JP2016158036A - Frame transfer device, link metric determination method and path determination method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a frame transfer device capable of delivering a link metric so that a path, suppressing the frequency of the frame requiring low delay transfer to compete, is determined.SOLUTION: In a frame transfer device, an adjacent node information collection unit 29 collects adjacent node information from each adjacent node, an adjacent node information advertisement unit 24 advertises the adjacent node information collected by the adjacent node information collection unit 29 in mesh network. Determination means, i.e., an Express traffic path determination unit 33 and a link metric determination unit 32, determines a first link metric for use when a frame requiring low delay transfer is transmitted or received, and a second link metric for use when a frame not requiring low delay transfer is transmitted or received, on the basis of the adjacent node information collected by the adjacent node information collection unit 29, and the adjacent node information advertised in the mesh network.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a frame transfer apparatus, a link metric determination method, and a path determination method for transferring a frame in a mesh network.

  In recent years, application of Ethernet (registered trademark) to a network such as a car, a factory, and a train has been studied. However, when Ethernet (registered trademark) is applied to these networks, there are problems such as suppression of frame transfer delay and ensuring network reliability. In order to solve these problems, attention has been paid to Interspersing Express Traffic (IET) and Shortest Path Bridging (SPB). IET is a technology that realizes low-latency transfer of frames, and standardization is underway in IEEE 802.3br. When an Express frame that requires low-delay transfer and a Normal frame that does not require low-delay transfer compete at the switch, the IET interrupts the transfer of the Normal frame, interrupts and transfers the Express frame, and the Express frame This is a technique for reducing the transfer delay of the Express frame by restarting the transfer of the Normal frame after the transfer of the frame is completed. On the other hand, SPB is a technique related to a method for determining a path in a mesh network, and standardization is underway in IEEE 802.1aq. SPB is a technology that enables high-speed path switching, redundant path configuration, and traffic distribution when a failure occurs by configuring ECMP (Equal Cost Multi Path).

  Further, Non-Patent Document 1 describes a link metric determination method as a technique for realizing traffic distribution in a mesh network. In the technique described in Non-Patent Document 1, a link metric that realizes load distribution of traffic is determined by solving a linear programming problem so that the maximum value of the utilization rate of all links is minimized.

“Internet Traffic Engineering without Full Mesh Overlaying”, Yufei Wang, Zheng Wang, Leah Zhang, IEEE INFOCOM 2001

  When the IET described above is applied, the Express frame can be transferred with low delay. However, IET can transfer an Express frame with low delay when a Normal frame and an Express frame compete with each other, but a delay occurs when different types of Express frames compete with each other. Therefore, it is necessary to determine a path so that the frequency of occurrence of Express frame contention is low. Here, when SPB is applied, it is possible to determine the path so that traffic is distributed, that is, the frequency of occurrence of frame contention at each node is reduced. However, in SPB, a path is determined without considering the type of frame transmitted / received in the mesh network, so it is difficult to use in combination with IET. That is, in SPB, even when a Normal frame and an Express frame are mixed, these frames are handled as the same type, and therefore, it is not always possible to determine a path so that the frequency of contention between Express frames is low. Also, the technique described in Non-Patent Document 1 does not determine the link metric on the assumption that a node that transmits and receives an Express frame and a node that transmits and receives a Normal frame coexist.

  The present invention has been made in view of the above, and provides an SPB for a network having a configuration in which nodes that transmit and receive frames that require low-delay transfer and nodes that transmit and receive frames that do not require low-delay transfer are mixed. It is an object of the present invention to obtain a frame transfer apparatus capable of deriving a link metric for determining a path in which the frequency of occurrence of contention between frames requiring low-delay transfer is suppressed even when applying.

  In order to solve the above-described problems and achieve the object, the present invention is a frame transfer apparatus constituting a mesh network, and includes an adjacent node information collecting unit, an adjacent node information advertising unit, and a determining unit. Neighboring node information collection means collects neighboring node information including attribute information indicating whether or not a node that transmits / receives a frame requiring low-delay transfer and communication partner node information indicating a communication partner node from each of the adjacent nodes. Then, the adjacent node information advertising means advertises the adjacent node information collected by the adjacent node information collecting means in the mesh network. The determining unit determines whether the link constituting the mesh network is low based on the adjacent node information collected by the adjacent node information collecting unit and the adjacent node information advertised into the mesh network by another frame transfer apparatus. A first link metric used when determining a path for transmitting and receiving a frame that requires delay transfer, and a second link used when determining a path for transmitting and receiving a frame that does not require low delay transfer Metrics are determined.

  According to the present invention, even when a node that transmits and receives a frame that requires low-delay transfer and a node that transmits and receives a frame that does not require low-delay transfer coexist, transfer processing of a frame that requires low-delay transfer competes As a result, the link metric can be derived so that a path with a low frequency is determined.

The figure which shows an example of the mesh network comprised by the frame transfer apparatus concerning Embodiment 1 of this invention The figure which shows the structural example of the switch which is a frame transfer apparatus of Embodiment 1. The figure which shows an example of the hardware constitutions which realize a switch Flow chart showing an example of operation of the frame processing unit The figure which shows the structural example of a Hello frame. A flowchart showing an operation example of the adjacent node information collection unit A flowchart showing an operation example of the adjacent node information advertising unit A flowchart showing an operation example of the node information collection unit The flowchart which shows the operation example of an Express traffic path determination part Flowchart showing an operation example of the link metric determination unit A flowchart showing an operation example of the SPB processing unit Flow chart showing an example of operation of the capsuler A flowchart showing an operation example of the frame transfer unit The flowchart which shows the operation example of a Hello production | generation part. The figure which shows the procedure in which the switch of Embodiment 1 determines the path | route between the nodes connected to the mesh network. The figure which shows the structural example of the frame which the switch advertises in the mesh network The figure which shows the structural example of the switch which is a frame transfer apparatus of Embodiment 2. Flow chart showing an example of the operation of the linear programming problem solving unit Flowchart showing an operation example of the link metric determination unit The figure which shows the procedure in which the switch of Embodiment 2 determines the path | route between the nodes connected to the mesh network.

  Hereinafter, a frame transfer device, a link metric determination method, and a path determination method according to an embodiment of the present invention will be described in detail with reference to the drawings. Note that the present invention is not limited to the embodiments.

Embodiment 1 FIG.
FIG. 1 is a diagram illustrating an example of a mesh network configured by the frame transfer apparatus according to the first embodiment of the present invention.

The mesh network of the present embodiment is configured by switches 2 ₁ to 2 ₇ that are frame transfer apparatuses. Switch 2 _1, switch 2 _2, 2 ₃ and conditions link is established between the 2 _4, i.e., it can communicate with the switch 2 _2, 2 ₃ and 2 _4. The switch 2 ₂ is in a state where a link is established between the switches 2 ₁ and 2 ₅ . The switch 2 ₃ is in a state where a link is established between the switches 2 ₁ and 2 ₆ . The switch 2 ₄ is in a state where a link is established between the switches 2 ₁ , 2 ₅ and 2 ₆ . The switch 2 ₅ is in a state where a link is established between the switches 2 ₂ , 2 ₄ and 2 ₇ . The switch 2 ₆ is in a state where a link is established between the switches 2 ₃ , 2 ₄ and 2 ₇ . The switch 2 ₇ is in a state where a link is established between the switches 2 ₅ and 2 ₆ .

The switch 2 _1, the best effort node 1 ₁ is a node for transmitting and receiving best effort traffic, and the Express node 3 ₁ and 4 ₁ is a node for transmitting and receiving Express traffic are connected. Here, the best effort traffic is data transmitted and received in a normal frame that is a frame that does not require low-delay transfer. The Express traffic is data transmitted and received in an Express frame that is a frame that requires low-delay transfer. In FIG. 1, the switch is represented by “◯”, the best effort node is represented by “Δ”, and the Express node is represented by “□”. In the following description, when only “frame” is described, both a normal frame and an express frame are shown. In addition, when “node” is simply described, it indicates both the best effort node and the express node.

A best effort node 1 ₂ is connected to the switch 2 ₂ . The switch 2 _3, best effort node 1 ₄ are connected. The switch 2 _7, a best effort node 1 _3, and Express node 3 ₂ and 4 ₂ are connected. It is assumed that the Express node 3 ₁ and the Express node 3 ₂ transmit and receive Express traffic to each other, and the Express node 4 ₁ and the Express node 4 ₂ transmit and receive Express traffic to and from each other.

Incidentally, there is a case where the best effort nodes ₁ 1, Express node 3 _1, 4 ₁ is connected to the switch 2 _1, referred to as switch 2 ₁ of the adjacent nodes. Similarly, there is a case where the best effort node 1 ₂ is referred to as switch 2 ₂ adjacent nodes. Sometimes the best effort node 1 ₄ is referred to as switch 2 ₃ adjacent nodes. The best effort node 1 ₃ and the express nodes 3 ₂ and 4 ₂ may be referred to as adjacent nodes of the switch 2 ₇ .

FIG. 2 is a diagram illustrating a configuration example of the switches 2 ₁ to 2 ₇ which are the frame transfer apparatuses according to the first embodiment. The configuration of the switches 2 _{1 to} 2 ₇ is the same. Hereinafter, when a configuration, operation, and the like common to the switches 2 _{1 to} 2 ₇ are described, the switches 2 ₁ to 2 ₇ may be collectively expressed as the switch 2.

In the switch 2, the receiving unit 21 receives a frame from another switch 2 or an adjacent node. The transmission unit 22 transmits the frame to another switch 2 or an adjacent node. The Hello generation unit 23 generates a Hello frame. The adjacent node information advertising unit 24 generates a frame including information on adjacent nodes to be advertised to other switches 2. The frame transfer unit 25 transfers a frame that is not addressed to the own switch among frames received from another switch 2 or an adjacent node. The transfer database 26 manages frame transfer destination information. The SPB processing unit 27 determines a path between nodes transmitting and receiving a frame by SPB (Shortest Path Bridging) standardized by IEEE 802.1aq. The frame processing unit 28 determines the output destination by confirming the contents of the frame received from another switch 2 or an adjacent node. When the adjacent node information collection unit 29 receives a frame including information on an adjacent node, the adjacent node information collection unit 29 extracts information on the adjacent node from the received frame. When the node information collection unit 30 receives a frame including information on adjacent nodes of another switch 2, the node information collection unit 30 extracts information on adjacent nodes of the other switch 2 from the received frame. The network topology information management unit 31 manages the topology information of the mesh network formed by the switches 2 ₁ to 2 ₇ . The link metric determination unit 32 determines a link metric for all links in the mesh network. The Express traffic path determination unit 33 temporarily determines an Express traffic transfer path, that is, an Express traffic path that is a path between Express nodes. The capsuler 34 performs frame encapsulation or decapsulation when transferring a frame received from another switch 2 or an adjacent node.

  FIG. 3 is a diagram illustrating an example of a hardware configuration that implements the switch 2. The switch 2 is realized by a transmitter 101, a receiver 102, a processor 103 such as a CPU (Central Processing Unit), and a memory 104 including a RAM (Random Access Memory), a ROM (Read Only Memory), and the like. Is possible. The transmitter 101, the receiver 102, the processor 103, and the memory 104 are connected to the bus 100, and can exchange data, control information, and the like via the bus 100.

  The transmission unit 22 of the switch 2 illustrated in FIG. The receiving unit 21 of the switch 2 is realized by the receiver 102. Hello generation unit 23, adjacent node information advertising unit 24 frame transfer unit 25, SPB processing unit 27, frame processing unit 28, adjacent node information collection unit 29, node information collection unit 30, network topology information management unit 31, link metric determination unit 32, the Express traffic path determination unit 33 and the capsuling unit 34 are realized by the processor 103 executing the corresponding programs. The transfer database 26 is realized by the memory 104 storing necessary information. The network topology information management unit 31 is realized by the memory 104. Note that a plurality of processors and a plurality of memories may be linked to realize each unit of the switch 2.

  Next, the operation of each part of the switch 2 will be described.

  The receiving unit 21 receives a frame from another switch 2 or an adjacent node constituting the mesh network, and outputs the received frame to the frame processing unit 28.

  The frame processing unit 28 operates according to the flowchart shown in FIG. Specifically, the frame processing unit 28 continuously monitors whether or not a frame has been received from the receiving unit 21 (step S11: No). When the frame processing unit 28 receives a frame (step S11: Yes), that is, when a frame is received from the receiving unit 21, the frame processing unit 28 checks the content of the frame and outputs the frame to an output destination corresponding to the type of the frame. Specifically, the frame processing unit 28 checks whether or not the frame received from the receiving unit 21 is a Hello frame sent from an adjacent node (Step S12). If the frame is a Hello frame (Step S12: Yes), the adjacent node The frame is output to the information collecting unit 29 (step S14). Neighboring node information is stored in the Hello frame. Here, the Hello frame will be described. FIG. 5 is a diagram illustrating a configuration example of a Hello frame. As shown in FIG. 5, the Hello frame is configured to include a header and adjacent node information, such as identification information, attribute information, a communication partner node, and bandwidth demand. The header is composed of Hello frame destination information, transmission source information, frame type information, and the like. The destination information is the MAC address of the device that receives the frame. The transmission source information is the MAC address of the device that transmitted the frame. The frame type information indicates the type of the received frame. In the case of a Hello frame, a value that uniquely indicates the Hello frame is set. When the frame is not a Hello frame, the value of the frame type information is a different value. The frame processing unit 28 can determine whether or not the received frame is a Hello frame by checking the frame type information. The identification information constituting the adjacent node information is identification information of the adjacent node, for example, a MAC address. The attribute information constituting the adjacent node information is information indicating whether the adjacent node corresponds to an Express node or a best effort node. The communication partner node is information (hereinafter referred to as communication partner node information) indicating a partner node with which an adjacent node transmits and receives a frame. The bandwidth demand is information included when the adjacent node is the best effort node, that is, when the attribute information indicates the best effort node. The bandwidth demand is information on the bandwidth required by the adjacent node (hereinafter, bandwidth demand for best effort traffic). Other information may be stored in the Hello frame. The identification information may be other than the MAC address. The identification information only needs to uniquely indicate the adjacent node. The Hello frame having the configuration shown in FIG. 5 is periodically transmitted from each node connected to the mesh network.

  Returning to the description of the operation of the frame processing unit 28, the frame processing unit 28 checks whether the frame received from the receiving unit 21 is not a Hello frame (step S12: No), or is a frame advertised by another switch. (Step S13). If the frame received from the receiving unit 21 is not a frame advertised by another switch (step S13: No), the frame processing unit 28 outputs the frame received from the receiving unit 21 to the encapsulation unit 34. If the frame received from the receiving unit 21 is a frame advertised by another switch (step S13: Yes), the frame processing unit 28 checks whether adjacent node information is stored (step S15). When adjacent frame information is stored in the frame received from the reception unit 21 (step S15: Yes), the frame processing unit 28 outputs the frame received from the reception unit 21 to the encapsulation unit 34 and the node information collection unit 30. (Step S17). When the adjacent node information is not stored in the frame received from the reception unit 21 (step S15: No), the frame processing unit 28 outputs the frame received from the reception unit 21 to the encapsulation unit 34 (step S16).

The adjacent node information stored in the frame advertised by the other switch 2 is the information collected from the adjacent node by using the Hello frame having the configuration shown in FIG. 5 by the other switch 2 that advertised the frame. It is. For example the frame processing unit 28 receives a frame switch 2 ₁ of Figure 1 is ads, if the frame to the adjacent node information received is stored, the adjacent node information stored in the received frame, the switch 2 ₁ Is adjacent node information collected from the best effort node 1 ₁ and the express nodes 3 ₁ and 4 ₁ . That is, the adjacent node information stored in this frame is the adjacent node information transmitted by the best effort node 1 ₁ and the Express nodes 3 ₁ and 4 ₁ in the Hello frame.

  The adjacent node information collecting unit 29 as the adjacent node information collecting means operates according to the flowchart shown in FIG. Specifically, the adjacent node information collection unit 29 continuously monitors whether or not a Hello frame is received from the frame processing unit 28 (step S21). When receiving the Hello frame (Step S21: Yes), that is, when receiving the Hello frame from the frame processing unit 28, the adjacent node information collecting unit 29 collects the adjacent node information stored in the Hello frame (Step S22). The extracted adjacent node information is output to the adjacent node information advertisement unit 24 and the network topology information management unit 31 (step S23).

  The adjacent node information advertising unit 24, which is an adjacent node information advertising means, operates according to the flowchart shown in FIG. Specifically, the adjacent node information advertising unit 24 continuously monitors whether adjacent node information has been received from the adjacent node information collecting unit 29 (step S31). When the adjacent node information is received (step S31: Yes), that is, when the adjacent node information is received from the adjacent node information collecting unit 29, the adjacent node information advertising unit 24 receives the received adjacent node information and the identification information of the own switch. The stored frame is generated (step S32), and the generated frame is advertised in the mesh network via the transmission unit 22 (step S33). The identification information of the own switch is a MAC address, but is not limited to this. The identification information of the own switch may be any identification information as long as the advertisement source of the frame can be recognized on the receiving side of the advertised frame. When a MAC address is used as identification information of the own switch, since the MAC address is stored as the frame transmission source information in the header of the frame to be advertised, the MAC address is not separately stored in the frame as the identification information of the own device. May be. That is, the frame transmission source information in the header and the identification information of the own device may be used together.

The node information collection unit 30 operates according to the flowchart shown in FIG. Specifically, the node information collection unit 30 continuously monitors whether a frame advertised from another switch is received from the frame processing unit 28 (step S41: No). When receiving the frame advertised from another switch (step S41: Yes), that is, when receiving the frame from the frame processing unit 28, the node information collection unit 30 receives the adjacent node information stored in the received frame and The identification information of the frame transmission source switch 2 is extracted (step S42), and the extracted adjacent node information and the identification information of the frame transmission source switch 2 are output to the network topology information management unit 31 (step S43). Here, the adjacent node information stored in the frame received by the node information collecting unit 30 from the frame processing unit 28 is adjacent node information of a node adjacent to the switch 2 that is the transmission source of the received frame. For example, if a sender of frame node information collecting unit 30 is received by the switch 2 ₁ in FIG. 1, the adjacent nodes are stored in the received frame information, the switch 2 ₁ best effort node 1 ₁ is at Hello frame transmitted neighbor node information to, the neighboring node information Express node 3 ₁ adjacent node information and the Express node 4 ₁ transmitted by Hello frames to the switch 2 ₁ is transmitted in Hello frames to the switch 2 _1. Incidentally, the node information collecting unit 30, when the sender receives a frame switch 2 _1, best effort node 1 ₁ of the adjacent node information, Express node 3 ₁ of the adjacent node information, Express node 4 ₁ adjacent node information and taking out the identification information of the switch 2 _1, and outputs each neighbor node information and the identification information retrieved to the network topology information management section 31. Network topology information managing unit 31, when receiving best effort node 1 ₁ of the adjacent node information, neighboring node information of Express node 3 _1, the adjacent node information and identification information of the switch 2 ₁ of the Express node 4 _1, switch 2 ₁ Are adjacent to the best effort node 1 ₁ and the express nodes 3 ₁ and 4 ₁ . That is, the network topology information management unit 31 grasps that the best effort node 1 ₁ and the Express nodes 3 ₁ and 4 ₁ are connected to the switch 2 ₁ . Switch 2 ₁ has been described when receiving the frame transmission source, but the same when other switch 2 ₁ receives the transmission source of the frame.

  The network topology information management unit 31 includes the identification information of each switch 2 constituting the mesh network, the adjacent node information collected from the adjacent nodes by each switch 2 constituting the mesh network, and advertised into the mesh network. Holding. Further, the network topology information management unit 31 establishes a link between which switch 2 and which switch 2 the network topology information collected by IS-IS (Intermediate System to Intermediate System) which is the SPB routing protocol. Information indicating whether or not As described above, the adjacent node information includes the identification information of the adjacent node, the attribute information of the adjacent node, the communication partner node information of the adjacent node, the bandwidth demand of the best effort traffic indicating the bandwidth required by the best effort node. Therefore, the network topology information management unit 31 includes, in addition to the network topology information, the attribute information of each node connected to each switch 2, which node communicates, Information indicating which node needs to set a path is also held. It also holds the bandwidth demand of each node that transmits and receives the best effort traffic (the bandwidth demand of the best effort traffic).

The Express traffic path determination unit 33, which is a low-delay transfer path determination unit, determines an Express traffic path according to the flowchart shown in FIG. Specifically, the Express traffic path determination unit 33 continuously monitors whether the start condition of the Express traffic path determination process is satisfied (step S51). Here, the process start condition is, for example, when a specified time has elapsed since the last Express traffic path was determined, or when a user performs an operation to instruct a process start, when a new node is connected to the own switch. In such a case, it may be possible to detect that a new node is connected to another switch that forms a mesh network with the own switch. When the start condition of the Express traffic path determination process is satisfied (step S51: Yes), the Express traffic path determination unit 33 reads the network topology information and adjacent node information held by the network topology information management unit 31. (Step S52). Next, the Express traffic path determination unit 33 determines the attribute information and the communication partner of the read information, specifically, the network topology information and the adjacent node information of each node connected to each switch 2. Based on the information, a path for transferring Express traffic between the switches 2 to which the Express node is connected is provisionally determined (step S53). The path temporarily determined in step S53 is referred to as an Express traffic path. In step S53, the Express traffic path determination unit 33 sets each Express traffic so that the number of times the duplicate traffic is used in the Express traffic path is reduced when a plurality of Express traffic paths are set for each of the links constituting the mesh network. Temporarily determine the path. Specifically, the Express traffic path determination unit 33 solves each Express traffic by solving the linear programming problem so that the link overlap that maximizes the link overlap used by each Express traffic path in the network is minimized. Temporarily determine the path. The Express traffic path determination unit 33 outputs the information on each Express traffic path that is provisionally determined to the link metric determination unit 32 (step S54). The Express traffic path information includes identification information of an Express node that transmits and receives Express traffic, and identification information of a switch 2 that configures the Express traffic path, that is, identification information of the switch 2 that relays the Express traffic. The Express traffic path for transmitting / receiving the Express traffic transmitted / received between the Express node 3 ₁ and the Express node 3 ₂ shown in FIG. 1 is configured by the switch 2 ₁ , the switch 2 ₂ , the switch 2 ₅ and the switch 2 _7. For example, the information of the Express traffic path can be as follows. “Express Traffic Path Information = Express Node 3 ₁ Identification Information + Switch 2 ₁ Identification Information + Switch 2 ₂ Identification Information + Switch 2 ₅ Identification Information + Switch 2 ₇ Identification Information + Express Node 3 ₂ Identification Information” In the link metric determination unit 32 that has received this information, the Express traffic path between the Express node 3 ₁ and the Express node 3 ₂ is configured by the switch 2 ₁ , the switch 2 ₂ , the switch 2 _5, and the switch 2 ₇ . Can be grasped.

  The link metric determination unit 32, which is a link metric determination unit, determines the link metric of each link in the mesh network according to the flowchart shown in FIG. Specifically, the link metric determination unit 32 continuously monitors whether or not the Express traffic path information has been received from the Express traffic path determination unit 33 (step S61). When the link metric determination unit 32 receives the information of the Express traffic path (step S61: Yes), the link metric determination unit 32 reads out the network topology information and the bandwidth demand information of the best effort traffic held by the network topology information management unit 31 ( Step S62). Next, the link metric determination unit 32 reads the network topology information read from the network topology information management unit 31 and the bandwidth demand information of the best effort traffic, and the information of each Express traffic path received from the Express traffic path determination unit 33. Based on the above, the link metric of each link in the mesh network is determined (step S63). Specifically, the link metric determination unit 32 determines the number of link metrics obtained by adding 1 to the number of Express traffic for each link. Here, the number of Express traffic is the number of combinations of Express nodes that transmit and receive Express traffic. The link metric determination unit 32 counts the number of Express traffic by confirming the information of each Express traffic path received from the Express traffic path determination unit 33, and adds the number of link metrics obtained by adding 1 to the number of Express traffic. Judge that it is necessary to decide. In the present embodiment, since the configuration of FIG. 1 is assumed, the number of Express traffic is two. That is, the link metric determination unit 32 determines that it is necessary to determine three link metrics. The link metric is information necessary for determining a traffic path, which is a communication path between nodes, using a routing protocol.

  The link metric determination unit 32 assigns one link metric, that is, the first link metric, to each Express traffic for each traffic, and assigns one link metric, that is, the second link metric, to the best effort traffic for the entire best effort traffic. . Specifically, the link metric determination unit 32 calculates the shortest path by the SPB based on these link metrics so that each Express traffic and the best effort traffic are transferred through a desired path. The link metric is determined so that the frequency of occurrence of the error is low. As described above, the link metric determining unit 32 conventionally determines one link metric for one link, but determines a plurality of link metrics for one link. Details of the operation in which the link metric determination unit determines the link metric will be described later. The link metric determination unit 32 is referred to as an extended link metric. The link metric determination unit 32 outputs the determined link metric to the SPB processing unit 27 (step 64). Note that the link metric determination unit 32 and the Express traffic path determination unit 33 constitute a determination unit.

  The SPB processing unit 27 determines a frame transfer path between nodes that transmit and receive frames via the mesh network according to the flowchart shown in FIG. Specifically, the SPB processing unit 27 continuously monitors whether or not a link metric has been received from the link metric determining unit (step S71). When receiving the link metric (step S71: Yes), the SPB processing unit 27 reads the network topology information from the network topology information management unit 31 (step S72). Next, based on the read topology information and the link metric of each link received from the link metric determination unit 32, the SPB processing unit 27 uses SPB to calculate the shortest path of a frame transmitted / received by each node ( Step S73). The link metric of each link received from the link metric determination unit 32 is the above-described link metric obtained by adding 1 to the number of Express traffic, that is, the link metric obtained by extending the link metric of one link to a plurality of links. . Next, based on the calculation result of the shortest path, the SPB processing unit 27 determines to which port of the own switch the received frame is transferred, that is, the output port of the frame transmitted / received between the nodes (step S74). . Specifically, the SPB processing unit 27 determines the port to which the next switch 2 or node on each determined shortest path is connected as the transfer destination of the received frame. When determining the transfer destination port of the received frame, the SPB processing unit 27 outputs the determination result to the transfer database 26 (step S75).

  The capsuler 34 operates according to the flowchart shown in FIG. Specifically, the capsuler 34 continuously monitors whether a frame has been received from the frame processor 28 (step S81). When receiving a frame (step S81: Yes), the capsuler 34 checks whether the frame is advertised from another switch 2 (step S82). When the frame transmission source is another switch 2 and the frame destination is a broadcast address, the capsuler determines that the frame has been advertised by another switch. If the frame received from the frame processing unit 28 is a frame advertised by another switch 2 (step S82: Yes), the frame is output to the frame transfer unit 25 as it is (step S88). If the frame received from the frame processing unit 28 is not a frame advertised by another switch 2 (step S82: No), the encapsulation unit 34 determines whether the received frame is a frame addressed to an adjacent node of the own switch. Is confirmed (step S83). When the received frame is addressed to the adjacent node of the own switch (step S83: Yes), the frame is in a state where a header for forwarding in the mesh network is added by another switch 2, that is, in an encapsulated state. Then, decapsulation for removing the header added by the other switch 2 is performed (step S86), and output to the frame transfer unit 25 (step S87). On the other hand, when the received frame is not addressed to the adjacent node of the own switch (step S83: No), it is confirmed whether or not the transmission source of the received frame is an adjacent node of the own switch (step S84). When the transmission source of the received frame is an adjacent node of the own switch (step S84: Yes), the encapsulation unit 34 performs processing for adding a header for transfer in the mesh network to the received frame, that is, encapsulation. Perform (step S86) and output to the frame transfer unit 25 (step S87). In addition, when the transmission source of the frame received from the frame processing unit 28 is not an adjacent node of the own switch (step S84: No), the encapsulation unit 34 does not perform any processing on the frame and sends it to the frame transfer unit 25. Output (step S87).

  In the encapsulation process of step S86, the encapsulation unit 34 sets the MAC address of its own switch as a transmission source for a frame whose adjacent node is a transmission source, and the destination node of the frame is connected to the destination. A header in which the MAC address of the switch is set is added. Further, in the decapsulation process in step S85, the encapsulation unit 34 removes the header added to the frame, that is, the header in which the MAC address of the own switch is set as the destination, and the original before the header is added. Return to the frame.

  The frame transfer unit 25 operates according to the flowchart shown in FIG. Specifically, the frame transfer unit 25 continuously monitors whether or not a frame has been received from the encapsulation unit 34 (step S91). When the frame transfer unit 25 receives the frame (step S91: Yes), that is, when receiving the frame from the capsuling unit 34, the frame transfer unit 25 reads the information on the transfer destination of the frame from the transfer database 26 (step S92). Determine the transfer destination of the received frame. Then, the frame transfer unit 25 transfers the frame received from the encapsulation unit 34 toward the transfer destination via the transmission unit 22 (step S93).

  The Hello generation unit 23 transmits a Hello frame according to the flowchart shown in FIG. Specifically, the Hello generation unit 23 continuously monitors whether or not a specified time has elapsed (step S101). In step S <b> 101, the Hello generation unit 23 confirms whether a specified time has elapsed since the previous Hello frame was generated. When the specified time has elapsed (step S101: Yes), the Hello generation unit 23 creates a Hello frame (step) and transmits it into the mesh network via the transmission unit 22 (step S103). At this time, the Hello generation unit 23 sets a broadcast MAC address as the destination of the Hello frame.

  Next, a procedure in which the switch 2 determines a link metric in the mesh network according to the first embodiment illustrated in FIG. 1 and determines a path between nodes based on the determined link metric will be described.

FIG. 15 is a diagram illustrating a procedure in which the switch 2 according to the first embodiment determines a path between nodes connected to the mesh network. Each switch 2 determines a path between nodes that perform communication by executing a procedure # 1-1 to a procedure # 1-4 described below. Here, in the mesh network having the configuration shown in FIG. 1, the switches 2 ₁ to 2 ₇ are the best effort nodes 1 ₁ , 1 ₂ , 1 ₃ , 1 ₄ , and the Express nodes 3 ₁ , 3 ₂ , 4 ₁ , 4 _2. The operation for determining the path for each of the above will be described.

<Procedure # 1-1>
The switches 2 ₁ to 2 ₇ constituting the mesh network transmit and receive Hello frames 5 _{1 to} 5 ₈ with adjacent nodes and collect various information from the adjacent nodes. Here, the Hello frames 5 _{1 to} 5 ₈ are only required to collect information from each adjacent node, and a Hello frame generated by a protocol such as LLDP (Link Layer Discovery Protocol) can be used. Information to be collected includes the MAC address that is identification information of the adjacent node, attribute information indicating whether the adjacent node corresponds to an Express node or a best effort node, and information of a partner node that each adjacent node transmits / receives a frame (communication) Partner node information). When the adjacent node is a best effort node, information on bandwidth demand is also collected. Explaining a specific operation, each switch 2 broadcasts a Hello frame storing a MAC address which is identification information of the own switch. Meanwhile, best effort nodes 1 _1, 1 _2, 1 ₃ and 1 _4, MAC address of the own node, the attribute information indicating that the node is a best effort node, a correspondent node information, the bandwidth demand information, etc. Broadcast the stored Hello frame. The Express traffic nodes 3 ₁ , 3 ₂ , 4 ₁ and 4 ₂ broadcast a Hello frame storing the MAC address of the own node, attribute information indicating that the own node is an Express node, communication partner node information, and the like. Each switch 2 receives the Hello frame transmitted by the adjacent node and collects various information from the Hello frame, thereby collecting the MAC address, attribute information, communication partner node information, bandwidth demand information, and the like of the adjacent node. Various information collected from the adjacent nodes by each switch 2 is adjacent node information.

In the switches 2 _{1 to} 2 ₇ , the Hello generation unit 23 generates and broadcasts a Hello frame. In addition, the adjacent node information collection unit 29 receives the Hello frame received from the adjacent node from the frame processing unit 28, and extracts the adjacent node information from the received Hello frame.

Next, the switches 2 ₁ to 2 ₇ that have transmitted / received the Hello frame and collected the adjacent node information from each adjacent node will then send the MAC address of the own switch and the adjacent node information collected from each adjacent node to other switches. Advertise to 2. This process is performed by the adjacent node information advertising unit 24. The frame used when the adjacent node information advertising unit 24 of the switches 2 _{1 to} 2 ₇ advertises the MAC address of the own switch and the adjacent node information collected from the adjacent node into the mesh network has the configuration shown in FIG. . FIG. 16 is a diagram illustrating a configuration example of a frame advertised by the switches 2 ₁ to 2 ₇ to another switch 2, that is, the mesh network. A frame used when the switches 2 _{1 to} 2 ₇ advertise the adjacent node information (hereinafter referred to as an adjacent node information advertising frame) includes a header, own device identification information, adjacent node information # 1, # 2, # 3,. It is configured to include. The header includes frame destination information, transmission source information, frame type information, and the like. Although the MAC address of the own device is set in the own device identification information, the present invention is not limited to this. Any information that uniquely identifies the device may be used. The adjacent node information collected from the adjacent nodes is stored in the adjacent node information # 1, # 2, # 3,. For example, when the switch 2 ₁ transmits the neighboring node information advertisement frame, the switch 2 _1, the MAC address of the switch is set to the own device identification information, neighbor nodes neighboring node information collected from best effort node 1 ₁ set information # 1, and sets the adjacent node information collected from Express node 3 ₁ to the adjacent node information # 2, the neighboring node information collected from Express node 4 ₁ generates a frame set in the adjacent node information # 3 To switch 2 _{2 to} 2 ₇ . Note that the order of storing adjacent node information in the adjacent node information advertisement frame is not limited to this. Here, it is assumed that the switches 2 _{1 to} 2 ₇ have already executed IS-IS, which is an SPB routing protocol standardized in IEEE 802.1aq, and hold network topology information.

In addition, when receiving the adjacent node information advertisement frame advertised from the other switch 2, the switches 2 _{1 to} 2 ₇ transfer the adjacent node information advertisement frame to the other switch 2 and receive the adjacent node information from the received adjacent node information advertisement frame. Remove and keep. In the switches 2 _{1 to} 2 ₇ , the frame transfer unit 25 performs the process of transferring the adjacent node information advertisement frame to another switch 2. The node information collection unit 30 performs processing for extracting adjacent node information from the adjacent node information advertisement frame. The network topology information management unit 31 holds adjacent node information and network topology information. When the network topology information management unit 31 receives adjacent node information, information indicating which switch in the network the Express traffic is transferred and the best effort traffic is transferred between which switch in the network. And information indicating whether or not.

<Procedure # 1-2>
Next, the switches 2 _{1 to} 2 ₇ determine the path of the Express traffic. Specifically, the switches 2 ₁ to 2 ₇ tentatively determine the path of the Express traffic by solving the linear programming problem so that the overlap of the links used in the path of the Express traffic is minimized. Thus, for each of the links constituting the mesh network, the number is reduced to be overlapped used in Express traffic path, Express traffic can be kept low frequency to compete in the switch 2 ₁ to 2 _7. Switch 2 ₁ to 2 _7, for example, as shown in FIG. 15, the path between the Express node 3 ₁ and Express node 3 ₂ which is the first Express node pair is temporarily determined path 6 _2, the second path between the Express node 4 ₁ and Express node 4 ₂ is Express node pair is temporarily determines the path _61.

In the switches 2 _{1 to} 2 ₇ , the Express traffic path determination unit 33 transfers Express traffic between the information held in the network topology information management unit 31, that is, between the network topology information and any switch in the network. The path of the Express traffic is provisionally determined by solving the linear programming problem based on the information indicating whether or not. Note that the switches 2 _{1 to} 2 ₇ exchange the adjacent node information acquired from the adjacent nodes by executing the above-described procedure # 1-1. That is, the switches 2 _{1 to} 2 ₇ are in a state of holding the same information. Therefore, the Express traffic path determining section 33 of the switch 2 ₁ to 2 ₇ will be temporarily determine the path of Express traffic using the same algorithm with the same information. Accordingly, the path of Express traffic is determined at each Express traffic path determining section 33 of the switch 2 ₁ to 2 ₇ is the same.

<Procedure # 1-3>
Next, the switches 2 _{1 to} 2 ₇ determine the link metric of each link in the network. Normally, there is one link metric, but in this embodiment, the number of link metrics is determined by adding 1 to the number of Express node pairs.

In the mesh network shown in FIG. 1, the Express nodes 3 ₁ and 3 ₂ transmit and receive Express traffic, and the Express nodes 4 ₁ and 4 ₂ transmit and receive Express traffic, so the number of Express node pairs is two. Therefore, the number of link metrics for each link is expanded to 3. The first link metric for each link is used to determine the path between the _first Express node pair, Express node 3 ₁ and Express node 3 ₂ , and the second link metric is the second was used to determine the path between the Express node 4 ₁ and Express node 4 ₂ is Express node pair, the third link metric used to determine the path of best effort traffic. In FIG. 15, the link metric is expressed in the format “(A, B, C)”. The numerical value described at the position “A” is the first link metric, the numerical value described at the position “B” is the second link metric, and the numerical value described at the position “C” is the third. Link metric.

The switches 2 _{1 to} 2 ₇ are the first Express node pair determined in step # 1-2 when the shortest path is calculated by the SPB for the first link metric of the link metrics 7 _{1 to} 7 ₉ of each link. A path between a certain Express node 3 ₁ and an Express node 3 ₂ becomes a path 6 ₂ , and the path cost of this path 6 ₂ is determined to be smaller than the link metric of other links. Here, "other links of the link metric" refers to first link metric for each link that does not constitute a path 6 _2. In the example shown in FIG. 15, large because the path 6 ₂ path cost is "3", the first link metrics other links are links that do not constitute a path 6 ₂ than "3" A value, for example, “4” is determined. The condition that the link metric should satisfy can be expressed by the following equation (1).

In the above equation (1), x _i ^m is the m-th link metric of link i, x _k ^m is the m-th link metric of link k, and X is the link used by the path between the m-th Express node pair. A set Y indicates a set of links other than X in the network.

The switches 2 _{1 to} 2 ₇ determine the second link metric of the link metrics 7 _{1 to} 7 ₉ of each link in the same manner as the first link metric. That is, when the shortest path is calculated by the SPB, the path between the node 4 ₁ and the node 4 ₂ , which is the second Express node pair determined in the procedure # 1-2, becomes the path 6 ₁ , and this path 6 ₁ The path cost is determined to be smaller than the link metric of other links.

The switches 2 _{1 to} 2 ₇ determine the third link metric of the link metrics 7 _{1 to} 7 ₉ of each link so that the load balancing of the best effort traffic can be realized. For example, the method disclosed in Non-Patent Document 1, specifically, by solving the linear programming problem so that the maximum value of the utilization rate of all links is minimized, the third link metric of each link is obtained. decide. As another determination method, a reciprocal of the link bandwidth may be determined as a link metric. The method for determining the third link metric of the link metrics 7 _{1 to} 7 ₉ for each link may be any method as long as a link metric capable of realizing load balancing of best effort traffic can be obtained.

In the switches 2 _{1 to} 2 ₇ , the link metric determination unit 32 determines each link based on the information held in the network topology information management unit 31 and the information on the Express traffic path determined by the Express traffic path determination unit 33. Determine the link metric for.

<Procedure # 1-4>
Next, the switches 2 _{1 to} 2 ₇ determine a path between the nodes using the SPB. Specifically, the switches 2 ₁ to 2 ₇ use the first link metric of each link to make the path between the Express nodes 3 ₁ and 3 ₂ , which is the first Express node pair, the shortest path. To be determined. In the example of FIG. 15, the path 8 is determined (procedure # 1-4 (1)). Further, the switches 2 _{1 to} 2 ₇ use the second link metric of each link to determine the path between the Express nodes 4 ₁ and 4 ₂ that are the second Express node pair to be the shortest path. To do. In the example of FIG. 15, the path 9 is determined (procedure # 1-4 (2)). Further, the switches 2 _{1 to} 2 ₇ use the third link metric of each link to determine the path between the best effort nodes to be the shortest path (procedure # 1-4 (3)).

In the switches 2 _{1 to} 2 ₇ , the SPB processing unit 27 determines the path between the nodes based on the information held in the network topology information management unit 31 and the link metric determined by the link metric determination unit 32. decide.

  In this embodiment, a case has been described in which an Express node transmits and receives a frame on a one-to-one basis. However, a frame may be transmitted and received on a one-to-many basis or a many-to-many basis. In this case, the number of link metrics obtained by adding 1 to the number of combinations of Express nodes that transmit and receive frames is determined as the link metric for each link. For example, in the mesh network, when the node N1 transmits and receives an Express traffic frame between the nodes N2 to N5, the switch 2 uses a link metric for determining a path between the node N1 and the node N2, and the node N1. Link metric for determining the path between the nodes N3, Link metric for determining the path between the nodes N1 and N4, Link metric for determining the path between the nodes N1 and N5, Best effort Determine the link metric for determining the traffic path. When nodes N1 and N2 transmit and receive Express traffic frames between nodes N3 and N4, switch 2 uses a link metric for determining a path between node N1 and node N3, node N1 and node N4. A link metric for determining a path between the nodes N2 and N3, a link metric for determining a path between the nodes N2 and N4, and a best effort traffic A link metric for determining a path is determined.

  As described above, the switch 2 according to the present embodiment includes, from each node connected to the mesh network, attribute information indicating whether it is an Express node or a best effort node, and a partner node that transmits and receives a frame. The communication partner node information shown is collected, and based on the collected information, first, the path between the Express nodes that transmit and receive Express traffic is determined so that the overlap of the links constituting the path is minimized. Based on the path cost of the determined path, for each pair of Express nodes that transmit and receive Express traffic, a link metric to be used when determining a path with SPB is determined individually. In addition, the link metric used when determining the path between the best effort nodes that transmit and receive the best effort traffic by the SPB is determined. As a result, when applying SPB to a mesh network having a configuration in which an Express node and a best effort node are mixed, each switch determines a path that suppresses the frequency with which Express traffic forwarding processing competes low. The link metric can be derived as follows.

Embodiment 2. FIG.
FIG. 17 is a diagram illustrating a configuration example of the switches 2a _{1 to} 2a ₇ which are frame transfer apparatuses according to the second embodiment.

It is assumed that the switches 2a _{1 to} 2a ₇ constitute a mesh network having the same configuration as that described in the first embodiment. That is, the switch 2a ₁ to 2A region _7, it is assumed that the switch 2 ₁ to 2 ₇ constitutes the structure of the mesh network by replacing the switch 2a ₁ to 2A region ₇ in FIG. 1. The connection relationship between the switches 2a _{1 to} 2a ₇ and the best effort node and the Express node is the same as that in the first embodiment. Thus, the switch 2a _1, a best effort nodes 1 _1, is connected to the Express node 3 ₁ and 4 _1. The switch 2a _2, best effort node 1 ₂ are connected. The switch 2a _3, best effort node 1 ₄ are connected. A best effort node 1 ₃ the switch 2a _7, are connected to the Express node 3 ₂ and 4 _2.

The switches 2a _{1 to} 2a ₇ of the present embodiment replace the Express traffic path determination unit 33 provided in the switches 2 ₁ to 2 _{7 of the first} embodiment shown in FIG. 2 with a linear programming problem solving unit 35, and The SPB processing unit 27 and the link metric determination unit 32 are replaced with an SPB processing unit 27a and a link metric determination unit 32a. In the present embodiment, the description of the parts common to the switches 2 ₁ to 2 _{7 of the first} embodiment will be omitted. The linear programming problem solving unit 35, the SPB processing unit 27a, and the link metric determining unit 32a can be realized by the hardware shown in FIG. The linear programming problem solving unit 35 is realized by the processor 103 executing a program for operating as the linear programming problem solving unit 35. The SPB processing unit 27a is realized by the processor 103 executing a program for operating as the SPB processing unit 27a. The link metric determination unit 32a is realized by the processor 103 executing a program for operating as the link metric determination unit 32a.

Also in the switches 2a _{1 to} 2a ₇ of the present embodiment, a link metric expanded as a link metric of each link, that is, a plurality of link metrics are allocated, and a path between nodes transmitting and receiving frames is determined.

  The linear programming problem solving unit 35, which is the first link metric determining means, determines the link metric used when determining the path of the Express traffic according to the flowchart shown in FIG. Specifically, the linear programming problem solving unit 35 continuously monitors whether or not the start condition of the process for determining the link metric is satisfied (step S111). Here, the process start condition is, for example, when a specified time has elapsed since the last time the link metric was determined, or when a user performs an operation to instruct the process to start, a new node is connected to the own switch. For example, a case where it is detected that a new node is connected to another switch that forms a mesh network with the own switch may be considered. When the start condition of the link metric determination process is satisfied (step S51: Yes), the linear programming problem solving unit 35 reads the network topology information and adjacent node information held by the network topology information management unit 31 ( Step S112). Next, the linear programming problem solving unit 35, among the read information, specifically, the topology information of the network and the adjacent node information of each node connected to each switch 2, attribute information, and each node Based on the communication partner information, the link metric of each link used when determining the path of the Express traffic is determined (step S113). More specifically, the linear programming problem solving unit 35 determines the link metric of each link used when determining the path of Express traffic by solving the linear programming problem. For example, the link metric of each link is determined by a method similar to the method described in Non-Patent Document 1. Specifically, Non-Patent Document 1 discloses a method for determining a link metric by solving a linear programming problem so that the maximum value of the utilization rate of all links is minimized. On the other hand, the linear programming problem solving unit 35 solves the Express traffic by solving the linear programming problem so that the link overlap that maximizes the link overlap used by each Express traffic path in the network is minimized. Determine the link metric for each link used in determining the path. The linear programming problem determination unit 35 outputs the determined link metric, that is, the link metric used when determining the path of the Express traffic to the link metric determination unit (step S114).

  The link metric determination unit 32a as the second link metric determination means determines the link metric of each link in the mesh network according to the flowchart shown in FIG. Specifically, the link metric determination unit 32a continuously monitors whether or not a link metric has been received from the linear programming problem solving unit 35 (step S121). When the link metric determination unit 32a receives the link metric determined by the linear programming problem solving unit 35 (step S121: Yes), the network topology information and the best effort traffic held by the network topology information management unit 31 The bandwidth demand information is read out (step S212). Next, the link metric determination unit 32a uses each of the network topology information read from the network topology information management unit 31 and the bandwidth demand information of the best effort traffic to determine the path of the best effort traffic. The link metric of the link is determined (step S213). Next, the link metric determining unit 32a determines the link metric received from the linear programming problem solving unit 35, that is, the link metric used when determining the path of the Express traffic, and the determined link metric, that is, the path of the best effort traffic. Together with the link metric used for the determination, the link metric is output as an extended link metric to the SPB processing unit (step S214). The extended link metric according to the present embodiment includes two types of link metric: a link metric used when determining the path of Express traffic and a link metric used when determining the path of best effort traffic. The link metric determining unit 32a constitutes a determining unit together with the linear programming problem solving unit 35.

  The SPB processing unit 27a uses the SPB based on the network topology information held by the network topology information management unit 31 and the extended link metric held by the link metric determination unit 32a to transmit and receive frames transmitted and received by each node. Calculate the shortest path of. The operation of the SPB processing unit 27a is the same as that of the SPB processing unit 27 described in the first embodiment. That is, as shown in FIG. 11, when the link metric is received (step S71: Yes), the SPB processing unit 27a reads the network topology information from the network topology information management unit 31 (step S72). Next, based on the read topology information and the link metric of each link received from the link metric determination unit 32, the SPB processing unit 27a uses SPB to calculate the shortest path of a frame transmitted / received by each node ( Step S73). Next, the SPB processing unit 27 determines which port of the own switch to transfer the received frame based on the calculation result of the shortest path (step S74), and outputs the determination result to the transfer database 26 (step S75). ). However, the link metric used when calculating the shortest path in step S73 is different from that of the first embodiment. Details of the shortest path calculation method will be described later.

In the switches 2 ₁ to 2 _{7 according to the first} embodiment, both the link metric determination unit 32 uses the link metric used when determining the path of the Express traffic and the link metric used when determining the path of the best effort traffic. However, in the switches 2a _{1 to} 2a ₇ of this embodiment, the linear metric problem solving unit 35 determines the link metric used when determining the path of the Express traffic, and the link metric determining unit 32a Determine the link metric to use when determining the path for best effort traffic. Note that the function of the linear programming problem solving unit 35 may be provided in the link metric determining unit 32a. That is, the link metric determination unit 32a may determine both the link metric used when determining the path of Express traffic and the link metric used when determining the path of best effort traffic.

Next, a procedure in which the switches 2a _{1 to} 2a ₇ determine the link metric in the mesh network of the second embodiment and determine the path between the nodes based on the determined link metric will be described.

FIG. 20 is a diagram illustrating a procedure in which the switch 2a according to the second embodiment determines a path between nodes connected to the mesh network. Each switch 2a determines a path between nodes that perform communication by executing a procedure # 2-1 to a procedure # 2-3 described below. Here, in a mesh network similar to the configuration shown in FIG. 1, the switches 2a _{1 to} 2a ₇ are connected to best effort nodes 1 ₁ , 1 ₂ , 1 ₃ , 1 ₄ , Express nodes 3 ₁ , 3 ₂ , 4 ₁ , 4 an example of determining the path for each of ₂ will be described.

<Procedure # 2-1>
This procedure # 2-1 executes the same process as the procedure # 1-1 described in the first embodiment, and each switch 2a is a best effort node 1 ₁ , 1 ₂ , 1 ₃ , 1 ₄ , an Express node 3. Information on each of ₁ , 3 ₂ , 4 ₁ , 4 ₂ , that is, adjacent node information is collected. Since it is the same process as procedure # 1-1, description of detailed operation is omitted.

<Procedure # 2-2>
Next, the switches 2a _{1 to a} 2 ₇ determine the link metric of each link in the network. Normally, there is one link metric, but in this embodiment, two link metrics are determined.

  The first of the link metrics for each link is used to determine the path for Express traffic, and the second of the link metrics for each link is used to determine the path of best effort traffic. In FIG. 20, the link metric is expressed in the format “(A, B)”. The numerical value described at the position “A” is the first link metric, and the numerical value described at the position “B” is the second link metric.

The switches 2a _{1 to} 2a ₇ have the minimum link overlap for the _first link metric of the link metrics 10 ₁ to 10 ₉ for each link, with the maximum link overlap used by each Express traffic path in the network. The link metric is determined by solving the linear programming problem as follows. The switches 2a _{1 to} 2a ₇ determine the second link metric of the link metrics 10 ₁ to 10 ₉ of each link so that the load balancing of the best effort traffic can be realized. For example, the second link metric of each link is determined by solving the linear programming problem so that the maximum value of the utilization rate of all links is minimized based on the bandwidth demand of each best effort node. As another determination method, a reciprocal of the link bandwidth may be determined as a link metric.

In the switches 2a _{1 to} 2a ₇ , the linear programming problem solving unit 35 determines the first link metric, and the link metric determination unit 32a determines the second link metric. The link metric determination unit 32a holds the first link metric and the second link metric together as an extended link metric.

<Procedure # 2-3>
Next, the switches 2a _{1 to} 2a ₇ determine a path between the nodes using the SPB. Specifically, the switches 2a _{1 to} 2a ₇ use the first link metric of each link to determine the path of the Express traffic to be the shortest path. Here, since there are two Express node pairs, two Express traffic paths, that is, ECMPs 11 and 12 shown in the figure, are determined. The switches 2a _{1 to} 2a ₇ assign the determined ECMPs 11 and 12 to each Express node pair. For example, VID (Virtual Local Area Network ID) is set so that ECMP11 is used for the pair of Express node 3 ₁ and Express node 3 ₂ , and ECMP12 is used for the pair of Express node 4 ₁ and Express node 4 _2. Set the VID. Also, the switches 2a _{1 to} 2a ₇ use the second link metric of each link to determine the best effort traffic path to be the shortest path.

In the switches 2a _{1 to} 2a ₇ , the SPB processing unit 27a performs inter-node communication based on the information held in the network topology information management unit 31 and the extended link metric held in the link metric determination unit 32a. Determine the path.

  As described above, the switch 2a according to the present embodiment, from each node connected to the mesh network, sets the attribute information indicating whether it corresponds to the Express node or the best effort node, and the partner node that transmits and receives the frame. Communication partner node information to be collected, and based on the collected information, a link metric used when determining a path between the Express nodes that transmit and receive Express traffic, and the best effort node that transmits and receives the best effort traffic Determine the link metric to use when determining the path between. When determining the link metric used to determine the path between Express nodes that send and receive Express traffic, the link overlap that minimizes the link overlap used by the path between each Express traffic node is minimized. In other words, the link metric is determined by solving the linear programming problem so that the number of overlapping links constituting the Express traffic path is minimized. As a result, when applying SPB to a mesh network having a configuration in which an Express node and a best effort node are mixed, each switch determines a path that suppresses the frequency with which Express traffic forwarding processing competes low. The link metric can be derived as follows.

  The configuration described in the above embodiment shows an example of the contents of the present invention, and can be combined with another known technique, and can be combined with other configurations without departing from the gist of the present invention. It is also possible to omit or change the part.

1 ₁ to 1 ₄ best effort _{node, 2 1 ~2 7, 2a 1} ~2a 7 switch, 3 _1, 3 _2, 4 _1, 4 ₂ Express node 21 receiving unit, 22 transmission unit, 23 Hello generator, 24 Adjacent node information advertising unit, 25 frame transfer unit, 26 transfer database, 27, 27a SPB processing unit, 28 frame processing unit, 29 adjacent node information collection unit, 30 node information collection unit, 31 network topology information management unit, 32, 32a Link metric determination unit, 33 Express traffic path determination unit, 34 encapsulation unit, 35 linear programming problem solution unit.

Claims (11)

A frame transfer apparatus constituting a mesh network,
Neighboring node information collecting means for collecting neighboring node information including attribute information indicating whether or not a node that transmits / receives a frame that requires low-delay transfer from each of neighboring nodes and communication partner node information indicating a communication partner node; ,
Neighboring node information advertising means for advertising the neighboring node information collected by the neighboring node information collecting means in the mesh network;
Based on the neighboring node information collected by the neighboring node information collecting means and the neighboring node information advertised into the mesh network by another frame transfer device, for each link constituting the mesh network, A first link metric used for determining a path for transmitting / receiving a frame that requires low-delay transfer, and a second link metric used for determining a path for transmitting / receiving a frame that does not require low-delay transfer A determination means for determining a link metric;
A frame transfer apparatus comprising: The determining means includes
A path between nodes that transmits and receives a frame that requires low-delay forwarding based on the neighboring node information, the topology information of the mesh network, and information on neighboring nodes of each frame forwarding device that constitutes the mesh network Low-delay transfer path determination means for temporarily determining
Based on the path cost of the path tentatively determined by the low-delay transfer path determination means, the topology information of the mesh network, and the information on the bandwidth required by the node that transmits and receives frames that do not require low-delay transfer, Link metric determination means for determining the first link metric and the second link metric;
The frame transfer apparatus according to claim 1, further comprising: The low-delay transfer path determination means, when temporarily determining a plurality of Express paths that are paths between nodes that transmit and receive frames that require low-delay transfer, express paths for each of the links constituting the mesh network. Tentatively determine each Express path so that the number of times it is used repeatedly in
The frame transfer apparatus according to claim 2, wherein: The low-delay transfer path determination means temporarily determines the Express path by solving a linear programming problem so that the link overlap that maximizes the link overlap used by the Express path is minimized.
The frame transfer apparatus according to claim 3.   The link metric determination means determines a plurality of link metrics as many as the number of pairs of nodes that transmit and receive a frame that requires low-latency transfer to be the first link metric, and each of the plurality of link metrics is 5. The frame transfer apparatus according to claim 2, wherein the frame transfer apparatus is associated one-to-one with any one of a pair of nodes that transmit and receive a frame that requires low-delay transfer. The determining means includes
A path between nodes that transmits and receives a frame that requires low-delay forwarding based on the neighboring node information, the topology information of the mesh network, and information on neighboring nodes of each frame forwarding device that constitutes the mesh network First link metric determination means for determining the first link metric by solving a linear programming problem so that the link overlap that maximizes the link overlap is minimized;
Second link metric determination means for determining the second link metric based on topology information of the mesh network and information on a bandwidth required by a node that transmits and receives a frame that does not require low-latency transfer;
The frame transfer apparatus according to claim 1, further comprising: Path determining means for determining a path between nodes transmitting and receiving frames via the mesh network based on the first metric and the second metric of each link determined by the metric determining means;
The frame transfer device according to claim 1, further comprising: A link metric determination method for determining a link metric used in a process for determining a path between each node that transmits and receives a frame via the mesh network, wherein the frame transfer apparatus constituting the mesh network includes:
An adjacent node information collecting step for collecting adjacent node information including attribute information indicating whether or not a node that transmits / receives a frame that requires low-delay transfer from each of adjacent nodes and communication partner node information indicating a communication partner node; ,
An adjacent node information advertising step of advertising the adjacent node information collected in the adjacent node information collecting step in the mesh network;
Based on the neighboring node information collected in the neighboring node information collecting step and the neighboring node information advertised into the mesh network by another frame transfer device, for each link constituting the mesh network, A first link metric used for determining a path for transmitting / receiving a frame that requires low-delay transfer, and a second link metric used for determining a path for transmitting / receiving a frame that does not require low-delay transfer A decision step for determining a link metric;
A link metric determination method comprising: The determining step includes
A path between nodes that transmits and receives a frame that requires low-delay forwarding based on the neighboring node information, the topology information of the mesh network, and information on neighboring nodes of each frame forwarding device that constitutes the mesh network A low-delay transfer path determination step that temporarily determines
Based on the path cost of the path tentatively determined in the low-delay transfer path determination step, the topology information of the mesh network, and the information on the bandwidth required by the node that transmits and receives frames that do not require low-delay transfer, A link metric determination step for determining the first link metric and the second link metric;
The link metric determination method according to claim 8, comprising: The determining step includes
A path between nodes that transmits and receives a frame that requires low-delay forwarding based on the neighboring node information, the topology information of the mesh network, and information on neighboring nodes of each frame forwarding device that constitutes the mesh network A first link metric determination step for determining the first link metric by solving a linear programming problem such that a link overlap that maximizes a link overlap is minimized;
A second link metric determination step for determining the second link metric based on topology information of the mesh network and information on a bandwidth required by a node that transmits and receives a frame that does not require low-latency transfer;
The link metric determination method according to claim 8, comprising: A method for determining a path in a frame transfer apparatus constituting a mesh network,
An adjacent node information collecting step for collecting adjacent node information including attribute information indicating whether or not a node that transmits / receives a frame that requires low-delay transfer from each of adjacent nodes and communication partner node information indicating a communication partner node; ,
An adjacent node information advertising step of advertising the adjacent node information collected in the adjacent node information collecting step in the mesh network;
Based on the neighboring node information collected in the neighboring node information collecting step and the neighboring node information advertised into the mesh network by another frame transfer device, for each link constituting the mesh network, A first link metric used for determining a path for transmitting / receiving a frame that requires low-delay transfer, and a second link metric used for determining a path for transmitting / receiving a frame that does not require low-delay transfer A decision step for determining a link metric;
A path determining step for determining a path between nodes transmitting and receiving frames via the mesh network based on the first link metric and the second link metric;
Including a path determination method.

Images