TE-related metric or weight, indicating the desirability of including the advertised link in a communication path. For example, a TE metric may comprise the number of hops traversed by the link, assuming that using short links in the routing path is preferable over using longer links. Another TE metric may comprise the RPR round-trip time (RTT) estimate described above, which also gives more weight to shorter links. Various policy-related or administrative attributes described above can also be used as TE metrics.
The maximum bandwidth of the advertised link.
The maximum reservable bandwidth over the advertised link. This value may be greater than the maximum bandwidth when a certain amount of overbooking is allowed over the link, in accordance with a predetermined overbooking profile.
The currently available bandwidth over the advertised link.
An indication of the ringlet used by the advertised link. Since the two unidirectional links connecting a pair of ring nodes (over the two ringlets) have identical local and remote IP addresses, this field can be used to distinguish between links traversing the CW and the CCW ringlets.
An identifier indicating the affiliation of the advertised link to a particular administrative group. For example, Awduche et al. describe the use of administratively assigned parameters referred to as resource class attributes in “Requirements for Traffic Engineering over MPLS,” IETF RFC 2702, September 1999, section 6.2, page 21, which is incorporated herein by reference.
Some of the TE-related attributes can be defined manually by a user, such as a network administrator or designer. Other attributes can be determined automatically by the advertising ring node. For example, in RPR networks, the number of hops traversed by a particular link can be measured automatically using the topology messages transmitted among the ring nodes. The total link capacity can also be deduced automatically using the physical layer properties known to the advertising ring node.
In some embodiments, each ring node obtains and maintains the information required for advertising its respective outgoing links, and advertises them without coordination with the other ring nodes. In these embodiments, each ring node should be aware of the current ring topology and the current status of bandwidth allocations over the entire ring network.
In alternative embodiments, one of the ring nodes is defined as a master. The master ring node obtains and maintains the TE-related information required for advertising all point-to-point links of the ring network. The master updates the other ring nodes with the up-to-date link status, so as to enable them to advertise their respective outgoing links correctly. A hybrid configuration, in which some of the TE-related attributes are determined locally by the advertising node and some attributes are provided by the master ring node, can also be used.
Typically, the ring node that currently operates bandwidth broker 48 is also chosen to serve as the master for maintaining advertisement-related information. When establishing a new communication path, the master ring node approves the resource allocations in the ring for the selected path. Typically, any of the ring nodes is capable of carrying out the master functionality, but only one master is active at any given time. If the current master fails, another ring node may replace it. Any suitable logic can be used to select the currently-active master or replace a failed master.
In order to reduce processing overhead in network nodes, the OSPF and OSPF-TE protocols allow the network to be partitioned into areas. LSA messages and OSPF/OSPF-TE operation in general do not cross area boundaries. In these scenarios, the advertisement of point-to-point links 52 of ring 28 is also limited to the area containing ring network 28.
Each ring node advertises its respective outgoing links by transmitting OSPF-TE advertisement (LSA) messages, at an advertising step 62. The advertisement messages comprise the TE-related attributes, thereby distributing the attributes to nodes of network 20. In general, the attributes are distributed both to the other ring nodes and to nodes outside ring network 28, if such nodes exist. The different nodes of network 20 receive the advertised links and attributes, and use them to update their TE databases 50.
When it is required to set up a new communication path (a LSP in the present example) from a source node to a destination node in network 20, a request is accepted at the source node, at a request acceptance step 64. In general, it is assumed that the optimal routing path traverses ring network 28, although the source and destination nodes may comprise any node in network 20, and not necessarily ring nodes in network 28. The source node determines the optimal routing path to the destination node, at a path determination step 66. The source node determines the optimal path based on its locally-stored TE database 50, which was previously constructed and updated with the TE-related attributes advertised at step 62 above. As noted above, the source node may also take into account additional constraints when determining the optimal routing path. These constraints are typically specified in the request, but may alternatively be defined in advance.
The source node then establishes the communication path using the optimal routing, at a path establishment step 68. As part of the path establishment, resources are reserved in the different nodes and links along the path, for example using the RSVP-TE protocol. Within ring network 28, resources are typically reserved by bandwidth broker 48.
Referring, for example, to
24E→24A→24D (via CCW ringlet 36).
24E→24A→24B→24C→24D (via CW ringlet 32).
24E→24B→24A→24D (via CCW ringlet 36).
24E→24B→24C→24D (via CW ringlet 32).
Using the advertised link attributes, source node 24E has all the information necessary to apply OSPF-TE and select, out of the four alternative paths, the optimal path to destination node 24D.
In some cases, TE-related attributes change over time. When such changes affect the TE-related attributes of one or more of point-to-point links 52, the method may return to advertising step 62 above in order to re-advertise the links whose TE-related attributes have changed.
Although the embodiments described herein mainly refer to the establishment of MPLS LSP over RPR networks using OSPF-TE, the methods and systems described herein can also be used in other applications, such as the IS-IS and IS-IS-TE protocols cited above.
Additionally or alternatively to ring networks, the methods and systems described herein can be used for representing and communicating over any layer 2 multi-access network in which resources are allocated in only part of the network. In such resource allocation mechanisms, network resources (e.g., bandwidth) are allocated to a particular packet or traffic flow in only a subset of the network's segments or topology. Since packets or traffic flows are confined to particular areas of the network, they do not consume the entire multi-access medium. Thus, a more efficient use of network resources is achieved. For example, resource allocation mechanisms may be employed in various local area network (LAN) configurations. In such configurations, the logical topology is typically different from the physical topology of the network.
It will thus be appreciated that the embodiments described above are cited by way of example, and that the present invention is not limited to what has been particularly shown and described hereinabove. Rather, the scope of the present invention includes both combinations and sub-combinations of the various features described hereinabove, as well as variations and modifications thereof which would occur to persons skilled in the art upon reading the foregoing description and which are not disclosed in the prior art.
