Method and apparatus for exchanging routing information and establishing communication through multiple network areas

FIELD OF THE INVENTION

The present invention relates to an Ethernet network, and more particularly to a method and apparatus for exchanging routing information and establish connectivity via a plurality of network segments.

PRIOR ART

The architecture of an Ethernet network devices connected to the network compete for the ability to share telecommunications paths at any given time. Where use multiple bridges or nodes to connect network segments, often there are many potential paths to the same destination. The advantage of this architecture is that it provides path redundancy between bridges and can increase the performance of the network in the form of additional channels. However to prevent loops spanning tree mainly used to limit the manner in which broadcast traffic is transmitted over the network. Since the routes are established by the broadcast frame and waiting for a response, and as both the request and response will follow the spanning tree, most, if not all traffic, followed the channels that were part of the spanning tree. This often resulted in excessive use of the channels that were on the spanning tree without using channels which are not part of the spanning tree.

To overcome some of the limitations inherent in Ethernet networks, Ethernet protocol network controlled by a channel state has been disclosed in U.S. Patent Application №11 / 537,775, reported 2 October 2006, entitled "Routing of channel provider", the content of which is incorporated herein by reference . As described in detail in this application, the nodes in the routing protocol as the channel controls the exchange of greetings, an Ethernet, to explore the neighboring nodes in the network and to transmit status messages channel such as "advertisements" to allow each node in the network to create a database on the state of channel. The packages of channels included a metric related to your channel. Typically, this is interpreted as a distance metric. Database of the channel can then be used to calculate the shortest signal path through the network. Each node then fills forwarding information base (FIB), which will be used by the node to decide on sending so that the frames would be transferred to the destination by the shortest route. As to a specific destination always use the shortest route network traffic to be distributed over a larger number of channels and follow the optimal path to more nodes than via a single spanning tree or even a plurality of spanning trees, which are used to carry traffic in the network.

When customer traffic injected into a network provider, client address assignment frame MAC (C-MAC DA) is transmitted by the provider MAC address (B-MAC DA), so that the provider may transmit the traffic on the network using the MAC address space provider. Additionally, provider network configured network elements to transmit the traffic based on Virtual LAN ID (VID) so that different frames would be addressed to the same recipient, but with different VID can be transmitted through the network along different paths. When the network Ethernet, Control Protocol channel state may associate one VID range with transmission on the shortest path for transmitting unicast and multicast traffic using a VID that range, and traffic can be transmitted in a network path, except for a shortest path using a second VID range . Using the developed traffic paths (TE) through a protocol of the channel controlled Ethernet network described in greater detail in US patent application №11 / 732,381, filed Apr. 3, 2007, entitled "Engineering the way the protocol and as of the channel in a managed network Ethernet", the content of which is incorporated herein by reference.

Routing protocols include channel as open shortest path (OSPF) and intermediate system to intermediate system (IS-IS). This network of channels can be increased only at the point where the time for the control plane pereskhodimosti channel state becomes unacceptable because of the complexity of calculations required that grows exponentially in proportion to the size of the network. To pass this point protocols used as a channel to separate network segments. As the IS-IS, OSPF and restricted to the two-level hierarchies: backbone single zone (Level 2 IS-IS) with a dead-end segments of the level 1 (L1).

Routing stateful channel provider (PLSB), wherein the IS-IS protocol is applied to the bridges in an Ethernet network providers, the bridge that connects the two (or more) segments, the boundary is called the bridge network segments (ABB). For reliability, it is desirable to have a set of bridges between any ABB segment L1 and a segment of a single Layer 2 (L2). Application IS-IS protocol in IP networks is well known in the art. However, there is a significant difference between the Internet Protocol (IP) and PLSB, which leads to a trial and the true paths of its IP traffic destined to the segment, are not always applicable to PLSB. For example, if the IP is based on subnets, is it possible to transmit the packet toward the boundary segments is simple.

IP is a connectionless performed, so that transmission of the packet the closest boundary segments router (ABR), where the IP network is equivalent to the nearest ABB, will always work. IP does not require symmetry path, thus, the package can exit from the region on one ABB, and a reverse packet may arrive at different ABB, whereas, for reasons relating to the multicast Ethernet and the operating tools, a PLSB path between two end points should be the same in both directions. In addition, the IS-IS protocol for the IP and OSPF do not support multicast routing, although multicast trees are a major part PLSB. The Ethernet is desirable (and necessary for project design PLSB), to multicast packets followed the same route as unicast packets to the same destinations.

Currently Protocol IS-IS allows the channel to be in the segment L1, and LANs L2, but PLSB not provide indication to ABB, to determine whether the incoming packet is to be processed as outputted from L1 or L2 in the determination of the next hop. Also, there are no conditions for the processing scenario in which a single ABB serves many non-overlapping segments L1.

Therefore, a system and method for forwarding packets without looping PLSB network with a plurality of segments, wherein the segments L1 can serve multiple ABB, and one ABB can serve a plurality of segments.

SUMMARY OF THE INVENTION

The present invention provides a method, apparatus and system to ensure that multicast packets follow the same nezatsiklennym way as the way in which follow unicast packets on the network packet. In general, to packet networks, where a segment (L1) can be maintained at any level bridges boundary plurality of network segments (ABB), when a single forwarding information base (FIB) is insufficient. The present invention provides the use of specific and different FIB, depending on whether the packet arrives at the port level L1 or the port layer 2 (L2).

In accordance with one aspect of the present invention, the process ensures that multicast packets follow the same path nezatsiklennomu as unicast packets in the packet network. The network packet includes at least one first segment, a certain first level. Each first segment includes a first plurality of nodes connected to the first set of channels. Each of the first segment is connected to at least one node through a second boundary segments of the segment defined by the second layer. The second segment includes a second plurality of nodes connected to the second set of channels. Each node boundary segment includes at least one port of the first level connected to each first segment, and the port of the second level connected to the second segment. Each multicast packet transmitted by nezatsiklennomu path includes a header having a root identifier identifying the root of the multicast tree. At least one data packet receiving node boundary segments of the network, responsive to receipt of a multicast packet to the port of the second-level node boundary network segments, wherein the root identifier of the multicast packet is analyzed. If the multicast packet to be transmitted at least at one of the ports of the first-level node segment boundary network, another root identifier is entered into the packet before transmitting the packet, at least one port of the first level.

In accordance with another aspect of the present invention, a node boundary segments for use in the packet network. The network packet includes at least one first segment, a certain first level. Each first segment includes a first plurality of nodes connected to the first set of channels. Each of the first segment is connected to a second segment defined by the second layer. The second segment includes a second plurality of nodes connected to the second set of channels. Node boundary segment includes at least one port of the first level, the first segment corresponding to each port of the second level, corresponding to the second segment, and at least one processor. Port of the first layer is used for receiving data packets from the corresponding segment, and for transmitting data packets corresponding to the first segment. Port of the second layer is used for receiving data packets from the corresponding segment and to transmit data packets to the second segment. The at least one processor is electrically coupled to each port of the first-level and second-level port. Responsive to receiving a multicast packet port of the second level receives a packet containing a header having a root identifier identifying the root of the multicast tree; at least one processor is used to analyze the root identifier multicast packet and determining whether the multicast packet must be transmitted at least at one of the ports of the first-level node segment boundary network. If the multicast packet to be transmitted at least at one of the ports of the first level, the processor replaces the root identifier in the packet to initiate transmission of a packet port of the first level.

In accordance with another aspect of the present invention, a packet communication system comprises a second segment, at least one first segment and at least one node of the boundary segments. At least one first segment is connected to the second segment. The second segment of the first segment, and each configured for link-state protocol, managed network segment Ethernet, comprising a plurality of nodes connected by a number of channels. At least one node boundary segments connecting each first segment to the second segment and is used to serve two or more first non-overlapping segments. Each node includes a port boundary segments of the second level, the at least one port of the first level and the at least one processor. The second port is used for receiving data packets from the corresponding segment, and for transmitting the data packets of the second segment. Each port of the first level is used for receiving data packets from the corresponding segment, and for transmitting data packets corresponding to the first segment. The at least one processor is electrically coupled to the second port level with each port of the first level. Responsive to receiving a multicast packet port of the second level receives a packet containing a header having a root identifier identifying the root of the multicast tree. A processor is used to analyze the root identifier multicast packet and determining whether the multicast packet to be transmitted, at least one of the ports of the first-level node segment boundary. If the multicast packet to be transmitted at least at one of the ports of the first level, the processor replaces the root identifier in the packet to initiate transmission of a packet port of the first level.

BRIEF DESCRIPTION OF THE DRAWINGS

Objects of the present invention are listed in the subclaims. The present invention is illustrated by the examples in the accompanying drawings, wherein like reference numerals refer to like elements. The drawings disclose various embodiments of the present invention for purposes of illustration and do not limit the scope of the invention. For clarity, not every component may be labeled by the symbol.

In the drawings:

Figure 1 - is a functional block diagram of one example, channel state protocol controlled Ethernet network;

Figures 2 and 3 - functional block diagram of an exemplary set of Ethernet network segments, controlled by protocol channel state according to one embodiment of the invention;

Figure 4 - a functional block diagram of the partition ABB, which performs the separation of network segments and hierarchical routing, and which shows the process used to place the identity information of common interests between network segments so that the way could be held between the control segments the Ethernet protocol channel state according to one embodiment of the invention;

Figure 5 - is a functional block diagram of a network element that can be used as a bridge boundary segments (ABB) on the border between two Ethernet networks, link-state protocol controlled according to one embodiment of the invention;

Figure 6 - is a functional block diagram of a network configured for use of recursion, to ensure separation of the network according to one embodiment of the invention; and

Figure 7 - functional block diagram of a two-level routing of network channel provider ("PLSB"), ABB has multiple interfaces, created in accordance with the principles of the present invention.

DETAILED DESCRIPTION

Standard IEEE 802.1ah-2008 defines the main bridges provider defines a new Ethernet header, informally known as the "MAC in MAC", provides a complete separation and Ethernet address of the client and the provider allows the network provider to offer large amounts of customer service facilities, such as a transparent service facilities segments of the LAN. Using a link-state protocol 802.1 ah management backbone Ethernet network provider allows Ethernet networks scale from space to the user LAN and WAN to (WAN), providing a more efficient use of network bandwidth to transfer nezatsiklennogo shortest path. Instead of using unexplored network view at each node using the Spanning Tree Protocol (STP), an algorithm combined with transparent bridging connections in the protocol as channel controls the routing of network Ethernet, forming a cellular network sharing broadcast messages about the state of the channel to allow each node to have synchronizes representation of the network topology. This is achieved through a well-understood mechanism of routing of the channel system. The bridges in the network have to sync representation of network topology, the required knowledge and capabilities unicast forwarding multicast connectivity can compute the shortest path between any pair of bridges on the network and individually may fill their forwarding information bases (FIB) according to the calculated representation of the network.

When all nodes to define its role in the synchronized network performance and filled their base FIB, the network will be unicast nezatsiklennoe tree toward any given bridge from the set of peer bridges (those who for whatever reason require shipment to this bridge); and congruent and nezatsiklennoe tree from one point to many points (r2mp) from any given bridge to the same set or subset of peer bridges at the service object placed in the bridge. The result is a path between a given pair of bridges, not limited to the transfer of the root bridge spanning tree, and the overall result may be to make better use of the width of the cell provider. Basically, each bridge is the root of one or more spanning trees which define unicast communication opportunity to the bridge, and the ability to multicast from the bridge.

Link-state protocol controlled Ethernet network provides the equivalent of Bridged Ethernet, but achieves this through a configuration FIB network element, rather than by addressing and avalanche training. As such it can be used in the new standards, such as standard IEEE 802.1ah, the Institute of Engineers of Electrical and Electronics Engineers, a draft of which is called "routing trunk line service provider, or MAC-in-MAC transmission configuration B-MAC (main MAC), and trivial modifications adaptation function BEB, to the broadcast behavior of the client in multicast, so that an Ethernet client can use the connectivity offered by the link-state protocol controlled Ethernet network without modification. Configuration The MAC may be used to create a communications capabilities along the shortest path without loops (for the purposes of unicast and multicast) between a number of (slightly modified standard IEEE 802.1ah) backbone provider bridges to provide transparent LAN service on the level of C-MAC (MAC client ) or other network-level, who can use the service transparent LAN.

Referring now to the drawings in which like characters denote the same elements. Figure 1 shows a functional block diagram of one example of part of the link-state protocol controlled Ethernet network 10. As shown in Figure 1, the network 10 in this example includes a plurality of bridge nodes 12 connected by channels 14. The nodes 12 communicate bridge "greeting" to explore the neighboring units and communicate the channel state, allowing each node to create a database of channel, which can be used to compute shortest paths between input and output nodes. Additional details associated with this example include the link-state protocol controlled Ethernet network, as described in U.S. Patent Application №11 / 537,775, registered 2 October 2006, entitled "Routing of channel provider", the content is incorporated herein by reference.

Two examples of routing protocols include channel as open shortest path protocol (OSPF) routing protocol and intermediate systems (IS-IS), can also be used at other routing protocols as the channel. IS-IS is described, for example, in ISO 10589, and IETF RFC 1195; the contents of each of which are incorporated herein by reference. Although there are current versions of this protocol, the invention is not limited to the embodiment based on the current version of the standard, as it can be adapted to work with future versions of the standards as they are developed. Similarly, the invention is not limited to the embodiment which operates in conjunction with one of these specific protocols, as other protocols may also be used to exchange routing information.

In addition to the setting of the unicast forwarding the shortest path nodes may also install forwarding state for multicast trees in the network. An example of a method for implementing multicast forwarding protocol state of the channel in a controlled Ethernet network is described in more detail in US patent application №11 / 702,263, registered on 5 February 2007, entitled "Multicast embodiment in the routing protocol as the channel in a controlled Ethernet network," the contents of which It incorporated herein by reference. As described in this application, alert channel state may be used to advertise multicast group sending initiate forwarding state the network. In particular, each tree supporting Multicast this group may be assigned a unique identifier, such as root identifier used as the destination MAC address (DA) in order to transmit the multicast frames on the network. Nodes in the network status of the shipment is set for a tree root / group of the shortest path from the multicast root node to one destination to the group Multicast. Figure 1 illustrates a multicast tree having a root node F, when the destination nodes (A, B, C, E and H) show an interest in one or more groups Multicast i.e. "Have an interest" in one or more multicast groups that have an element in F. Node D, for example, is set in the tree (set forwarding state for the root), because it is on the shortest path between the node F and node A.

Interest in a multicast transmission may be based on a common interest in a security identifier (SID), such as the I-SID, such that a node on the network will install forwarding state for a multicast group sending, when it is on the shortest path between the source and the destination when they report an interest in the community of interest identifier associated with the Multicast group. However, the state re Shortcuts based on the multicast recipient address (DA) and VLAN ID (VID), associated with the multicast transmission. In operation, when the internal node receives the frame, it will search in its database to send information (FIB) based on DA and VID, associated with the frame and forward the frame. As mentioned above, although there will be described an embodiment of the invention in which I-SID is used as an identifier of a community of interests, the invention is not limited to this embodiment, as also may use other types of identifiers common interests.

Traffic Engineering can be used to create a path that is not necessarily the shortest route in the report as a channel controlled the Ethernet. Delivery condition for establishing paths can differentiate traffic by forwarding state that has been set in connection with an embodiment of the shortest-path routing protocol, identifying organizing traffic to a state of delivery, using a different VID. One way to create a path through the traffic network controlled Ethernet protocol channel conditions disclosed in U.S. Patent Application №11 / 732,381, filed April 3, 2007, entitled "Engineered path in link-state protocol controlled Ethernet network", the contents of which are incorporated herein by reference .

When the frame reaches the network element, for example, if the item is a custom Network I must transmit frame Elements Network I, a frame is received by the network element F provider. Network element F will determine whether it is known which of the nodes in the network provider is able to reach the MAC address of the destination node J (S-MAC). If F is already aware that the service provider network element E can reach user network element J, the network element F will add the title of MAC, to perform encapsulation of Mac-in-Mac client frame. Exterior header will include the MAC address of the target network element E, to provide a transmission frame over the network.

Similarly, if the frame is a multicast frame, the network provider determines element F multicast DA provider that should be used to transmit a frame on the network provider. Input network element frame F then passes through a network provider on the shortest path, or alternatively, using any available route through the network traffic engineering. The input node performs resolution → C-MAC and B-MAC frame encapsulates the client using the new MAC header so that the encapsulated frame is addressed, the addressing space using B-MAC. Encapsulation MAC-in-MAC is known in the art and therefore a detailed description of the processes included in this type of encapsulation is omitted.

If the input node F does not know which site providers can reach a client node J, the input node will simply use the multicast tree related to the common interests (or I-SID), to avalanche send the packet to all the other edge backbone bridges (BEB) in the community interests. Any subsequent message by J F allows to know which DA provider to be used for the outer header MAC. Furthermore, the distributed hash table may be used as a repository correlations C-MAC to B-MAC so that the ingress node can transmit a request to one or more nodes using a distributed hash table instead of a broadcast request for the addresses. One method of using a distributed hash table is disclosed in U.S. Patent Application №11 / 714,508, registered March 6, 2007, entitled "Distributed storing routing information in link-state protocol controlled Ethernet network", which is incorporated herein by reference.

When increasing the size of the network and the network includes a large number of nodes, it may be desirable to divide the network into two or more small segments. This will split control plane and associated network database at the two or more objects so that detailed routing updates can be contained within a small segment of the network, and changes within one segment does not affect the adjacent segments. This is economically advantageous because the number of broadcast alert channel state can be reduced, the size of the database as of a channel can be reduced, and the overall speed of convergence of the network to change in topography can be increased. However, the division of the network into two or more segments has the disadvantage of the need for communication, which covers all segments of the network.

Once the network reaches a certain size, divided into segments, it may be enough to solve the problem of scaling, and it may be necessary to reduce the value of the state in the core network (network of L2), to continue to build the network. This can be achieved, and hierarchically recursively transforming network (MACinMACinMAC) in the control plane and data plane, and in the preferred embodiment, again using the MAC assay according to 802.1ah, to establish a link between the level of the B-MAC and the MAC layer in the recursive .

Looping in the Ethernet transmission path can be disastrous, especially if the transmission is a multicast route, as this could lead to unlimited replication packets. It is therefore advantageous to limit the hierarchical relationship segments compared with the resolution cell connect segments to facilitate the solution to the problem of cycling. Routing systems have a concept of a concept level 1 / level 2 (L1 / L2) in IS-IS, which segments L1 are connected to only one segment L2.

Figure 2 shows one example of a communication network 11 in which a plurality of channels controlled Ethernet network the bridges 20 are connected through the boundary segments (ABB) 30. Specifically, Figure 2 network 11 includes a first set of protocol controlled Ethernet network channel state L1A, L1S and L1V. The first set of protocol controlled Ethernet network channel status can be, for example, the central area networks, although the invention is not limited to this particular example. Network L1A, L1B and L1C are connected to a managed network Ethernet L2 protocol state of another channel. Segment L2 network may be, for example, the core network provider configured for a network connection L1. The invention is not limited to the particular example shown in Figure 2 as the network of Figure 2 is shown merely as one illustration of an environment in which the invention may be implemented. In IS-IS formal interface between L1 and L2 is defined as a set in the connection outside the assembly. In this document, is defined as ABB bridge having an interface, at least one channel L1 and the at least one channel L2.

Clients connected to backbone network through edge bridge (BEB) 32. Within the network connections are established through the root backbone bridges (BCB) 34. Assume that, as shown in Figure 2, the client 40, which is connected to the network via L1 BEB-A can communicate with the client 42, which is connected to the network via the L1-B-BEB B, and can communicate with the client 44, which is connected to the network via the L1-B-C BEB. To ensure such a connection, it will be necessary to establish a route between A and B through the network segments L1-A, L2 and L1-B, and also establish a route between A and C through a network segments L1-A, L2, and L1-B.

In accordance with one embodiment of the present invention, the network 10 includes a single segment L2. Although ABB can serve multiple disjoint segments L1, each port on ABB is designed for only one segment. However, if there is a direct physical link between ABB, serving the same segment, and it is desirable to use a traffic channel for L1 and L2, there are two logical port with multiplexing scheme. Each segment L1 is a dead-end segment, ie there are no ABB L1 between the two segments, which are also not connected with the segments L2. To facilitate the calculation of nezatsiklennyh routes traffic between segments L1 must not use channel L2. Nodes L2 L1 channels not used as a transit point to other nodes L2, even if L2 is divided into segments otherwise; However, the node L2 may use the path of the main transit provider (PBT) through a segment L1: in this case, the traffic segment L1 L2 intersects with an additional level of encapsulation of Ethernet and most remote of the same variety VID traffic L1. When traffic entering from different segments, the signals always come in different physical or logical ports, and ABB can easily maintain and use a variety of information about the shipment base (FIB), one for each segment served. Thus, when a packet arrives at port L2, ABB consults FIB L2, to determine where it should be sent.

There are a number of limitations that should be considered when deciding on a plurality of segments. For example, as opposed to telephone numbers, MAC Ethernet address can not be summed, thereby reducing the group (such as code segments 613, for example, code segments defining all telephone numbers in Ottawa, Canada). In addition, segments of the network must provide symmetrical transmission so that traffic can travel through the network on the same path in both directions.

In the example of Figure 2 segments L1-A, L2 and L1-B segments are controlled Ethernet network protocol state of the whole channel, each performing its own example of a routing protocol for channel state. Thus, the routing information is mainly contained in different local area networks, and is only limited by the total number of routing information exchange between the segments. However, as described in more detail below, ABB may allow identifiers of community of interests, such as the I-SID and certain related information BEB passed between the segments thus routes associated with BEB with the I-SID, may be set to a greater degree than one segment. Specifically, as the interest in I-SID can be passed through the network edge, the segments of the route can be established for the I-SID in each local area networks, which together form the path between the segments. Since the passage of I-SID can be done without the intervention of the network management system, the routes between the segments may be set automatically control plane plurality of network segments.

According to one embodiment of the invention, ABB on the border between two networks distributed in each network segment with the ability to reach the other network. Thus, for example, in Figure 2 and ABB, ABBA-d, each located on the boundary between network segments L1-A and L2. Accordingly, each of the ABB will report on the possibility of achieving a network segment within the L2 network segment Ll-A, and report on the possibility of achieving a network segment Ll-A within the network segment L2. According to one embodiment of the invention, ABB may announce network segment L2 as "pseudoknot" (also known as virtual BEB) in segment L1 so that VSV can automatically determine which ABB must handle traffic for a given set closest BEB by establishing state Shipping for the shortest paths between the closest and virtual BEB BEB, announces ABB. Thus, the network L1 itself may choose ABB, to submit a set of BEB in the adjacent network segment L2. If all ABB announced L2 network segment as one and the same virtual BEB, the shortest path from the BEB network segment L1 will be automatically installed by ABB, which is closest to the virtual BEB and, therefore, on the set of BEB that are nearest to a particular ABB .

Bridges ABB, serving some L1, self-selected to represent certain BEB in L2 each ABB, determining what BEB to L1 closer to him than any other ABB. Thus, in Figure 2, ABB and is closest to the A-BEB. Thus, routes A, which should come out of the network segment L1-A, will be established through the root backbone bridges (ESE), such as ENE-A ', to go through the ABB-A. Similarly, routes from BEB-D are set via the ABB-d. There are many ways to do this, but the simplest (and does not require any special rules for BEB and BCB in L1) is that L2 to L1 is represented as a single node ABB pseudoknot, ie Virtual BEB connected to ABB equivalent channels. As mentioned above, the traffic between the segments L1 should not use channels L2: value "channels" to pseudoknot representing L2, must be high enough that the shortest path between any pair of nodes in the segment L1 would not include virtual BEB. In one embodiment, this is achieved by setting the metric value for the distance "channels" that is greater than half the diameter of segment L1. The diameter of the segments L1 is the largest distance between any two nodes in the segment L1.

There are certain rules in order, as ABB information passed between the segments. ABB, the closest to the BEB in L1, circulate addresses I-SID and MAC BEB related to this segment in the L2, without a priori knowledge of what I-SID is of interest for a variety of segments. ABB only miss information BEB and I-SID, gathered from other segments L1, from L2 to L1, where one or more BEB to L1 has indicated its interest in the I-SID. Therefore, the nodes in L2 will be a comprehensive display of I-SID and BEB in the control plane. Nodes on L1 will have a display only interest BEB and I-SID restricted area and those who are really lots of segments.

From the foregoing it can be seen that in the respective compound L2 data plane will be created on the community of interest identifier, i.e. on I-SID, between ABB, opting for representation related BEB in L1. Also, L1, ABB, representing another BEB L1s, create corresponding compounds include local BEB that are part of the same community of interest, as identified community of interest identifier.

BEB network segment L1 advertise interest in the community of interest identifier such as the I-SID, by notifying the channel condition, or using other communications network segment L1. This example assumes that the ID is a community of interest I-SID. Also, other identifiers can be used by common interests.

ABB receive reports indicating that one or more network segment BEB interested L1 I-SID. Miss ABB I-SID, learned on the network segment L1, which have been advertised BEB those which are closest to a network segment L2. When advertising I-SID, a designated set of BEB, only the closest to the BEB network L2 may obtain information about what needs to be ABB, used to transmit traffic on the route to the BEB. ABB will also listen to the air in search of I-SID, advertised by other ABB network segment L2. Where more than one ABB, respectively connected to another network segment and L2, advertised interest in the same I-SID, where the I-SID of interest a plurality of segments. Finding I-SID in more than one L1 guarantee that the network will not establish L2 forwarding state between ABB in the same network L1. If a single L1 has more than one ABB, the internal topology of L1 can initiate several ABB to promote I-SID to L2, but it should be ignored in the L2, L1 and if the other does not endorse the I-SID. In this case, ABB, which I-SID advertised network L2, also be given I-SID back connected thereto network segment L1 so that the compound segment L1 may be installed on the network segment L1 from BEB to ABB. If the set of ABB sends I-SID back to the L1, the connection between themselves ABB for this I-SID in the L1 is not installed. In the example of Figure 2, the connections between ABB and ABB-b-c to L1-B is not set. In the example shown in Figure 2, it is assumed that BEB-A advertised interest in I-SID-x in segment L1-A, and that BEB-B and BEB-C shown interest in I-SID-x in segment L1 -B. ABB, ABB-b, ABB, together with advertised interest in all I-SID in L2, BEB that advertise that they represent. Thus, in this example, ABB, and advertise the MAC-BEB-A / I-SID-x, ABB-b advertise MAC-BEB-B / I-SID-x and ABB C advertise MAC-BEB-C / I -SID-x. ABB-a, ABB-b and ABB-c collectively decide that I-SID-x of interest for the plurality of segments to give warning to other ABB on L2, and determining that the I-SID-x advertised as from L1-A and from L1-B. Accordingly, ABB advertise MAC-BEB-B / I-SID-x, and MAC-BEB-C / I-SID-x in the network segment L1-A, and ABB-b and ABB-to advertise MAC-BEB-A / I-SID-x in the network segment L1-B. As explained below, these advertising messages in the segment L1 made as if they were obtained from a pseudoknot L2, ABB common in the segment L1. Initiating each ABB advertising all the I-SID, received from its adjacent network segment L1 to the network segment L2, ABB on L2 can determine which I-SID must be extended between network segments L1 to selectively provide information about MAK / I-SID only these routes in their network segment L1.

ABB will miss all the interest I-SID for their set BEB to L1 from L1 to L2, ABB in L2 advertise all the I-SID L1 between them, but advertised I-SID from L2 to L1 only when the same I-SID is and this has been circulated L1. Thus, the end result is that within L1 BEB all interested in a particular I-SID, the compounds will be set by the routing system. Only if the I-SID exists in another segment, ABB will promote interest in the I-SID in the L1 (when the connection will be established through segment ABB). The segment L2 ENE establish connections between the various segments L1 ABB, which showed interest in the same I-SID, so that the compound could be mounted within the L2 network. If a network has more than one L1 ABB, reported in I-SID L2, a compound of this I-SID between ABB are installed in L2.

ABB promotes all I-SID and disseminates relevant information on BEB on the L1 L2. Information about the I-SID, which comes from the network segment L1 L2 network segment, will be in the form of MAC addresses ABB, I-SID and BEB MAC addresses associated with I-SID. When ABB gets publicity I-SID from another ABB in L2 and also receives advertising from local L1 with an interest in the same I-SID, he touts the I-SID and disseminates information on BEB derived from L2 to L1.

I-SID will be advertised on the network L2. Like a single segment, NE segment L2 will install forwarding state to provide the shortest path between ABB, connected with the various segments L1, which declare their interest to the same I-SID. For example, assume that the ABB, a, b and ABB-ABB-c together announce interest in I-SID = x. ENE 1 recognizes, on what is the shortest path between two ABB to promote interest in the overall I-SID, and sets the forwarding state to allow the transfer of frames from ABB, and ABB to-b and vice versa. Similarly, VSV 2 install forwarding state to allow transmission of frames to and ABB-ABB-c and vice versa.

ABB and ABB-b-miss with I-SID L2 network segment of the network segment L1-B, as if it had been received from BEB Virtual situated behind ABB b & c. ENE within the network L1-B then set the forwarding state if they are on the shortest routes between BEB, which advertised interest in the I-SID and virtual BEB (where ABB also announced its interest in I-SID). ABB create advertising messages that seem to come from virtual BEB, when there are two or more ABB, skipping I-SID from the segment L2 network segment L1. In one embodiment, ABB configured to always be sent to the segment L1 advertising messages that are presented as advertising messages from virtual BEB. In another embodiment, ABB configured for use only virtual BEB pass I-SID in a certain L1, where there are a plurality of ABB, connected to that segment L1. There are other possible embodiments in which ABB decides that he is the one who has to send advertising messages I-SID to the segment L1 (such as ABB in Figure 2), and thus promotes the interest in I-SID from itself.

At the same time, we note that, by providing ABB to choose what BEB present in conjunction with routes that go from the L1-B, parallel paths have been established between ABB and BEB-b-B, and ABB-to-c and BEB. However, the use of multiple ABB, to achieve a variety of BEB, will not cause conflicts in transit, which actually takes place in the spanning tree to the virtual BEB, which is L2, which naturally leads to routes between BEB and ABB, is only installed by BEB to the nearest ABB. Where there are equivalent ways of data between BEB and two or more of ABB, the routing system will use the normal mechanism of the destruction of the links between segments, to determine which ABB must submit BEB in an adjacent segment.

I-SID normally has the ability to multicast communication. Specifically, multicast mode can be set to the network, initiating BEB interested in Multicast advertisements of interest to I-SID, associated with Multicast. Then multicast forwarding state is established is described in more detail in U.S. Patent Application №11 / 702,263, mentioned above. Instead of I-SID can be used by other identifiers community of interest, and the invention is not limited to the embodiment that uses the I-SID as the community of interest identifier. As mentioned above, it is desirable to place information SSUs between the segments, but by using a mechanism which minimizes changes in a single segment, affecting other segments. One way to achieve this is in connection with BEB ABB with peer segment so that as if they are arranged side by side, and that no knowledge of the topology of the peer segment (in the form of actual metrics) would not be used together with other segments. It has been simplified to a simple connection with BEB closest ABB. The result was that a multicast tree for a given I-SID, located in ABB, identical for all BEB that are behind ABB. This means that the scalability can be improved by using a common target multicast address to the multicast stream for the I-SID, which transmit ABB transit.

Since the ABB may be introduced into a plurality of multicast messages in set up the nearest L2 BEB, it may sum multicast message routing information flowing in the adjacent segment L2. For example, ABB can summarize information on multicast routing mMAS (BEB, I-SID), spreading instead mMAC (ABB, I-SID). Specifically, ABB can replace its own DA DA BEB for the I-SID. It can also be repeated at the boundary between the L1 and L2.

To illustrate:

- Moving from L1 to L2, L2 multicast tree in the roots in the ABB characteristic of all BEB in L1, which was closest to the ABB.

- Moving from L2 to the specific L1, multicast tree in the L1 rooted in this characteristic of all ABB ABB in L2, whose root is common to any other L1. Note that this tree will expand to L1 only to BEB, which are the closest to this ABB.

- No ABB segments on this boundary will never be a leaf on the multicast tree rooted at another ABB on the boundary segments in the L1, and a L2.

From the standpoint of design paths in the network L1-ENE A-A 'detects that it is on the shortest path from A to BEB-L2 (through ABB-a). NE-A 'will also determine that the BEB-A and ABB, and together have the I-SID. Thus, NE-A 'form, and set the address of the multicast group for BEB-A / I-SID = x. He also set individual addresses for remote BEB, who expressed interest in the I-SID-X (BEB-B and BEB-C in this example), set a unique address for the local BEB-A, and form and establish a multicast address for the ABB-a / I-SID = x.

The L2 network NE 1 determines that it is on the shortest path between ABB-ABB-a and b in L2 and that both of them together are I-SID (I-SID = x). ENE 1 form, and set the multicast address for ABB-a / I-SID = x and ABB-b / I-SID = x and set individual addresses for BEB and BEB-A-B.

This network L1, such as the network L1-B, a plurality of ABB may take an interest in or knowledge of the I-SID. To allow the ENE in the network (network L1-B) to set the forwarding state, ABB advertise the I-SID in conjunction with the virtual BEB, representing a network of L2. This will allow the ASP to establish forwarding state only for routes that pass through between the segments closest to the ABB interested BEB. It also prevents the establishment of various data paths between BEB and more than one ABB, since it will have only one shortest path from the virtual to the BEB BEB representing networking L2, which automatically pass to the BEB. NE can be configured to not install forwarding state between ABB on a network boundary (eg, LI A-L2), even though two or more ABB may be interested in the same I-SID.

In the L2 network ABB may have many BEB behind what is presented in the segment L2. To simplify the calculation of the shortest path on a network segment NE ENE L2 will perform routing computations ABB, rather than BEB, which represent ABB. In this case, each NE in L2 may determine whether it is on the shortest path between two ABB, and if so, whether the ABB I-SID in common use. If there are both of these conditions, ENE can then install forwarding state for the multicast MAC address mMAC (ABB, I-SID = x) and unicast MAC addresses uMAS (BEB) for the BEB involved in the set I-SID, characteristic of these Two ABB.

ABB initiating an independent choice, unicast forwarding can be set through a variety of domains without requiring the establishment of precise ways.

Rather, the system can realize the unicast routing path and allow the shipment status to be set for unicast routes even where unicast path passes through multiple network segments.

Since each segment of the network has its own control plane topology changes frequently can be isolated within a given segment. However, when there is a topology change, and these changes occur to some extent in the ABB closest to BEB, a topology change will also affect the adjacent network. In particular, assume that there was a failure in network L1-A, which changed the shortest path to L2 for BEB-A so that it transits ABB-d. In this case, the routing system in the L1-A set a new shortest path from A to BEB-ABB-d and initiates ABB-d on advertising BEB-A / I-SID = x in L2. This leads to the establishment of a new shortest paths L2 between ABB-c and d-ABB, and between ABB-c and d-ABB. However, a change in the network will not affect other segments L1 so that local failure will not lead to cascading changes in the destinations of all segments of the network. Furthermore, although some network failures L1-A could affect the routing system in L2, many failures in the network L1-A did not influence the selection of ABB for SSUs, as a result, these failures are localized within L1-A, without affecting the routing L2.

One consequence of modeling as a virtual BEB L2 to L1 is that from L2 to L1 can be transferred multiple copies of the multicast packet. However, since the overall behavior is the behavior of spanning tree rooted at BEB in the virtual L2, each BEB to L1 get one and only one copy of the multicast packet.

Although an example was shown and described in detail in connection with a specific network shown in Figure 2, the invention is not limited to this example, since the techniques described herein can be used in many different network settings, to create a path through the plurality of segments. Thus, the invention is not limited to the embodiment in a network that connects network segments, as shown in Figure 2, but rather can be used in connection with any network in which segments controlled Ethernet network protocol state of the two or more channels connected to one or more ABB. Similarly, although the I-SID of the exemplary community of interest identifier that can be used to define communities of interest located between the segments, the invention is not limited thereto, as well as other identifiers may be used common interests.

If the BEB has two or more paths, which are the equivalent of two or more of ABB and disperse, it may be necessary to use different VID, to differentiate the traffic to various ABB. It may also be used and other methods of resolving conflicts between ABB, and the invention is not limited to the embodiment that uses different VID to identify traffic assigned to different ABB.

ABB and ENE in L2 have additional requirements, consisting in the fact that ABB in this segment boundaries can not be multicast tree leaf ABB's on the same border segments. This prevents the loop on the borders of segments.

When traffic is transferred from one network segment to another network segment, such as segment L1 segment L2, traffic may be encapsulated in order to transfer through the second segment has occurred, using the MAC address space segments. For example, when a frame is received node BEB-A from the client 16, which is addressed to a client in the BEB 18-B, the frame will initially have a destination address DA = C-MAC client 18. A BEB determine which BEB can achieve the client MAC address and encapsulate the client frame using the header provider Ethernet. For example, the BEB-A can perform encapsulation MAC-in-MAC so that the frame can be transmitted over the network L1-A, using the MAC address space of the provider, not the client MAC address space. For BEB-A, there are several ways to determine which BEB network client 18 can reach, and the invention is not limited to a particular way in which this information is advertised.

After transmission of the frame across the network segment L1-A it reaches the ABB-A, where it will be transferred to the network segment L2. In this regard, it is assumed that the paths have been set, as described in detail above. According to one embodiment of the invention, ABB may further encapsulate the frame for transmission through the network L2, performing encapsulation "MAC-in-MAC-in-MAC" so that the transfer frame within the L2 network may use the MAC address space L2. Specifically, ABB can determine which other ABB on L2 can transmit the frame to the desired destination (B-MAC address) to determine the MAC address of the target ABB network L2 (A MAC address), and then add the header MAC L2, to then encapsulated for transmission frame L2 in the network. This allows to summarize addresses L1 to L2 from ABB through encapsulation so that VSV L2 network must only establish routes on the basis of MAC address space L2 (MAC-A).

Information C-MAC / B-MAC L1 within the network can be extended in a conventional manner. Similarly, information L1-MAC / L2-MAC (B-MAC-address → A-MAC-address) may be extended normal learning process, such as flooding request L1-MAC / L2-MAC, and waiting for an answer, or by using a distributed hash table.

Figure 3 visually shows what happens in connection with the process of encapsulation. Specifically, the metric L1-A are local to the segment of the L1-A. L2 only filters routes inter-L1 segments of I-SID. This includes kongruetnost uMAC / mMAC in L1, L2 and "MAC-in-MAC-in-MAC". Multicast MAC address from the L1-A via the display I-SID through the tree in the L2. ABB must know that the path to BEB-E passes through ABB's. This association can be studied, avalanche sending a request and waiting for an answer. Flooding in the network segments is limited in the boundary nodes ABB, however, the association requests the B-MAC / A-MAC should not avalanche sent to other network segments. After examining the association of input ABB B MAC / A-MSS, ABB can use this address to encapsulate the frames for transmission to the L2 network. In addition, a self-assigned multicast MAC address L2 may be used where the I-SID advertised able to target more than one network ABB L2.

Figure 4 shows an adaptive, learning and Interlevel function between the levels when the system is recursive routing change. As mentioned above, the L2 network may become too large, and it may be desirable to further modify network recursively to break L2 network to a second network level L1 / L2 / L1, as shown in Figure 6. Figure 4 illustrates a process that allows the frame to encapsulate transmission recursively modified L2 from L1 (sometimes encapsulated level is called "level X" and the encapsulated layer is referred to as "level x + 1"), and also shows the process that allows to encapsulate the frame upon receipt of recursively modified segment L2 network for the transmission of segment L1 network at a given level.

Figure 4 is a functional block diagram of ABB, which performs separation of the network into segments, and hierarchical routing. As such, this node is connected to other peers in every part of the L1 and L2 of the current level, respectively. He is also working on a recursive level X + 1.

FIB levels L1 to be filled through the routing exchange with peer devices in L1 (including L2 transmitted through), so the FIB L1 to level X + 1 (encapsulation layer) is filled through the routing exchange with the peer level X + 1.

As shown in Figure 4, when a frame is received from L1 to level X, ABB check whether the level X destination MAC pass the level X + 1 MAC through the search mapping from X to FIB X + 1, or whether the frame is a broadcast or multicast frame . In these cases, it will be encapsulated using a level X + 1 MAC BEB as a source and the multicast MAC address for the I-SID, used BEB level X + 1 as the destination, and transferred to the FIB level X + 1. If the X target MAC address may be allowed to transmit at the level X + 1 of the MAC address, the packet is encapsulated with the MAC address BEB as a source, and the level of the address X + 1 MAC, received from the mapping X + 1 X to FIB, as the place of destination, and forwarded according to prior FIB X + 1.

When a packet is received from the level X + 1, is connected with the source MAC levels MAC source and inserted into the binding FIB mapping from x to x + 1. De-encapsulates the packet and sent according to the information on "level X" FIB. This study is the result of ligaments X to X + 1 by reusing 802.1ah protocol learning process MAC, which eliminates the need for the exclusive binding intermediate level to level routing system X + 1.

It should be noted that the network can actually use the technique recursion arbitrary number of times. It may also be noted that in the example mentioned may also be divided without recursion so that the mixture recursion unit and at each level of recursion can be used to scale the network. This is shown in Figure 6. For example, as shown in Figure 6, the L2 network may be formed as a level X + 1, L1 / L2 / L1 network having a plurality of networks L1 (X + 1) connected to a network segment. Similarly, the network segment L2 (X + 1) can be formed as a set of L1 / L2 / L1 network segments (X + 2). The process described with reference to Figure 4 may be used to determine the boundary between the level L1 (X) and the level L1 / L2 / L1 (X + 1), the boundary between the level L1 (X + 1) and the level L1 / L2 / L1 (X + 2) or more boundary segment between the network and another recursive level L1 / L2 / L1 (X + n).

From the standpoint of routing UNI interface at the network side ABB X will store level information XI-SID, received through the X level network protocol routing state channel at X FIB. Similarly, at NNI interface (X + 1) on the network side ABB retain information about the level of I-SID X + 1 obtained through the level X + 1, the network routing protocol according to the channel status at FIB X + 1. However, according to one embodiment of the invention, the information about I-SID is passed between the levels and the level X + 1 network to allow the level X + 1, the network is selectively set the routes through the level X + 1 network for I-SID, which are characteristic for different segments X-level networks.

From the viewpoint of the control plane, information about the control plane stack / aggregated through the level X + 1 network to reduce the amount of information that must be processed in the control plane and is set at the forwarding table X + 1. This is advantageous from the viewpoint of scaling, since NE level X + 1 networks require only keeping information transmitted to the level of the MAC address X + 1.

At the level of exchanges at the level of X X + 1 exchange is an exchange of information on the membership of a peer I-SID, allowing other ABB know what I-SID to be missed. Information about the I-SID is then used multicast connections at the network segment x + 1 to examine inter-level communication. Where the level of X uses a network encapsulation Mac-Mac, and the level X + 1 network uses encapsulation Mac-in-Mac-in-Mac, using information about the I-SID enables ABB to study the links Mac-in-Mac / Mac-in- Mac-in-Mac so that ABB could encapsulate traffic based on the "one SID".

If the interconnection network L1 / L2 to be used, alternative ABB, ABB alternative can provide a large metric to the fact that it is not similar to ABB, selected to provide the shortest path to any BEB network segment L1. However, alternative ABB may still miss the information about I-SID to the network segment L1 and, on the other hand, allow the network elements to have information about the ABB and provide faster convergence in case of failure of the primary ABB.

When ABB fails all traffic to I-SID has to be rebuilt. Traffic for the I-SID must be linked to other ABB, which require ENE network L1 to install the new status of the shipment. One method by which this can be accomplished is to set the new condition of delivery, using a different VID, to install two sets of communication paths: a first set of paths for a primary ABB and second set of secondary paths to ABB. Forwarding state may be established after determining the malfunction or, alternatively, may be pre-calculated and established before failure occurs. Installing the backup status of the shipment using another VID, allows you to set different network forwarding state in advance so that the failure ABB traffic can be automatically switched to an alternative path to the tagged traffic using the alternate VID.

Figure 5 shows an example of a network element which may be used to implement an embodiment of the invention. As shown in Figure 5, network element 50 includes a data plane and a control plane 60. The plane of data 50 basically includes a card input / output used to communicate with the channels in the network interface, the data card 54 used to perform the functions of the data obtained on the input-output cards 52, and the structure of the switch 56 is used to switch data between cards IO. The control plane includes a processing unit 62 comprising a control logic adapted to implement the routing process as the channel 64 in the L1 and the routing process as the channel 66. The L2 control logic can also be implemented in other processes.

Data and commands associated with the routing process by L1 of L1 channel 64 and the process of routing L2 channel 66 may be stored as software routing L1 72 and L2 routing software 74 in the memory 70. One or more databases or tables can AVV-supported 30 and initiate ABB to store data associated with the routes that have been installed in the networks L1 and L2. For example, ABB 30 may comprise L1 FIB 80, L2 FIB 82, data base channel state 84, L2 database channel state L1 86, and FIB 88 L1 / L2, containing the identifier community of interest (e.g., I-SID) with an association between the information transmission in the two networks.

ABB may contain other software, processes, and data storage to perform the functions described above and to perform other functions typically implemented in a network element connection.

The functions described above may be implemented as a set of program instructions that are stored in memory, consider a computer and executed in one or more processors on the computer platform associated with the network element. However, to the skilled person it is obvious that all the herein described logic may be implemented using discrete components, integrated circuitry such as an application specific integrated circuit (ASIC), programmable logic used in connection with a programmable logic device such as a programmable in place an array of logic elements ( FPGA) or microprocessor, a state machine, or any other device including any combination thereof. Programmable logic can be built temporarily or permanently in a tangible medium, such as read-only memory chip, a computer memory, disk, or other storage medium. Programmable logic can also be embedded in a computer data signal at the carrier frequency, allowing to transmit programmable logic interface, such as a computer bus or communications network. All such embodiments are within the scope of the present invention. Routing of the channel

Suggesting changes U.S. Application №11 / 537,775, reported October 2, 2006, entitled "Routing of channel provider" of both the source and the multicast group of interest, in the plane of encoded data, which can be placed major techniques to create a tree of the shortest paths as described above, but with slight modifications in the plane of the transfer function data executed in ABB.

In a modification of the address of the multicast group for the group of interest common to all groups BEB that support interest groups and specific source BEB or ABB (multicast source) is encoded in the VLAN. In this case, the seal group of MAC addresses is not possible, but the seal VLAN information between segments is possible. This is useful, because this method is not economical for the VLAN and, hence, the decision on a plurality of segments may substantially increase network scalability. Sealing can be performed well known conversion output VLAN ABB, whereby ABB overwrites VLAN multicast VLAN packet with the value, which has been assigned to ABB as a source of the multicast. The invention is not limited to the particular way in which the value of VLAN assigned to ABB as a multicast source.

In this variation shortest path tree from this BEB will have a unique add-on a VLAN on each tree so that the tree is the shortest path from BEB A will see (for example) all packets from the tagged BEB of VLAN 1, all packets from the tagged BEB in a VLAN, 2 etc. Then will check the route the way back (RFPC) on a VLAN instead of the source address of the MAC. The packages that are required for transit between the segments, will pass through the ABB and in the shortest path tree in an adjacent segment. Packets sent on the shortest path tree from ABB, would simply be re-tagged with the VID, assigned to ABB as a multicast source so that ABB was the "bottleneck" for a set of multicast sources that pass through this segment of transit ABB. Thus, given that there are 4,000 odd available tags VLAN, the end result is that each "segment" or "level" can have 4,000 nodes (the sum of BEB, ENE and ABB), though the seal ABB (and replacement VID ABB) allows each of the segments have their own space VID, and the network can grow in size up to 4,000 units per segment.

In another modification of the address of the multicast group is advertised as described above but is encoded only in the source MAC address, and VLAN is used for all characteristic BEB. In this case, no seal at ABB multicasting impossible, and the packets are transmitted in their original form.

Referring now to Figure 7, there is shown an exemplary communications network 100 PLSB, wherein ABB can be "placed" in a plurality of segments L1. In other words, ABB can serve a lot of disjoint network segments L1. The communication network 100 is shown with PLSB L2 network segment 110, which is geographically expanding to cover more segments L1 116, that would be the case when the segments L1 are the central network and segment L2 is the national backbone network. A single segment L2 110 includes five ABB, i.e. 1 ABB 112a, 112b ABB-2, 3-ABB 112c, ABB 4112-d and 5 ABB 112e (collectively designated as 112 ABB), and three other explosives, i.e. BB 1 114a, 114b, and BB 2 BB 3 114c (collectively designated as BB 114). The communications network 100 also includes three PLSB deadlock segment L1, ie, 11-116a, L1-B 116b and 116c L1-C (collectively designated as L1 segment 116). L1-A 116a is serviced by a single ABB, namely, ABB-2 112b. L1-B 116b served by two ABB, namely, ABB 1 ABB 112a and 112b-2. L1-C 116 is serviced by three ABB, namely ABB-3 112c, 112d ABB 4 and 5 ABB 112e. Note that the L2 segment 110 is illustrated in Figure 7 as a pseudoknot PN L2 in each of the segments 116 L1.

When a segment L1 116 is served by more than one ABB 112, the nodes in the segment L1 116 are divided into non-overlapping "subset" of nodes, one for each ABB, where all the nodes in this section are "closer" to the particular ABB than other ABB segment L1. Typically, in the field of routing protocols "closer" in this specification means that the sum of the metrics of the channel for the shortest path between the node and a particular ABB less than or equal to the shortest path to any other ABB and where there is a bundle which is the sum of the metrics of the channel, is the same between the node and two or more of ABB, then failure mechanism ligament decides that certain ABB is "closer". In the communication system 100 because L1-B 116b served by two ABB, it is divided into two subsets, as shown by dividing lines 118a. A subset L1-B1 120a served by ABB 112a-1, and the subset L1-B2-120b served by ABB 2 112b. Similarly, as the L1-C 116c is serviced by three ABB, it is divided into three subsets lines 118b and 118 c. L1-CI subset 122a served by ABB-3 112c, the subset L1-C2 served by ABB 122b and 112d subset 4, L1-C3 122c served by ABB-5 112e.

It should be noted that the ABB 2-112b serves two disjoint segments L1, namely L1-A 116a and B 116b-L1. Usually, when ABB 112 serves a single segment L1 116, ABB 112 relates to a single FIB L1 for transmitting data packets, as described above. However, ABB, serving multiple segments L1, should have plenty of FIB L1, one FIB L1 packets arriving at all ports.

Channel pseudoknots L2 PN to 110, representing the segment L2, advertised every 112 ABB segment L1 116. The metric value associated with advertising, usually identical for all ABB. However, in this case more than half of the metric of the maximum diameter L1 segment 116, so that L2 PN 110 does not appear on any of shortest paths between the segments. This great performance metric divides segment L1 116 in disjoint subsets of nodes that are "closest" to each ABB 112. "Port MAC" and ISID for a complete set of "external" MAC also advertised with L2 PN. For each subset of the proposed network segments L1 each ABB 112 transmits a message to the level 2 "port MAC" and for a subset of ISID. The root identifier for the subset is included in the package of the channel for that subset. It can be seen that L2 PN 110 is the root node for a general tree, thus using it as the name of the root identifier for all multicast traffic can enter into the segment L1.

Pseudoknot L2 110 performs many functions, including the following three. Firstly, the use of a large metric ensures that traffic within the segments L1 does not pass in transit level 2. Second, the calculation of the "nearest" subset of the nodes of L1 ABB easier for nodes on the fast track to L2PN. Finally, all the external port MAC associated with a single node.

For unicast message traffic entering port L2, sent FIB L2, and traffic to the ports forwarded L1 FIB L1. These FIB different for the case where the destination is in the segment L1, and not in the "near" a subset of ABB. In this case, FIB L1 dictates another packet transmission port L1, though FIB L2 already transmitted packets through the port L2 other ABB.

L2 Multicast tree with source ABB 2 112b shown in Figure 7 by the solid bold lines 124. For multicast packets "next" to L1 subset 116 are needed to guarantee the receipt of a copy of a single package for packages arriving in the ports of L2 in a variety of ABB same network L1 . In the example shown multicast tree L2, wherein the multicast packet generated by ABB 112b-2, will be given three ABB, serving segment L1 C 116c: namely ABB 3-112c, ABB 4112-d and 112e 5 ABB. Tree with roots from ABB 112 is not limited to "nearest" subset, so that an identifier for the root of the tree, which covers only the "near future" subset can not be an alias ABB. However, the root identifier can be an alias L2PN. Advertising L2PN, as described above, of course, creates a "near" subset and multicast trees. Note that the "near future" for each subset of B-VID does not necessarily include the same set of nodes, ie, equivalent multipath routing (ESMR) to L2PN 110 can use a variety of ABB 112.

Thus, in one embodiment of the present invention, when a multicast packet reaches ABB 112 through port layer 2 examines the root identifier of the incoming packet. If the root ID is the root identifier of another "next" a subset of the same segment L1, the packet is dropped. Otherwise, the root ID is replaced by the root identifier and transmitted L2PN wood L1, cover it "near" a subset of ABB.

To ensure symmetry, multicast from the node L1 to L2 switches 110, 112 only when the ABB maintains its "near" subset. This means that the multicast tree should be isomorphic L2 with respect to a multicast tree rooted at L2 ABB 112. However, the root identifier may be an alias ABB, to avoid re-entry multicast traffic segment L1 from another ABB 116 122. Thus, again referring to the exemplary network 100 in Figure 7, ABB-1 112a should replicate packets from ABB-2 112b in L1-B 116b, if they came from the L1-A 116a, but not to do so if they come from the L1-B 116b .

The root identifier in L2 for all trees with roots from ABB 112, serving the same segment L1 116, should not be the same, because the trees from each ABB is a crossing. The root identifier should be different and easy test for identifying the segments in order to ABB 112 could be removed, rather than transmitting packets which come from its own segment. Thus, the multicast packet level 1 if the root identifier of the packet belongs to the "nearest" Subset ABB, this identifier is pumped to a unique "nearest" root identifier subset and is transmitted to all ports of the level 2, which are part of the "nearest" multicast tree subsets package ISID.

A typical combination of hardware and software could be a specialized computer system having one or more processing elements and a computer program stored on a carrier which, after loading and executing the control computer system for implementing the methods described herein. The present invention can also be embedded in a computer program product, which comprises all the features that would allow the embodiment methods described herein, and when loaded in a computer system may perform these methods. The carrier can be any device in the form of volatile or nonvolatile memory.

Computer program or application in the present context means any expression, in any language, code or record a series of commands that initiate a system for processing information to perform a particular function either directly or after the following (a) conversion to another language, code or entry; (b) reproduction in a different material form.

In addition, if the above was not mentioned otherwise, we note that all the drawings should not be scaled. Importantly, this invention may be embodied in other specific forms without departing from the spirit or scope of the invention and, accordingly, the emphasis should be placed on the following claims, rather than the foregoing description of the invention.

It should be understood that various changes and modifications of the embodiments shown in the drawings and described above may be made without departing from the spirit and scope of the present invention. Accordingly, all matter contained in the above description and shown in the accompanying drawings is illustrative and not restrictive. The invention is limited only by the appended claims.