JP4123437B2 - Network system, spanning tree configuration method, spanning tree configuration node, and spanning tree configuration program - Google Patents

Description

  The present invention relates to a network system, and more particularly to a network system, a spanning tree configuration method, and a spanning tree configuration node that do not stop a network when reconfiguring a spanning tree and further have a load distribution function.

  Conventionally, this type of spanning tree is used in a network formed in a loop shape to prevent data from circulating forever.

  For example, in the standardized document published by IEEE titled “1998, IEEE Std 802.1D”, data circulates forever in a loop-shaped network. In order to prevent this problem, control information called Bridge Protocol Data Unit (BPDU) is exchanged between nodes, and a part of the network that is physically looped is logically disabled and logically A control method called a spanning tree that forms a tree-like topology is defined. This is referred to as Prior Art 1.

  Also, in the standardized document issued by IEEE titled “2001, iStdE Standard 802.1w (IEEE Std 802.1w)”, the tree creation in the prior art 1 is made possible by extending the control information exchange method. A control method called high-speed spanning tree is defined in which a high-speed detour path is set in advance by setting a detour path in advance, thereby setting a high-speed detour path when a failure occurs. This is referred to as Conventional Technology 2.

A standardized document issued by IEEE titled “1998, iStd 802.1D” A standardized document issued by IEEE entitled "2001, iStd E Standard 802.1w (IEEE Std 802.1w)"

  The conventional techniques described above have the following problems.

  First, there is a problem that frame arrival delays and omissions occur due to congestion.

  In the prior art 1, when adding and deleting nodes and links belonging to the spanning tree, the spanning tree is stopped and rebuilt from the beginning. Therefore, the entire network is stopped for a long time during the reconstruction, and congestion occurs. There were cases where the arrival of frames was delayed or lost.

  In the prior art 2, when adding or deleting nodes and links belonging to the spanning tree, the data frame transfer is stopped locally and the spanning tree is gradually rebuilt, so part of the network is stopped during the reconstruction. In some cases, congestion may occur, and the arrival of frames may be delayed or lost.

  Secondly, there is a problem that the network is stopped when a spanning tree is reconfigured such as addition and deletion of nodes belonging to the spanning tree.

  In the prior art 1, when adding or deleting a node belonging to the spanning tree, the spanning tree is stopped and rebuilt from the beginning. Therefore, the entire network may be stopped for a long time during the rebuilding.

  In Prior Art 2, when adding or deleting a node belonging to a spanning tree, the spanning tree is rebuilt gradually while stopping forwarding data frames locally. was there.

  Third, there is a problem that the load distribution of traffic cannot be performed.

  In the prior art 1 and the prior art 2, since the cost is calculated using the link capacity and used for the path selection when the spanning tree is constructed, the path cannot be changed for dynamic load distribution according to the traffic. It was.

  Fourth, when trying to distribute the load, there is a problem that the network stops due to spanning tree reconfiguration.

  In the prior art 1, when the cost is dynamically changed according to the traffic situation, the entire network is stopped for a long time during the reconfiguration because the span is changed after the spanning tree is stopped and then rebuilt. There was a case.

  In the prior art 2, when the cost is dynamically changed according to the traffic situation, the path is changed by gradually reconstructing a part of the spanning tree while locally stopping the transfer of the data frame. During the rebuilding, part of the network might stop.

  Fifth, there is a problem that the lowest cost route to the destination is not always selected.

  In prior art 1 and prior art 2, only one system of spanning tree is set on the network, and a single root node on the network is determined based on a priority value and a MAC address set in advance for each node. In order to create a tree, when nodes located at the end of the tree communicate with each other, even if there is another shortest path, the node is blocked and may pass through a redundant path.

  Sixth, there is a problem that the load is concentrated near the root node while the link utilization rate is low.

  In prior art 1 and prior art 2, only one spanning tree is set on the network, a single root node is determined on the network from the priority value and MAC address set for each node, and a single tree is created. Because it was created, links that were installed at the end of the tree but were not used appeared and the link utilization rate decreased. On the contrary, in the vicinity of the root node, there is a case where traffic is concentrated and the possibility of occurrence of congestion is increased.

  Seventh, there is a problem that it takes time to construct a tree when a root node fails, and the network stops during that time.

  In the prior art 1, since only one spanning tree is set on the network and there is only one root node, if a failure occurs in the root node, the spanning tree is stopped and rebuilt from the beginning. There was a case where the whole network stopped for a long time during construction.

  In the prior art 2, when a failure occurs in the root node, the spanning tree is gradually reconstructed while stopping the transfer of data frames locally, so part of the network may stop during the reconstruction. .

  Eighth, the section using IEEE 802.1D has a problem that the route switching at the time of failure is slow and it takes time to reconfigure the spanning tree.

  This is because, in the prior art 1, it may take several tens of seconds until data can be exchanged when a tree is constructed.

  Ninth, since both the prior art 1 and the prior art 2 have only a single tree, the traffic is concentrated in the vicinity of the root node and the arrival of frames is delayed or lost. was there.

  SUMMARY OF THE INVENTION A first object of the present invention is to reduce the probability of occurrence of congestion and reduce the frequency of frame arrival delay or loss due to congestion, a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree To provide a configuration program.

  A second object of the present invention is a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree capable of reconfiguring a spanning tree such as adding and deleting nodes belonging to the spanning tree without stopping the network. To provide a configuration program.

  A third object of the present invention is to provide a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree configuration program capable of distributing traffic load.

  A fourth object of the present invention is to provide a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree configuration program capable of load balancing without stopping the network for spanning tree reconfiguration associated with a path change. Is to provide.

  A fifth object of the present invention is to provide a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree configuration program in which a lowest cost route to a destination is selected.

  A sixth object of the present invention is to provide a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree configuration program, which can increase the link utilization rate and can be distributed without concentrating the load near the root node. There is.

  A seventh object of the present invention is to provide a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree configuration program capable of avoiding a network stop due to a root node failure.

  The eighth object of the present invention is to prevent the spanning tree from being set through the IEEE802.1D usage section, speed up switching and route change at the time of failure, and possible congestion and frame loss It is an object of the present invention to provide a network system, a spanning tree configuration method, a spanning tree configuration node, and a spanning tree configuration program that can reduce performance.

According to the first aspect of the present invention, in a node constituting a spanning tree on a network in which a plurality of nodes are connected, a spanning tree in which each node in the network is a root node and a destination is a root node is generated. It is characterized by using this method to transfer a frame .

The node of the present invention according to claim 2 corresponds to a plurality of tree managers that generate a plurality of spanning trees that operate independently in a node constituting a spanning tree on a network in which a plurality of nodes are connected, and a forwarding tree. A tag table that returns the received tag, a tag inserter that inserts the tag that has responded from the tag table into the frame, and a number of tree managers that are the same as the number of links that are using a slow-action protocol that exists in the network A tree selector, a forwarding table that records a frame transfer output destination for each destination, a frame transfer device that transfers a frame to a transfer output destination specified in the forwarding table, and a sorting that determines a tree manager of the transfer destination according to the tag And a vessel .

In the node according to the third aspect of the present invention, the tree selector transmits a main controller in the tree selector that creates or deletes a tree manager, a tag deleter that deletes a tag added to the frame, and a control frame. It comprises a GVRP transceiver and a tag inserter for adding a tag to a frame .

The node of the present invention according to claim 4 is characterized in that the tree manager deletes a tag added to a frame, a BPDU transmitter / receiver that transmits / receives a BPDU, a tag inserter that adds a tag to the frame, and spanning A tree controller that creates a spanning tree in accordance with a tree protocol and a tree table that holds parameters used in the spanning tree protocol are provided .

The present invention of claim 5 is a spanning tree configuration program that operates on each node constituting a spanning tree on a network in which a plurality of nodes are connected, and generates a spanning tree in which each node in the network is a root node; A frame is transferred using a tree whose destination is a root node .

The spanning tree configuration program of the present invention according to claim 6 generates a plurality of spanning trees that operate independently in a spanning tree configuration program that operates on each node that constitutes the spanning tree on a network in which a plurality of nodes are connected. A plurality of tree manager processes, a tag table process that returns a tag corresponding to the spanning tree used for forwarding, a tag insertion process that inserts a reply tag from the tag table into the frame, and a network presence Tree selector processing for generating the same number of tree managers as the number of root nodes to be transmitted, forwarding table processing for recording the frame transfer output destination for each destination, and frame transfer processing for transferring the frame to the transfer output destination specified in the forwarding table And the tag Characterized in that it comprises a sorting process for determining the destination tree manager of the frame I.

In the spanning tree configuration program of the present invention according to claim 7, the tree selector process includes a main controller process in a tree selector that creates or deletes a tree manager, a tag deletion process that deletes a tag added to a frame, A GVRP transmission / reception process for transmitting a control frame for switching a spanning tree and a tag insertion process for adding a tag to the frame are executed .

The spanning tree configuration program of the present invention according to claim 8 , wherein the tree manager process includes a tag deletion process for deleting a tag added to a frame, a BPDU transmission / reception process for transmitting / receiving a BPDU, and a tag insertion for adding a tag to the frame. And a tree controller process for creating a spanning tree in accordance with the spanning tree protocol, and a tree table process for holding parameters used in the spanning tree protocol .

The spanning tree configuration program of the present invention according to claim 9 generates a plurality of spanning trees that operate independently in a spanning tree configuration program that operates at each node that configures the spanning tree on a network in which a plurality of nodes are connected. Multiple tree manager processing, tag table processing that responds to tags corresponding to the transfer tree, tag insertion processing that inserts a tag that responds from the tag table into the frame, and a slow protocol that exists in the network A tree selector process that generates the same number of tree managers as the number of links in use, a forwarding table process that records a transfer output destination of a frame for each destination, and a frame that transfers a frame to the transfer output destination specified in the forwarding table Transfer processing and the tag Accordingly, characterized in that it comprises a sorting process for determining the destination tree manager frame.

In the spanning tree configuration program of the present invention according to claim 10, the tree selector process includes a main controller process in a tree selector that creates or deletes a tree manager, a tag deletion process that deletes a tag added to a frame, A GVRP transmission / reception process for transmitting a control frame and a tag insertion process for adding a tag to the frame are provided .

12. The spanning tree configuration program according to claim 11 , wherein the tree manager process includes a tag deletion process for deleting a tag added to a frame, a BPDU transmission / reception process for transmitting / receiving a BPDU, and a tag insertion for adding a tag to the frame. And a tree controller process for creating a spanning tree in accordance with the spanning tree protocol, and a tree table process for holding parameters used in the spanning tree protocol .

A network system according to a twelfth aspect of the present invention is a network system in which a forwarding path is set by a spanning tree on a network in which a plurality of nodes are connected, and a spanning tree in which each node in the network is a root node is generated. A frame is transferred using a tree as a root node .

A network system according to a thirteenth aspect of the present invention is a network system in which a forwarding path is set by a spanning tree on a network in which a plurality of nodes are connected, a plurality of tree managers that generate a plurality of spanning trees that operate independently, A tag table that returns a tag corresponding to the transfer tree, a tag inserter that inserts a tag returned from the tag table into a frame, and a tree selector that generates the same number of tree managers as the number of nodes existing in the network; A forwarding table that records a frame transfer output destination for each destination, a frame transfer device that transfers a frame to a transfer output destination specified in the forwarding table, and a classification that determines a tree manager of the frame transfer destination according to the tag and a vessel And wherein the door.

The present invention according to claim 14 includes a plurality of tree managers for generating a plurality of spanning trees that operate independently in a network system for setting a forwarding path by a spanning tree on a network in which a plurality of nodes are connected, and a forwarding tree. A tag table that responds to the tag corresponding to the tag, a tag inserter that inserts the tag that responds from the tag table into the frame, and the same number of tree managers as the number of links that are using a slow protocol existing in the network. A tree selector to be generated, a forwarding table for recording a frame transfer output destination for each destination, a frame transfer device for transferring a frame to a transfer output destination specified in the forwarding table, and a tree of the transfer destination of the frame according to the tag A separator to determine the manager and Characterized in that it comprises.

The network system according to a fifteenth aspect of the present invention is characterized in that the forwarding table has a broadcast output port field .

The network system according to a sixteenth aspect of the present invention is characterized in that the forwarding table has a spare output port field .

The network system of the present invention according to claim 17 is characterized in that an output destination port is determined using a port type determined by a spanning tree .

The network system of the present invention according to claim 18 is characterized in that the port type determined by the spanning tree is one of a root port and a designated port .

  According to the present invention, the following effects are achieved.

  First, it is possible to reduce the probability of occurrence of congestion and reduce the frequency of frame arrival delay or omission due to congestion. The reason is that while the spanning tree before the configuration change is used, a spanning tree including the newly added node is generated, the spanning tree used after stabilization of the new spanning tree is switched, and multiple spanning trees with different root nodes are used. This is because the system is set.

  Second, spanning tree reconfiguration such as addition and deletion of nodes belonging to the spanning tree can be performed without stopping the network. The reason is that a spanning tree including a newly added node is generated while operating the spanning tree before the configuration change, and the spanning tree used after the new spanning tree is stabilized is switched.

  Third, traffic load can be distributed. The reason is that the link cost is calculated based on dynamic information such as free bandwidth and server load.

  Fourth, load redistribution can be performed without stopping the network due to spanning tree reconfiguration accompanying path change. The reason is that a tree after the cost change is generated while operating the tree before the change, and the tree used after the stabilization of the new tree is switched.

  Fifth, the lowest cost route to the destination can be selected. This is because the frame is transferred using a tree whose destination is the root node.

  Sixth, it is possible to increase the link utilization rate and distribute without concentrating the load near the root node. This is because a plurality of spanning trees having different root nodes are set.

  Seventh, it is possible to avoid a network stop due to a root node failure. The reason is that by transferring a frame using a tree whose destination is the root node, frames other than the frame whose destination is the root node are not disabled for a long time due to the influence of the root node failure. is there.

  Eighth, the spanning tree is prevented from being set through the IEEE802.1D usage section, speeding up the switching and route change at the time of failure, and reducing the possibility of congestion and the loss of frames. Is possible. The reason is that the cost of the IEEE802.1D usage section is set large, and the spanning tree is prevented from being set through the IEEE802.1D usage section.

  Ninth, the detouring operation can be performed at high speed, and congestion and frame loss can be prevented. The reason is that tree managers are created as many as the number of IEEE 802.1D usage sections, a tree in which the cost of one different IEEE 802.1D usage section is set large is created for the IEEE 802.1D usage section, and the section is defined by a failure or the like. This is because when it is necessary to make a detour, the tree is switched to use a tree with a large cost for the section.

  Tenth, a failure detector that transmits and receives failure detection frames at short intervals has been added, because high-speed failure detection has not been possible due to the long transmission interval of HELLO frames conventionally used in the spanning tree protocol. Therefore, it is possible to detect a failure faster than the HELLO frame. As a result, it is possible to reduce the possibility of congestion and frame loss.

  Eleventh, the lowest cost route to the destination was not always selected during broadcasting, but in the present invention, the broadcast frame is transferred by using a tree in which the transmission source node is the root node. The broadcast frame can be transferred by selecting the lowest cost route to all nodes.

  Twelfth, by transferring a broadcast frame using a tree in which the transmission source node is the root node, broadcast frames other than the frame in which the root node is the transmission source node cannot be transferred for a long time due to the influence of the root node failure. Therefore, it is possible to avoid a network stop due to a root node failure. As a result, it is possible to reduce the possibility of congestion.

  13thly, the lowest cost route to the destination has not always been selected, but the lowest cost route to the destination is determined by transferring the unicast frame using a tree in which the destination is the root node. You can choose.

  Fourteenth, there has been a problem in the prior art that it takes time to switch the output destination port when the root port side link fails, and frame transfer stops during that time. By setting the output link in the forwarding table, it is possible to change the route at a high speed when the root port side link, that is, the output link fails. As a result, it is possible to reduce the possibility of congestion.

  In the following description, a description will be given using a tag as an identifier for identifying a plurality of spanning trees and a plurality of nodes. This tag is applied to Japanese Patent Application No. 2002-204673 by the present applicant in addition to the VLAN tag. It means a single or a combination of one or more of the expansion tag and other tags or identification means disclosed in No. 1.

  Here, among the tags used in the present invention, the format of a frame with an extension tag disclosed in the above Japanese Patent Application No. 2002-204673 will be described.

  FIG. 1 shows an Ethernet (R) frame format with a VLAN tag defined by IEEE 802.1Q. The Ethernet (R) frame 3200 with a VLAN tag includes a transmission destination MAC address 3201, a transmission source MAC address 3202, a VLAN tag 3203, Ethernet (R) attribute information 3204, a payload 3205, and an FCS 3206.

  On the other hand, FIG. 2 shows the format of an Ethernet (R) frame with an extension tag according to the present invention. The Ethernet frame with extension tag 3300 includes a transmission destination MAC address 3201, a transmission source MAC address 3202, an expansion tag storage area 3301, Ethernet (R) attribute information 3204, a payload 3205, and an FCS 3206. The VLAN tag 3203 of the existing Ethernet (R) frame with VLAN tag 3200 is replaced with the expansion tag storage area 3301.

  Further, as shown in FIG. 3, there is an Ethernet frame 3400 with an extension tag having another configuration, which includes a destination MAC address 3201, a source MAC address 3202, an extension tag storage area 3301, and a VLAN tag 3203. , Ethernet (R) attribute information 3204, payload 3205, and FCS 3206, and the extension tag storage area 3301 is inserted after the source MAC address 3202.

  One or a plurality of expansion tags can be stored in the expansion tag storage area 3301. The size of the expansion tag is 4 bytes, which is the same size as the VLAN tag 3203. The uppermost expansion tag of the Ethernet (R) frames 3300 and 3400 with the expansion tag and the VLAN tag of the Ethernet (R) frame 3200 with the VLAN tag are stored in the same position and with the same size. Are distinguished by changing the value stored in the upper 2 bytes of the data (details will be described later).

  As a result, Ethernet frames with extended tags (R) 3300 and 3400 are compatible with Ethernet frames with VLAN tags (R) 3200, and both frames can be processed by both existing nodes and nodes that support extended tags. It is.

  FIG. 4 shows the expansion tag storage area 3301. In the storage example shown in FIG. 4, eight expansion tags 3500-3507 are stored.

  The forwarding tag 3500 stores an identifier of the destination node and a label up to the destination (for example, MPLS label). In addition to the forwarding tag 3500 that stores the identifier of the destination node, the identifier of the transmission source node may be stored. Each node determines a frame transfer destination with reference to the forwarding tag. The forwarding tag 3500 is always stored in Ethernet frames (R) frames 3300 and 3400 with extension tags.

  As types of expansion tags, a customer separation tag 3501, a protection tag 3502, an OAM & P tag 3503, a quality information tag 3504, a frame control tag 3505, a security tag 3506, and a user expansion tag 3507 are stored.

  The customer separation tag 3502 stores an identifier for separating information for each customer accommodated in each node. As customers, the customers belonging to the same VLAN are the same customers, the customers accommodated in specific ports of two or more nodes are the same customers, or two connected to the nodes in the network. There are cases where the above hosts are the same customer. A separation identifier is assigned to these customers, and this separation identifier is stored in a customer separation tag 3502 in a frame from each customer. By identifying a customer by the customer separation tag 3502, it is possible to provide an additional service (for example, priority control for a specific customer) in units of customers. Further, a plurality of customer separation tags 3502 can be used in a stack. In this case, the number of separable customers can be greatly increased. When the customer separation tag 3502 is stacked, a special customer separation tag indicating that the customer separation tag 3502 at the last stage stacked is the last stage is used.

  The protection tag 3503 stores failure information at the time of failure occurrence and detour route information for failure recovery. The OAM & P tag 3504 stores operation / management information.

  The quality information tag 3505 stores quality information such as delay, jitter, packet loss rate, time stamp indicating the inflow time of the frame into the network, and band control information. When the time stamp value is stored in the quality information tag 3505, the node that has received the frame can calculate the delay in the network (the stay time in the network) of the frame from the current time and the time stamp value. When a guaranteed value for the delay in the network is determined, priority processing can be performed so that the guaranteed value can be realized. In addition, when the quality information tag 3505 stores bandwidth control information such as required bandwidth, accumulated data amount, traffic class, etc., the request is made in consideration of the accumulated data amount of the flow, the traffic class and the traffic status of other flows. Band control for securing the band can be performed.

  The frame control tag 3506 stores information such as a hop counter (TTL: Time To Live) that limits the lifetime of the frame in the network and a CRC for error detection. When the TTL is stored, the TTL value is subtracted for each passing node, and this frame is discarded when TTL = 0. As a result, it is possible to prevent the frame from continuing to circulate even when the loop path is reached. When the CRC is stored, the CRC calculation result of the extension tag storage area 3301 in the entry side node is stored, and the CRC calculation is performed again in the exit side node and compared with the stored value. An error in the storage area 3301 can be detected.

  The security tag 3507 stores information for ensuring the reliability of the frame and the confidentiality at the time of network construction or network configuration change. Examples of usage of the security tag 3507 include the following examples. A security identifier for each customer is set in advance for the customers communicating in the network, and the identifier is held in each node to which the customer connects. Each customer always stores the set security identifier in the security tag 3507 at the time of frame transfer, thereby preventing frame transmission / reception from a malicious customer who has altered the information of the customer separation tag 3501. In addition, when a network is constructed or a network configuration is changed, a common security identifier is set by negotiating between nodes. By always storing the set security identifier in the security tag 3507 during frame transfer between the nodes, it is possible to prevent a malicious node from being connected to the network.

  The user expansion tag 3507 stores arbitrary information defined by the user. When the user uniquely defines the tag format and stored information and defines the processing contents, the user can expand the functions unique to the user and increase the flexibility of the network.

  Expansion tags 3501 to 3507 other than the forwarding tag 3500 are stored as necessary. The forwarding tag 3500 is stored at the top of the expansion tag storage area 3301, and the other expansion tags 3501 to 3507 are stored behind it. As long as it is behind the forwarding tag 3500, it may be a predetermined fixed position or an arbitrary position.

  After that, out of the two spanning trees, the spanning tree used for transferring data frames entering the network is the working tree or working tree, and the non-working tree is the spare tree or spare tree. It expresses.

  In addition, a tree manager that generates a working tree is called a working tree manager, and a tree manager that creates a protection tree is called a protection tree manager.

  A tag group refers to a group of nodes identified using tags and other identifiers, that is, a collection of a plurality of nodes. When a tag group is formed using a VLAN tag as an identifier, the tag group is called a VLAN.

  BPDU (Bridge Protocol Data Unit) refers to control data described in IEEE 802.1D (Prior Art 1) and IEEE 802.1w (Prior Art 2) exchanged for spanning tree generation, and the working system of the present invention. This refers to a control frame including backup system identification information.

  FIG. 5 is a format diagram showing the structure of the Configuration BPDU frame 2205 described in IEEE 802.1D (Prior Art 1) and IEEE 802.1w (Prior Art 2).

  The MAC DA 2201 is an area for storing a destination MAC address.

  The MAC SA 2202 is an area for storing a source MAC address.

  A tag area 2203 is an area into which a tag as an identifier for identifying a plurality of spanning trees is inserted. Although not described in the prior art, the tag is any one of an expansion tag disclosed in Japanese Patent Application No. 2002-204673 and other tags or identification means other than a VLAN tag. One or more combined tags may be used.

  Type 2204 is an area for storing a frame type identifier.

  The BPDU area 2205 is an area for storing information corresponding to Configuration BPDU parameters described in IEEE 802.1D (Prior Art 1) and IEEE 802.1w (Prior Art 2).

  The FCS 2206 is an area for storing a frame check sequence.

  The Protocol Identifier 22051 is an area for storing information equal to the Protocol Identifier described in IEEE 802.1D (Prior Art 1) or IEEE 802.1w (Prior Art 2).

  The Protocol Version Identifier 22052 is an area for storing information equal to the Protocol Version Identifier described in IEEE 802.1D (Prior Art 1) or IEEE 802.1w (Prior Art 2).

  The BPDU Type 22053 is an area for storing information equal to the BPDU Type described in IEEE 802.1D (Prior Art 1) or IEEE 802.1w (Prior Art 2).

  Flags 22054 is an area for storing information equal to Flags described in IEEE 802.1D (conventional technology 1) or IEEE 802.1w (conventional technology 2).

  The Root Identifier 22055 is an area for storing information equal to the Root Identifier described in IEEE 802.1D (Prior Art 1) or IEEE 802.1w (Prior Art 2).

  The Root Path Cost 22056 is an area for storing information equal to the Root Path Cost described in IEEE 802.1D (conventional technology 1) or IEEE 802.1w (conventional technology 2).

  The Bridge Identifier 22057 is an area for storing information equal to the Bridge Identifier described in IEEE 802.1D (Prior Art 1) or IEEE 802.1w (Prior Art 2).

  The Port Identifier 22058 is an area for storing information equal to the Port Identifier described in IEEE 802.1D (conventional technology 1) or IEEE 802.1w (conventional technology 2).

  Message Age 22059 is an area for storing information equal to Message Age described in IEEE 802.1D (conventional technology 1) or IEEE 802.1w (conventional technology 2).

  The MAX Age 2205A is an area for storing information equal to the MAX Age described in IEEE 802.1D (conventional technology 1) or IEEE 802.1w (conventional technology 2).

  The Hello Time 2205B is an area for storing information equal to the Hello Time described in IEEE 802.1D (Prior Art 1) or IEEE 802.1w (Prior Art 2).

  The Forward Delay 2205C is an area for storing information equivalent to the Forward Delay described in IEEE 802.1D (Prior Art 1) or IEEE 802.1w (Prior Art 2).

  FIG. 6 is a format diagram showing the structure of a topology change notification BPDU frame described in IEEE 802.1D (Prior Art 1) and IEEE 802.1w (Prior Art 2).

  The MAC DA 2201 is an area for storing a destination MAC address.

  The MAC SA 2202 is an area for storing a source MAC address.

  The tag area 2203 is an area in which a tag as an identifier for identifying a plurality of spanning trees is inserted, although not described in the prior art. In addition to the VLAN tag, the tag may be a combination tag of any one or more of an expansion tag disclosed in Japanese Patent Application No. 2002-204673 and other tags or identification means.

  Type 2204 is an area for storing a frame type identifier.

  The BPDU area 2205 is an area for storing information corresponding to Topology Change Notification BPDU parameters described in IEEE 802.1D (Prior Art 1) and IEEE 802.1w (Prior Art 2).

  The FCS 2206 is an area for storing a frame check sequence.

  GVRP refers to a control frame that is transmitted and received for managing tag groups, identifying active and standby systems, and exchanging various setting information between nodes.

  In FIG. 53, another frame format of the extension tag frames 3300 and 3400 will be described. Hereinafter, the frame format of the extension tag frames 3500 to 3508 described in FIG. 4 is referred to as an extension tag frame format (1), and the frame format described in FIG. 53 is hereinafter referred to as an extension tag frame format (2). To do.

  The upper part of FIG. 53 shows a detailed frame format of the VLAN tag 3203. A value “0x8100” is set in a TPID (Tag Protocol Identifier) 2800. Note that the value “0x9100” may be used although it is not standard. The TCI 2801 includes a Priority field 2802, a CFI 2803, and a VLAN-ID field 2804.

  The priority field 2802 stores the priority of the frame, and the priority value is defined in IEEE 802.1p. The CFI stores a value indicating the presence / absence of special routing information or the MAC address format type, and the VLAN-ID field 2804 stores a VLAN-ID.

  On the other hand, in the extended tag frame format (2) shown in the lower part of FIG. 50, the CFI 2803 in the TPID 2800 and the TCI 2801 is the same as the VLAN tag 3203, and the Priority field 2802 is set in the Priority / tag Type field 5003. The ID field 2804 is changed to the extension tag information field 5004. The size of the corresponding field is the same.

  In this extended tag frame format (2), the types of extended tags 3500-3508 are stored in the Priority / Tag Type field 5003. A part of the Priority value in the Priority field 2802 (IEEE 802.1p) of the existing VLAN tag 3203 is used as the type of the extension tags 3500-3508 so that IEEE 802.1p can be supported even when the extension tags 3500-2307 are used.

  Specifically, 110, 100, 001, 000 are used for expansion tags 3500-3508, 111 (for reservation), 101 (for conversational multimedia), 011 (for critical application), 010 (standard stream) For IEEE 802.1p.

  Therefore, the number of expansion tags 3500-3508 that can be used is limited to four. For example, the forwarding tag 3500, the broadcast forwarding tag 3508, the customer separation tag 3501, the OAM & P tag 3503 are used, and the correspondence with the priority value is 001 = forwarding tag 3500,000 = broadcast forwarding tag 3508, 110 = customer separation tag 3501, 100 = OAM & P tag 3503. As a result, the four types of extension tags can be identified, and four priorities in IEEE 802.1p can be supported. The selection of the extension tag to be used and the setting of the priority value corresponding to the selection are not limited to this example.

  In the extended tag frame format (2), information such as address information corresponding to the tag type of the extended tags 3500 to 3508 is stored in the extended tag information field 5004. For example, in the case of the forwarding tag 3500, address information of the destination node is stored, in the case of the broadcast forwarding tag 3508, address information of the transmission source node is stored, and in the case of the customer separation tag 3501, customer identification information is stored.

(First embodiment)
Hereinafter, a first embodiment of the present invention will be described in detail with reference to the drawings.

  Referring to FIG. 7, the first embodiment of the present invention includes nodes 11-16, clients 91-96, links 81-86, and links 21-28.

  The node 11 is realized by a program-controlled CPU or the like, and has the following functions.

  (1) The frame arrived from the link 21 or the link 24 is transferred to the link 24 or the link 21.

  (2) A tag necessary for transfer is added to a frame arriving from the link 81, and the frame is transferred to the link 21 or the link 24.

  (3) The frame arrived from the link 21 or the link 24 is transferred to the link 81 after the tags necessary for the transfer are deleted.

  (4) In order to construct a spanning tree, control frames are exchanged with other nodes, and link ports are closed as necessary.

  (5) Monitor the flow rate of the frame flowing through the link.

  Nodes 12 to 16 are nodes similar to the node 11. Hereinafter, the node 11 will be described as a representative of the nodes 11 to 16, but the description of the node 11 can be similarly realized in the other nodes 12 to 16 unless otherwise specified.

  The client 91 is a set of one or more clients, and has a function of transmitting / receiving a frame to / from the node 11 through the link 81.

  The clients 92 to 96 are a client group similar to the client 91. Hereinafter, the client 91 will be described as a representative of the clients 91 to 96, but the description of the client 91 can be similarly applied to the other clients 92 to 96 unless otherwise specified.

  The link 81 is a bidirectional link that connects the client 91 to the node 11 and the node 11 to the client 91.

  The links 82 to 86 are the same links as the link 81. Hereinafter, the link 81 will be used as a representative of the links 81 to 86, but the description of the link 81 can be similarly applied to the other links 82 to 86 unless otherwise specified.

  The link 21 is a bidirectional link that connects the node 11 to the node 12 and the node 12 to the node 11.

  The links 22 to 26 are the same links as the link 21. In the following, description will be made using the link 21 on behalf of the links 22 to 26, but the description about the link 21 can be similarly applied to the other links 22 to 26 unless otherwise specified.

  FIG. 8 is a diagram showing the configuration of the node 11 in detail. The node 11 includes a frame transfer unit 111, a tag insertion unit 112, a tag deletion unit 113, a forwarding table 114, a classifier 1150, a tree manager 1151, a tree manager 1152, a tree selector 116, and a tag table 117. And a setting interface 118.

  The frame transfer unit 111 transfers the frame received from the link 21 or the link 24 and the tag insertion unit 112 to the link 21 or the link 24 and the tag deletion unit 113 or the tree selector 116 according to the description of the forwarding table 114. .

  The tag inserter 112 inserts a tag into the frame received from the link 81 in accordance with the description of the tag table 117 and transfers it to the frame transfer unit 111. Depending on the description of the tag table 117, it is possible to transfer the received frame as it is to the frame transfer unit 111 without inserting a tag, or to insert zero or more tags in the same frame, or It is also possible to copy the arrived frame and insert zero or more plural same or different tags for each copied frame.

  The tag deletion unit 113 removes the tag added to the frame received from the frame transfer unit 111 and transfers it to the link 81. Depending on the setting, the received frame can be transferred to the link 81 as it is without removing the tag.

  The forwarding table 114 responds to an inquiry from the frame transfer unit 111 by sending one or more frame transfer destination ports to the frame transfer unit 111 using one or more combinations as well as a MAC address, a tag, or an input port as a key. To do. The key and the transfer destination port are set by the tree manager 1151 or the tree manager 1152.

  The discriminator 1150 determines an output destination port according to the tag of the received frame, and transfers the frame to the tree manager 1151 or the tree manager 1152. The tree selector 116 can set which tag is added to the frame manager 1151 or the tree manager 1152.

  The tree manager 1151 receives the BPDU from the classifier 1150 using the spanning tree algorithm in accordance with the instruction of the tree selector 116, transmits the BPDU to the frame transfer unit 111, and sets the forwarding table 114. Furthermore, setting information is received from the tree selector 116 and used as a BPDU parameter. Also, control information included in the BPDU is extracted and notified to the tree selector 116.

  The tree manager 1152 is a tree manager similar to the tree manager 1151. Hereinafter, the tree manager 1151 to the tree manager 1152 will be used to describe the tree manager 1151, but the description of the tree manager 1151 can be similarly applied to the tree manager 1152 unless otherwise specified.

  The tree selector 116 receives a setting frame such as GVRP from the frame transfer unit 111, receives control information included in the BPDU from the tree manager 1151 or 1152, receives notification of link information from the resource monitor 119, or sets interface In response to the setting notification from 118, the tree manager 1151, the tree manager 1152, and the tag table 117 are set in accordance with the information included in the setting frame or notification. Also, the setting frame is transmitted to the frame transfer unit 111.

  In response to the inquiry from the tag inserter 112, the tag table 117 responds to the tag inserter 112 with information on the tag to be inserted or a command to transfer without adding a tag. A tag to be inserted or a transfer command without tag insertion is set by the tree selector 116. You can also set to insert zero or more tags in the same frame, or copy the arrived frame, and insert zero or more tags that are the same or different for each copied frame. is there.

  The setting interface 118 transmits a tree selection command, a node deletion request, a link cost, a spanning tree parameter value, and the like from the user to the tree selector 116 through a serial connection, a command line interface such as TELNET, or a WEB server.

  The resource monitor 119 monitors the status of each link port of the node, and transmits a link up notification to the tree selector 116 when link connection is detected. In addition, one or more values of the cumulative number of bytes of frames passing through the link, the number of passing TCP sessions, and the number of HTTP requests are counted and held. In addition, the stored value is reset to 0 by an instruction from the tree selector 116. Further, the passage of a frame of a type designated in advance by the tree selector 116 is monitored, and when the monitored frame passes, the tree selector 116 is notified.

  FIG. 9 is a configuration example of the forwarding table 114 that determines an output port using a tag as a key in FIG. 8 of the present embodiment.

  The tag field 1141 is a field serving as a search index, and it is checked whether or not the information in this field matches the content written in the tag of the received frame.

  The output port 1142 is a field for describing to which port the frame should be transferred when the contents written in the tag of the received frame and the contents of the field 1141 match the tag.

  Note that this embodiment determines not only the tag forwarding that determines the transfer destination port according to the contents of the tag, but also the transfer destination based on the MAC address that has been conventionally used, as shown in this operation example. The same applies to normal MAC address transfer. In this case, a plurality of ports are described in the output port field 1142.

  FIG. 10 is a diagram showing in detail the configuration of the tree manager 1151 in FIG. 8 according to the first embodiment of this invention. The tree manager 1151 includes a tag deletion unit 11511, a BPDU transceiver 11512, a tag insertion unit 11513, a tree controller 11514, and a tree table 11515.

  The tag deletion unit 11511 deletes the tag inserted in the frame input from the classifier 1150 and transfers the tag to the BPDU transceiver 11512. If a tag is not attached to the frame received from the classifier 1150, the received frame is transferred to the BPDU transceiver 11512 as it is.

  The BPDU transmitter / receiver 11512 receives the BPDU from the tag deleter 11511 and notifies the tree controller 11514 of information included in the frame by BPDU reception notification. In addition, upon receiving a BPDU transmission notification from the tree controller 11514, a frame is generated and transmitted to the tag inserter 11513.

  The tag inserter 11513 receives a frame from the BPDU transceiver 11512, inserts a preset tag, and transmits it to the frame transfer unit 111. It is also possible to set the frame to be transferred as it is without inserting a tag.

  The tree controller 11514 has the following four functions.

  (1) Stop operation (initial state): Stops BPDU transmission notification to the BPDU transceiver 11512 according to a stop command from the tree selector 116. Further, the state of the port is registered in the tree table 11515 assuming that all links are down.

  (2) Start operation: In response to a start command from the tree selector 116, a BPDU transmission notification to the BPDU transceiver 11512 is started. Also, an up-state port is registered in the tree table 11515 based on information included in the start command.

  (3) BPDU reception operation: BPDU reception notification is received from the BPDU transceiver 11512 and the tree table 11515 is updated. Further, the identification information of the active system tree and the protection system tree included in the BPDU reception notification is extracted and notified to the main controller 1164 in the tree selector 116.

  (4) Topology update operation: After the stop operation, the start operation, and the BPDU reception operation, the tree table 11515 is referred to according to the spanning tree protocol shown in the prior art 1 or the prior art 2, and if necessary, the tree table Setting of 11515, setting of the forwarding table 114, and transmission of BPDU are performed. The BPDU to be transmitted also includes identification information for the active system tree and the standby system tree. In addition, the tree selector 116 is notified of whether or not the topology is changed as a result of the tree recalculation.

  The tree table 11515 is a table in which parameters relating to port states and node states necessary for the spanning tree protocol shown in the prior art 1 or the prior art 2 are described. The priority of each port or link and the link cost are also described in this table. In the first embodiment, the case where the thickness of the link band is used as the link cost will be described.

  FIG. 11 is a diagram showing in detail the configuration of the tree selector 116 in FIG. 8 according to the first embodiment of this invention. The tree selector 116 includes a tag deleter 1161, a GVRP transceiver 1162, a tag inserter 1163, a main controller 1164, a stability timer 1165, and an arrival interval timer 1166.

  The tag deletion unit 1161 deletes the tag inserted in the frame input from the frame transfer unit 111 and transfers it to the GVRP transceiver 1162. If a tag is not attached to the frame received from the frame transfer unit 111, the received frame is transferred to the GVRP transceiver 1162 as it is.

  The GVRP transceiver 1162 receives the control frame from the tag deleter 1161 and notifies the main controller 1164 of information included in the frame by a GVRP frame reception notification. Further, upon receiving a GVRP frame transmission notification from the main controller 1164, a frame is generated and transmitted to the tag inserter 1163.

  The tag inserter 1163 receives a frame from the GVRP transceiver 1162, inserts a preset tag, and transmits it to the frame transfer unit 111. It is also possible to set the frame to be transferred as it is without inserting a tag.

  The main controller 1164 has the following four functions.

  (1) Link-up detection: The main controller 1164 receives a link-up notification from the resource monitor 119 and notifies a link-up (start command) to the tree manager that is currently in the standby system. The start command stores information on the link that is up. In addition, after the start command is transmitted, the stability timer 1165 is set. When the expiration notification of the stability timer 1165 is received, the tag table 117 is notified of the insertion tag change command, and the GVRP transceiver 1162 is informed of the new tree so as to reverse the registration of the standby system and the active system. Commands the root node to send a tree switch request frame. Further, the stability timer 1165 is set, and the own node is added to the old tree after the timer expires.

  (2) Node deletion request reception: When a node deletion request notification is received from the setting interface 118, a stop command is transmitted to the tree manager that is currently in the standby system. After the stop command is transmitted, the stability timer 1165 is set, and when the expiration notification of the stability timer 1165 is received, the insertion tag change command is notified to the tag table 117, and the registration of the standby system and the active system is reversed. The GVRP transceiver 1162 is instructed to transmit the usage tag group change GVRP frame to the root node of the new tree. Furthermore, the stability timer 1165 is set, and after the timer expires, the setting interface 118 is displayed with permission to delete its own node.

  (3) Usage tag group change GVRP reception: When the local node is the root node of the new tree, when the usage tag group change GVRP frame is received, the active system flag is added to the BPDU transmitted from the local node and transmitted. To the tree manager 1151 or the tree manager 1152. Further, it instructs the GVRP transceiver 1162 to cancel the attachment of the working system flag to the root node of the old tree.

  (4) Reception of working system bit change notification: The tree manager 1151 or the tree manager 1152 confirms the working system flag added to the BPDU when receiving the BPDU, and confirms whether the own group is the working system or the standby system. , Notify the main controller. As a result of receiving the notification, if there is a change between the active system and the standby system, an insertion tag change notification is transmitted to the tag table 117 to reverse the registration of the active system and the standby system.

  (5) Receiving specified frame passing notification: When receiving a notification from the resource monitor 119 that a preset monitoring target frame has passed, a set notification is transmitted to the arrival interval timer 1166. From the arrival interval timer 1166, when the timer expiration notification arrives, it can be seen that the arrival interval of the monitoring target frame is longer than the time set by the set notification. As a result, it is possible to detect that the arrival interval of the BPDU is increased, the arrival interval of the frames flowing through the backup system is increased, and the like.

  The stability timer 1165 transmits a timer expiration notification to the main controller 1164 when a preset time has elapsed from reception of the set notification transmitted from the main controller 1164.

  When the arrival interval timer 1166 receives a set command from the main controller 1164, it resets the currently held time to 0 and activates the timer, and when the time specified by the set command has elapsed, Send timer expiration notification.

  FIG. 12 is a flowchart showing in detail the state transition of the main controller 1164 in FIG. 11 according to the first embodiment of the present invention.

  After that, out of the two spanning trees, the spanning tree used for transferring data frames entering the network is the working tree or working tree, and the non-working tree is the spare tree or spare tree. It expresses.

  In addition, a tree manager that generates a working tree is called a working tree manager, and a tree manager that creates a protection tree is called a protection tree manager.

  The state 11641 is a state in which it is not possible to determine whether the working tree manager is the tree manager 1151 or the tree manager 1152, and the BPDU transceiver 11512 in the tree manager 1151 and the BPDU in the tree manager 1152 The BPDU transmission function of the transceiver 11522 is disabled, and only the BPDU reception function is enabled.

  In the state 11642, the active tree manager is the tree manager 1151, the standby tree manager is the tree manager 1152, the BPDU transmission function of the BPDU transceiver 11512 in the tree manager 1151 is disabled, and the The BPDU transmission function of the BPDU transceiver 11522 is also disabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In state 11643, the active tree manager is the tree manager 1151, the standby tree manager is the tree manager 1152, the BPDU transmission function of the BPDU transceiver 11512 in the tree manager 1151 is disabled, and the The BPDU transmission function of the BPDU transceiver 11522 is enabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In the state 11644, the active tree manager is the tree manager 1152, the standby tree manager is the tree manager 1151, the BPDU transmission function of the BPDU transceiver 11522 in the tree manager 1152 is enabled, and the The BPDU transmission function of the BPDU transceiver 11512 is disabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In the state 11645, the active tree manager is the tree manager 1152, the standby tree manager is the tree manager 1151, the BPDU transmission function of the BPDU transceiver 11522 in the tree manager 1152 is enabled, and the tree manager 1151 The BPDU transmission function of the BPDU transceiver 11512 is also enabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In the state 11646, the active tree manager is the tree manager 1151, the standby tree manager is the tree manager 1152, the BPDU transmission function of the BPDU transceiver 11512 in the tree manager 1151 is enabled, and the The BPDU transmission function of the BPDU transceiver 11522 is also enabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In the state 11647, the active tree manager is the tree manager 1151, the standby tree manager is the tree manager 1152, the BPDU transmission function of the BPDU transceiver 11512 in the tree manager 1151 is enabled, and the tree manager 1152 In this state, the BPDU transmission function of the BPDU transceiver 11522 is disabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In state 11648, the active tree manager is the tree manager 1152, the standby tree manager is the tree manager 1151, the BPDU transmission function of the BPDU transceiver 11522 in the tree manager 1152 is disabled, and the This is a state where the BPDU transmission function of the BPDU transceiver 11512 is enabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In the state 11649, the active tree manager is the tree manager 1152, the standby tree manager is the tree manager 1151, the BPDU transmission function of the BPDU transceiver 11522 in the tree manager 1152 is disabled, and the This is a state where the BPDU transmission function of the BPDU transceiver 11512 is disabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  Next, the operation of the main controller 1164 will be described with reference to FIG.

  When the main controller 1164 receives a notification from the resource monitor 119 that a new connection has been made to the network, the main controller 1164 waits for the arrival of an active notification from the tree manager 1151 or the tree manager 1152. When the active system notification included in the BPDU is received from the tree manager 1151 or the tree manager 1152, the tree manager 1151 or the tree manager 1152 specified in the notification is used as the active, and the tree manager 1151 or 1152 specified for the standby is used as the standby. Set and transition to state 11642 or state 11649. Here, the case where the state transitions to the state 11642 will be described as an example, but the following description is the same when the state transitions to the state 11649. (State 11641)

  The main controller 1164 sets the tree manager 1151 as active and the tree manager 1152 as backup. Further, a BPDU transmission stop command is issued to each of the tree managers 1151 and 1152. (State 11642)

  If the main controller 1164 receives a node addition request from the setting interface 118 in the state 11642, the main controller 1164 transitions to the state 11443. If the active system notification is received from the tree manager 1151 or the tree manager 1152 and the relationship between the active system and the standby system has changed, the state 11649 is entered. (State 11642)

  The main controller 1164 notifies the tree manager 1152 of the link up, and at the same time, permits the tree manager 1152 to transmit BPDUs. Further, the stability timer 1165 is activated. (State 11643)

  When the main controller 1164 receives the timer expiration notification from the stability timer 1165, the main controller 1164 replaces the tree manager 1151 registered for the current use with the tree manager 1152 registered for the backup, and newly uses the tree manager 1152 for the current use and the spare for the spare. The tree manager 1151 is assumed. Further, the usage tag group change notification is transmitted to the root node of the new tree through the GVRP transceiver 1162. The contents of the usage tag group change notification are reflected in the BPDU and transmitted to all nodes. Thereafter, the stability timer 1165 is activated. (State 11644)

  When the main controller 1164 receives the timer expiration notification from the stability timer 1165, the main controller 1164 notifies the tree manager 1151 of the link up, and at the same time, permits the tree manager 1152 to transmit the BPDU. Usually, it is stable in this state. (State 11645)

  In the state 11645, the main controller 1164 receives the active system notification contained in the BPDU from the tree manager 1151 or the tree manager 1152, and transitions to the state 11646 if the relationship between the active system and the standby system has changed. . Then, the tree manager 1152 registered for the current use is replaced with the tree manager 1151 registered for the backup, and the new work is set as the tree manager 1151 and the backup is set as the tree manager 1152. (State 11645)

  If the main controller 1164 receives a node deletion request from the setting interface 118 in the state 11645, the main controller 1164 transitions to the state 11644. (State 11645)

  The main controller 1164 notifies all connected link links down to the tree manager 1151, and simultaneously issues a BPDU transmission stop command to the tree manager 1151. Further, the stability timer 1165 is activated. (State 11644)

  When the main controller 1164 receives the timer expiration notification from the stability timer 1165, the main controller 1164 replaces the tree manager 1152 registered for the current use with the tree manager 1151 registered for the backup, and newly uses the tree manager 1151 for the current use and the spare for the spare use. The tree manager 1152 is assumed. Further, the usage tag group change notification is transmitted to the root node of the new tree through the GVRP transceiver 1162. The contents of the usage tag group change notification are reflected in the BPDU and transmitted to all nodes. Thereafter, the stability timer 1165 is activated. (State 11643)

When the main controller 1164 receives a timer expiration notification from the stability timer 1165, the main controller 1164 notifies the tree manager 1151 of link down, and simultaneously issues a BPDU reception stop command to the tree manager 1151. Furthermore, the state transits to the state 11641 unconditionally and waits until the node is disconnected. (State 11642)
FIG. 13 is a configuration example of the tag table 117 that determines a tag to be inserted using the destination MAC address as a key in FIG. 8 of the present embodiment.

  The destination MAC address 1171 is a field that becomes an index for search, and it is checked whether the information in this field matches the contents written in the destination MAC address field of the received frame, that is, the MAC DA field. In this case, the tag described in the insertion tag field 1172 is inserted into the received frame.

  The insertion tag field 1172 is a field in which a tag to be inserted with respect to the destination MAC address field 1171 is described. In the present embodiment, a tag of a tag group that is currently active is inserted. The insertion tag field 1172 is rewritten by the tree selector 116 to a tag that is currently used.

  With reference to FIG. 8, FIG. 14, FIG. 15 and FIG. 16, the operation when adding the node 17 in this embodiment will be described in detail using a specific example.

  The two spanning trees in the initial state (the state before the node 17 is connected to the links 29 and 30) have the same connection relationship. This is because the spanning tree setting protocol sets the spanning tree based on information such as node and link priority values, so if two spanning trees are set for the same network, the result is the same connection relationship. is there. In this initial state, one of the spanning trees is set as active and the other is set as spare (specifically, each of the tree managers 1151 and 1152 rewrites the tree table 11515 and the tree selector 116 rewrites the tag table 117) This network is operated using the current spanning tree.

  In the network shown in FIG. 10, it is assumed that two spanning trees 51 indicated by bold lines are set in the initial state.

  Referring to FIG. 14, this operation example includes nodes 11 to 17, links 21 to 30, and a tree 51. However, the node 17, the link 29, and the link 30 are not connected in the initial state.

  FIG. 15 shows the state of the spanning tree 52 after the node 17 is added in this operation example. The tree 52 is indicated by a bold line in FIG.

  Two tag groups to which all nodes and all ports of the nodes 11 to 16 belong are already set, and the first tag group is called a tag group 41 and the second tag group is called a tag group 42.

  Basically, all nodes and all ports are joined to two tag groups, but a tag group composed of only some ports or nodes may be created. In the following description, it is assumed that all nodes and all ports join two tag groups.

  The nodes 11 to 16 have two spanning tree circuits that operate independently. The spanning tree that operates on the tag group 41 is referred to as a tree 51, and the spanning tree that operates on the tag group 42 is referred to as a tree 52.

  Although two spanning trees are always created, it is not always necessary to create two tag groups. Only the tag group 41 is set and the tag group 42 is not used. A spanning tree operating on the tag group 41 can be called a tree 51, and a spanning tree operating without belonging to the tag group can be called a tree 52. On the other hand, only the tag group 42 is set, the tag group 41 is not used, the spanning tree that operates without belonging to the tag group is called the tree 51, and the spanning tree that operates on the tag group 42 is called the tree 52. it can.

  Here, the case where both the tag group 41 and the tag group 42 are used will be described, but the operation when both the tag group 41 and the tag group 42 are used is only the tag group 41 or only the tag group 42. The same applies to the case where is used.

  In this network, the initial setting from the setting interface 118 is set so that the tree 51 becomes an active spanning tree and the tree 52 becomes a standby spanning tree. The BPDU of the tree 51 has an active flag and The tag of the tag group 41 is attached, and the BPDU of the tree 52 is attached with the reserve system flag and the tag of the tag group 42.

  The spanning of the active system, which is the tree 51, is achieved by all the nodes transmitting each other's BPDU frame with the active system flag or the standby system flag to each other at a fixed period defined by IEEE Std 802.1D or IEEE Std 802.1w. A spare spanning tree that is a tree or tree 52 is constructed.

  The working system flag or the protection system flag can be expressed by using fields such as a tag area 2203, a Type 2204, and a BPDU Type 22053 among the fields of the BPDU frame shown in FIG.

  At present, a sufficient amount of time has passed since the start of the network, and each of the trees 51 and 52 is a node that is the result of a sufficient exchange of BPDU frames to which the tags of the tag group 41 and the tag group 42 are added. Assume that 11 is stable as a root node.

  Stable means a state in which the tree structure of the spanning tree has not changed for a sufficiently long time.

  The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52. Specifically, a value indicating that this BPDU belongs to the tag group 41 is described in the tag area 2203 of the BPDU transmitted using the tree 51, and the tag area 2203 of the BPDU transmitted using the tree 52 is used. Describes a value indicating that this BPDU belongs to the tag group 42.

  At this time, since the tree 51 is set as the active system, the tag of the tag group 41 is added to the data transmitted from the client to the nodes 11 to 16 by the tag inserter 112. Specifically, a value indicating belonging to the tag group 41 is described in the tag area of the data signal. The data to which the tag is added is transferred by the frame transfer unit 111 along the tree 51 that is currently set as the active system.

  When node 17 is connected to link 29 and link 30, node 17 does not join any tag group and starts receiving BPDUs. (The state of the main controller 1164 of the node 17 at this time is the state 11641 of FIG. 12)

  When the node 17 receives the BPDU of each tag group, the node 17 confirms that the current active system is the tag group 41 and the current standby system is the tag group 42. (The state of the main controller 1164 of the node 17 at this time is the state 11642 of FIG. 12). Then, the own node is allowed to participate only in the tag group 42 to perform transmission / reception of BPDU, and the tag group 41 only receives BPDU. Set to not send. (The state of the main controller 1164 of the node 17 at this time is the state 11643 of FIG. 12)

  When the node 17 is added, a change occurs in the members of the tag group 42. Therefore, an operation of updating the tree 52 is started by the spanning tree protocol. That is, when the node 17 transmits a BPDU and is received by an adjacent node, the adjacent node recognizes that the topology status has changed, and starts an update operation of the tree 52. Since there is no change in the members of the tag group 41, the tree 51 is not updated.

  The frame transmitted from the client is added with the tag of the tag group 41 by each of the nodes 11 to 16 and continues to be transferred along the tree 51.

  Here, it is assumed that the tree 52 is stabilized with the root node as the node 11 by the update operation of the tree 52 described in the paragraph 0135. The structure of the tree 52 at this time is shown in FIG.

  The node 17 determines that the tree 52 has become stable after a lapse of a certain period of time after being connected to the network, transmits a usage tag group change notification to the node 11 that is the root node of the tree 52, and the tree 52 is used as a standby system. Command to transition from active to active. For this command, for example, a control frame (GVRP) is used. (The state of the main controller 1164 of the node 17 at this time is the state 11644 of FIG. 12).

  Note that the detection that the tree 52 is stable is detected when a certain time elapses after the node 17 is connected to the network, and is also detected when the BPDU arrival interval of the tree 52 at the node 17 exceeds a certain time. You can also.

  The node 11 that has received the usage tag group change notification transmits a usage tag group change notification to the node 11 that is the root node of the tree 51 and instructs the node 51 to transition to the standby system. For this usage tag group change notification, for example, a control frame (GVRP) is used. Then, the tag of the tag group 42 is attached, and an active system flag is set in the BPDU transmitted for the tree 52. The BPDU propagates to all nodes while being transferred at each node.

  The node 11 of the tree 51 receives the usage tag group change notification, transitions the tree 51 to the standby system, attaches the tag group 41 flag, and sets the standby system flag in the BPDU transmitted for the tree 51. This backup system flag is set, for example, in a predetermined field in the BPDU of FIG. The setting is executed when the tree managers 1151 and 1152 rewrite the tree table 11515 and the tree selector 116 rewrites the tag table 117. The BPDU propagates to all nodes while being transferred at each node.

  The nodes 11 to 17 confirm that the working flag is added to the BPDU with the tag of the tag group 42, and switch the tag to be added to the frame transferred from the client from the tag group 41 to the tag group 42. . At this time, the insertion tag field 1172 in the tag table 117 is rewritten. The frame to which this tag is added is transferred along the tree 52.

  After a while after the above switching is completed, no frames flow through the tree 51.

The node 17 determines that there is no longer a node to which the tag of the tag group 41 is added after a lapse of a certain time from the transmission of the usage tag group change notification, and joins the own node to the tag group 41 to prepare for the next topology change. . In order to join the own node to the tag group 41, the node 17 permits BPDU transmission to the tree controller in the tree manager 1151 and permits BPDU transmission from the BPDU transceiver 11512. (The state of the main controller 1164 of the node 17 at this time is the state 11645 of FIG. 12)
At this time, since the reconfiguration is performed on the spanning tree 51, when the network is viewed focusing on the spanning tree 51, the network stops. However, since communication in the network is performed using the spanning tree 52 during this time, problems such as congestion and frame arrival delay do not occur with the addition of the node 17.

  The operation of joining the node 17 to the tag group 41 may be performed by the node 11 that is the root node of the tree 52 or the node 11 that is the root node of the tree 51.

  When the node 11, which is the root node of the tree 52, joins the node 17 to the tag group 41, the tag of the tag group 41 is added after a certain period of time has elapsed since the node 11 received the usage tag group change notification from the node 17. It is determined that there is no node to be added, and a GVRP frame is transmitted to the node 17 to instruct to join the tag group 41.

  When the node 11, which is the root node of the tree 51, joins the node 17 to the tag group 41, the tag of the tag group 41 is added after a certain period of time has elapsed since the node 11 received the usage tag group change notification from the node 11. It is determined that there is no node to be added, and a GVRP frame is transmitted to the node 17 to instruct to join the tag group 41. In this case, the node 11 of the tree 51 transmits a GVRP frame directly to the node 17 (not via the node 11 of the tree 52). This is because in the operation of joining the node 17 to the tag group 41, the flag inserted into the BPDU does not change, so there is no need to go through the root node.

  As described above, the node 17 can be added without stopping the network. Thereafter, the same operation is repeated when adding a node. However, the tag group 41 and the tag group 42 in the above description are appropriately switched.

  FIG. 16 is a sequence diagram showing the addition operation of the node 17 described above.

  An arrow 31 indicates the flow of the BPDU with active system flag into which the tag indicating the tag group 41 is inserted.

  An arrow 32 indicates the flow of the BPDU with the backup system flag into which the tag indicating the tag group 42 is inserted.

  An arrow 33 indicates the flow of the BPDU with the active system flag into which the tag indicating the tag group 42 is inserted.

  An arrow 34 indicates the flow of a BPDU with a backup flag, in which a tag indicating the tag group 41 is inserted.

  An arrow 35 indicates a flow of a usage tag group change notification by a GVRP frame or the like in which a tag indicating a tag group is not inserted.

  Next, with reference to FIG. 15 and FIG. 14, the operation when the node 17 is deleted in the present embodiment will be described in detail using a specific example.

  Referring to FIGS. 15 and 14, this operation example includes nodes 11 to 17 and links 21 to 30.

  Two tag groups to which all the ports of the nodes 11 to 17 belong are already set. The first tag group is called a tag group 41 and the second tag group is called a tag group 42.

  The nodes 11 to 17 have two spanning tree paths that operate independently. The spanning tree that operates on the tag group 41 is referred to as a tree 51, and the spanning tree that operates on the tag group 42 is referred to as a tree 52.

  In FIG. 15, it is assumed that the tree 52 is represented by a thick line and is stable with the node 11 as a root node.

  The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52.

  Currently, the tag 52 of the tag group 42 is added to the data transmitted from the client to the nodes 11 to 17 because the tree 52 is the active system. Data to which the tag is added is transferred along the tree 52.

  It is assumed that the node 17 has received the BPDU of each tag group and has already confirmed that the current active system is the tag group 42 and the current standby system is the tag group 41.

  When the node 17 receives a deletion request through the setting interface or other means, the node 17 sets the node 17 to participate only in the active tag group 42 and not to participate in the tag group 41. At this time, the node 17 stops the BPDU transmission of the tag group 41.

  With this setting, it is recognized that the node 17 has been deleted by no longer receiving the BPDU in the adjacent node of the node 17, and a change occurs in the member of the tag group 41, so that an operation of updating the tree 51 is started. . Since there is no change in the members of the tag group 42, the tree 52 is not updated.

  The frames from the client continue to be transferred along the tree 52 with the tags of the tag group 42 added at the nodes 11 to 17 as before.

  Here, the tree 51 shows a stable state in which the root node is the node 11 and the node 17 is not joined. The configuration of the tree 51 is shown in FIG.

  Here, the term “stable” refers to a state in which the tree structure of the spanning tree has not changed for a sufficiently long time.

  The node 17 determines that the tree 51 has become stable after a certain period of time has elapsed since the setting of non-participation in the tag group 41, transmits a usage tag group change notification to the node 11, which is the root node of the tree 51, and reserves the tree 51. Command to transition from the system to the working system. (The state of the main controller 1164 of the node 17 at this time is the state 11644 of FIG. 12).

  The node 11 that has received the usage tag group change notification transmits a usage tag group change notification to the node 11 that is the root node of the tree 52 and instructs the node 52 to transition to the standby system. Then, the flag of the tag group 41 is attached, and the working system flag is set in the BPDU transmitted for the tree 51. The BPDU propagates to all nodes while being transferred at each node.

  The node 11 of the tree 51 receives the usage tag group change notification, shifts the tree 52 to the standby system, attaches the flag of the tag group 42, and sets the standby system flag in the BPDU transmitted for the tree 52. The BPDU propagates to all nodes while being transferred at each node.

  The nodes 11 to 17 confirm that the working flag is added to the BPDU with the tag of the tag group 41, and switch the tag to be added to the frame transferred from the client from the tag group 42 to the tag group 41. . The frame to which this tag is added is transferred along the tree 51.

  After a while after the above switching is completed, no frames flow through the tree 52.

  The node 17 determines that there is no longer a node to which the tag of the tag group 42 is added after a lapse of a certain time from the transmission of the usage tag group change notification, and notifies the node 17 that the own node may be deleted from the network. Is output to the setting interface 118. (The state of the main controller 1164 of the node 17 at this time is the state 11643 of FIG. 12)

  Thereafter, the same operation is repeated when deleting a node. However, the tag group 41 and the tag group 42 in the above description are appropriately switched.

  Referring to FIG. 14 and FIG. 15, in this embodiment, the operation when adding node 17 in the case where a tag is added only to BPDU and no tag is added to data will be described in detail using a specific example. Explained.

  In FIG. 14, it is assumed that the tree 51 is represented by a bold line and is stable with the node 11 as a root node. Stable means a state in which the tree structure of the spanning tree has not changed for a sufficiently long time.

  The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52.

  Currently, no tag is added to the data transmitted from the client to the nodes 11 to 16. Data is transferred along the tree 51 according to the setting in the forwarding table of each node.

  When node 17 is connected to link 29 and link 30, node 17 does not join any tag group and starts receiving BPDUs. (The state of the main controller 1164 of the node 17 at this time is the state 11641 of FIG. 12)

  When the node 17 receives the BPDU of each tag group, the node 17 confirms that the current active system is the tag group 41 and the current standby system is the tag group 42 based on the flag in the tag added to the BPDU. Then, BPDU transmission / reception is performed only in the tag group 42, and in the tag group 41, BPDU is only received and transmission is not performed. (The state of the main controller 1164 of the node 17 at this time is the state 11642 of FIG. 12)

  When the node 17 is added, a change occurs in the members of the tag group 42. Therefore, an operation of updating the tree 52 is started by the spanning tree protocol. Since there is no change in the members of the tag group 41, the tree 51 is not updated. (The state of the main controller 1164 of the node 17 at this time is the state 11643 of FIG. 12)

  A frame transmitted from the client is not added with a tag as before, and continues to be transferred along the tree 51.

  Here, it is assumed that the tree 52 is stable with the root node as the node 11. The structure of the tree 52 is shown in FIG.

  The node 17 determines that the tree 52 has become stable after a lapse of a certain period of time after being connected to the network, transmits a usage tag group change notification to the node 11 that is the root node of the tree 52, and the tree 52 is used as a standby system. Command to transition from active to active. (The state of the main controller 1164 of the node 17 at this time is the state 11644 of FIG. 12).

  The node 11 of the tree 52 that has received the usage tag group change notification transmits the usage tag group change notification to the node 11 that is the root node of the tree 51 and instructs the node 51 to transition to the standby system. Then, the flag of the tag group 42 is attached, and the working system flag is set in the BPDU transmitted for the tree 52. The BPDU propagates to all nodes while being transferred at each node.

  The node 11 of the tree 51 receives the usage tag group change notification, transitions the tree 51 to the standby system, attaches the tag group 41 flag, and sets the standby system flag in the BPDU transmitted for the tree 51. The BPDU propagates to all nodes while being transferred at each node.

  The nodes 11 to 17 confirm that the working flag is added to the BPDU with the tag of the tag group 42, and change the routing table to the setting according to the tree 52. As a result, the frame transferred in the node is transferred along the tree 52.

  When the routing tables of all nodes are switched to those using the tree 52, there are no frames flowing through the tree 51.

  The node 17 determines that there is no longer a node for setting the table according to the tag group 41 after a certain period of time has elapsed from the transmission of the usage tag group change notification, and joins the own node to the tag group 41 to change the next topology. Prepare for. (The state of the main controller 1164 of the node 17 at this time is the state 11644 of FIG. 12).

  As described above, the node 17 can be added without stopping the network. Thereafter, the same operation is repeated when adding a node. However, the tag group 41 and the tag group 42 in the above description are appropriately switched.

  In the above description, the stable state of the tree is confirmed using a timer. However, the method of confirming the stable state is not limited to this, and as shown below, the arrival interval of BPDUs or frames is measured, thereby It is also possible to check the stable state.

  In FIG. 14, it is assumed that the tree 51 is represented by a bold line and is stable with the node 11 as a root node.

  The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52.

  Currently, a tag of the tag group 41 is added to data transmitted from the client to the nodes 11 to 16. Data to which this tag is added is transferred along the tree 51.

  When node 17 is connected to link 29 and link 30, node 17 does not join any tag group and starts receiving BPDUs.

  When the node 17 receives the BPDU of each tag group, the node 17 confirms that the current active system is the tag group 41 and the current standby system is the tag group 42. Then, the own node is added only to the tag group 42 to perform transmission / reception of BPDU, and the tag group 41 is set to perform only reception of BPDU and not to transmit. Then, a frame notifying that the node 17 has been added is transmitted to each of the node 11 that is the root node of the tag group 42 and the node 11 that is the root node of the tag group 41.

  When the node 17 is added, a change occurs in the members of the tag group 42. Therefore, an operation of updating the tree 52 is started by the spanning tree protocol. Since there is no change in the members of the tag group 41, the tree 51 is not updated.

  The tag transmitted from the client is added with the tag of the tag group 41 as before, and continues to be transferred along the tree 51.

  Here, it is assumed that the tree 52 is stable with the root node as the node 11. The structure of the tree 52 is shown in FIG.

  When the node 11 that is the root node of the tree 52 detects that the BPDU arrival interval of the tree 52 is equal to or longer than a certain time, the node 11 determines that the tree 52 is stable and uses the node 11 that is the root node of the tree 51. A tag group change notification is transmitted, and an instruction is given to transition the tree 51 to the standby system. Then, the flag of the tag group 42 is attached, and the working system flag is set in the BPDU transmitted for the tree 52. The BPDU propagates to all nodes while being transferred at each node.

  The stability detection of the tree 52 may be performed by the node 11 that is the root node of the tree 51. In this case, when the node 11 that is the root node of the tree 51 detects that the BPDU arrival interval of the tree 52 is equal to or longer than a predetermined time, the node 11 determines that the tree 52 is stable, and the node 11 that is the root node of the tree 52 A usage tag group change notification is transmitted to the server, and an instruction is given to transition the tree 52 to the standby system. The node 11 which is the root node of the tree 51 attaches the flag of the tag group 42 and sets the working system flag in the BPDU transmitted for the tree 52. The BPDU propagates to all nodes while being transferred at each node.

  The node 11 receives the usage tag group change notification, transitions the tree 51 to the standby system, attaches the flag of the tag group 41, and sets the standby system flag in the BPDU transmitted for the tree 51. The BPDU propagates to all nodes while being transferred at each node.

  The nodes 11 to 17 confirm that the working flag is added to the BPDU with the tag of the tag group 42, and switch the tag to be added to the frame transferred from the client from the tag group 41 to the tag group 42. . The frame to which this tag is added is transferred along the tree 52.

  After a while after the above switching is completed, no frames flow through the tree 51.

  The node 11 that is the root node of the tag group 41 adds that the tag of the tag group 41 is added, and when the arrival interval of frames flowing through the tree 51 exceeds a certain time, there is no node to which the tag of the tag group 41 is added. Judgment is made and a GVRP frame is transmitted to the node 17 to instruct to join the tag group 41 to prepare for the next topology change.

  The operation of joining the node 17 to the tag group 41 may be performed by the node 11 that is the root node of the tree 52.

  When the node 11 which is the root node of the tree 52 joins the node 17 to the tag group 41, the tag group 41 is added when the arrival interval of frames flowing through the tree 51 with the tag of the tag group 41 added is longer than a certain time. It is determined that there is no longer a node to which the tag is added, and a GVRP frame is transmitted to the node 17 to instruct to join the tag group 41 to prepare for the next topology change.

  As described above, the node 17 can be added without stopping the network. Thereafter, the same operation is repeated when adding a node. However, the tag group 41 and the tag group 42 in the above description are appropriately switched.

  Referring to FIGS. 14 and 15, in this embodiment, the operation when adding node 17 when detecting the transition to the standby system by receiving the notification of the completion of the transition to the standby system is received. This will be described in detail using a specific example.

  In FIG. 14, it is assumed that the tree 51 is represented by a bold line and is stable with the node 11 as a root node.

  The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52.

  Currently, a tag of the tag group 41 is added to data transmitted from the client to the nodes 11 to 16. Data to which this tag is added is transferred along the tree 51.

  When node 17 is connected to link 29 and link 30, node 17 does not join any tag group and starts receiving BPDUs.

  When the node 17 receives the BPDU of each tag group, the node 17 confirms that the current active system is the tag group 41 and the current standby system is the tag group 42. Then, the own node transmits / receives BPDU only to / from the tag group 42, and the tag group 41 is set so as to only receive BPDU and not transmit.

  When the node 17 is added, a change occurs in the members of the tag group 42. Therefore, an operation of updating the tree 52 is started by the spanning tree protocol. Since there is no change in the members of the tag group 41, the tree 51 is not updated.

  The tag transmitted from the client is added with the tag of the tag group 41 as before, and continues to be transferred along the tree 51.

  Here, it is assumed that the tree 52 is stable with the root node as the node 11. The structure of the tree 52 is shown in FIG.

  The node 17 determines that the tree 52 has become stable after a lapse of a certain period of time after being connected to the network, transmits a usage tag group change notification to the node 11 that is the root node of the tree 52, and the tree 52 is used as a standby system. Command to transition from active to active.

  The node 11 that has received the usage tag group change notification transmits a usage tag group change notification to the node 11 that is the root node of the tree 51 and instructs the node 51 to transition to the standby system. Then, the flag of the tag group 42 is attached, and the working system flag is set in the BPDU transmitted for the tree 52. The BPDU propagates to all nodes while being transferred at each node.

  The node 11 receives the usage tag group change notification, transitions the tree 51 to the standby system, attaches the flag of the tag group 41, and sets the standby system flag in the BPDU transmitted for the tree 51. The BPDU propagates to all nodes while being transferred at each node.

  The nodes 11 to 17 confirm that the working flag is added to the BPDU with the tag of the tag group 42, and switch the tag to be added to the frame transferred from the client from the tag group 41 to the tag group 42. Further, a switch completion notification is transmitted to the node 11 that is the root node of the tree 52.

  After a while after the above switching is completed, no frames flow through the tree 51.

  When the node 11 receives the switching completion notification from all the nodes 11 to 17, the node 11 determines that there is no node to which the tag of the tag group 41 is added, and transmits a GVRP frame to the node 17. Command 41 to join.

  The operation of joining the node 17 to the tag group 41 may be performed by the node 17 that is a newly added node or the node 11 that is the root node of the tree 51.

  When the node 17 which is a newly added node joins the node 17 itself to the tag group 41, the nodes 11 to 16 add the active flag to the BPDU with the tag of the tag group 42. When the tag to be confirmed and added to the frame transferred from the client is switched from the tag group 41 to the tag group 42, a switching completion notification is transmitted to the node 17 which is a newly added node. However, when the switching completion notification is received from all the nodes 11 to 16, it is determined that there is no node to which the tag of the tag group 41 is added, and the node 17 itself joins the tag group 41.

  When the node 11 as the root node of the tree 51 joins the node 17 to the tag group 41, the nodes 11 to 17 confirm that the working system flag is added to the BPDU with the tag of the tag group 42. When the tag added to the frame transferred from the client is switched from the tag group 41 to the tag group 42, a switch completion notification is transmitted to the node 11 which is the root node of the tree 51, and the node 11 When the switching completion notification is received from all the nodes 11 to 17, it is determined that there is no node to which the tag of the tag group 41 is added, and a GVRP frame is transmitted to the node 17 to join the tag group 41. Command.

  As described above, the node 17 can be added without stopping the network. Thereafter, the same operation is repeated when adding a node. However, the tag group 41 and the tag group 42 in the above description are appropriately switched.

  Next, the effect of this embodiment will be described.

  Conventionally, when a node belonging to a spanning tree is added or deleted, the whole or a part of the data frame is stopped and the spanning tree is reconstructed, so that the network may be stopped during the reconstruction.

  In this embodiment, the spanning tree including the newly added node is generated while the spanning tree before the configuration change is operated, and the spanning tree used after the new spanning tree is stabilized can be switched without stopping the network. Spanning tree reconfiguration such as addition and deletion of nodes belonging to the spanning tree is possible.

  As a result, it is possible to reduce the possibility of congestion.

(Second Embodiment)
Hereinafter, a second embodiment of the present invention will be described in detail with reference to the drawings.

  The second embodiment of the present invention uses, in the first embodiment, the capacity of free bandwidth or the number of passing TCP flows, the number of HTTP requests, etc., instead of the thickness of the link bandwidth when calculating the cost, Further, when the cost is changed, the difference between the active system and the standby system is the same as in the case of adding / deleting a node. In the following, the case where the capacity of the free bandwidth is used as the cost will be described, but the number of passing TCP flows and the number of HTTP requests can be similarly realized unless otherwise specified.

  In IEEE 802.1D and IEEE 802.1w, the cost of the link is determined by the reciprocal of the thickness of the link bandwidth. That is, the cost cannot be changed dynamically according to the load.

  In the present embodiment, the cost of the link is determined by the reciprocal of the free bandwidth of the link, thereby dynamically changing the cost according to the load.

  FIG. 17 shows the configuration of the tree selector 116 in the second embodiment. Referring to FIG. 17, in the second embodiment of the present invention, a cost reference timer 1167, a function calculator 1168, and a smoothing circuit 1169 are added to the tree selector of FIG. 11 in the first embodiment. Are different.

  When the main controller 1164α receives the expiration notification from the cost reference timer 1167 in addition to the operation in the first implementation, the flow rate information of the frame that has flowed through the link after the previous expiration of the cost reference timer or the number of TCP flows from the resource monitor An operation of acquiring the number of HTTP requests, calculating the cost based on the flow rate, the number of flows, or the number of requests, and notifying a tree manager registered for backup (hereinafter referred to as a backup system tree manager) is performed. In the case of the flow rate, the free bandwidth of the link is obtained from the flow rate and the thickness of the link bandwidth, and the reciprocal of the free bandwidth of the ring is used as the cost. In the case of the number of TCP flows and the number of HTTP requests, take the difference between the maximum allowable flow number and the maximum allowable request number set in advance and the number of TCP flows or requests that actually pass through the link, and take the reciprocal of this difference. Use as cost.

  After calculating the cost by the above means, the main controller 1164α passes the cost to the function calculator 1168 for evaluation, and further passes the evaluation result of the function calculator to the smoothing circuit 1169 to smooth the smoothed result. The value is sent to the standby tree manager.

  The function calculator 1168 uses the cost value input from the main controller 1164α as a parameter, determines an output cost value using an arbitrary function such as a proportional function, a hysteresis function, and a staircase function. The function of the function calculator 1168 returned to the controller 1164 can prevent state transition vibration. This is to suppress a rapid fluctuation of the cost value and smoothly change the cost.

  The smoothing circuit 1169 uses a low-pass filter or the like to smooth the previous input parameter stored in advance and the input parameter newly passed from the main controller 1164α, and notifies the result to the main controller 1164α. By the function of the smoothing circuit 1169, it is possible to prevent rapid fluctuations in cost and vibration of state transition.

  The cost reference timer 1167 transmits a timer expiration notification to the main controller 1164α when a preset time has elapsed since the reception of the set notification transmitted from the main controller 1164α.

  FIG. 18 is a flowchart showing in detail the state transition of the main controller 1164A in FIG. 17 according to the second embodiment of the present invention. Referring to FIG. 18, the second embodiment of the present invention is different in that a state 1164A and a state 1164B are added to FIG. 12 in the first embodiment.

  In state 1164A, the active tree manager is the tree manager 1152, the standby tree manager is the tree manager 1151, the BPDU transmission function of the BPDU transceiver 11522 in the tree manager 1152 is enabled, and the The BPDU transmission function of the BPDU transceiver 11512 is also enabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  In state 1164B, the active tree manager is the tree manager 1151, the standby tree manager is the tree manager 1152, the BPDU transmission function of the BPDU transceiver 11512 in the tree manager 1151 is enabled, and the The BPDU transmission function of the BPDU transceiver 11522 is also enabled. Note that the BPDU reception functions of the BPDU transceiver 11512 and the BPDU transceiver 11522 are always valid regardless of whether the transmission function is valid or invalid.

  Hereinafter, with reference to FIG. 18, the flow of cost calculation is shown with the state 11645 as the base point, but this operation can be similarly applied even when the state 11646 is the base point.

When the main controller 1164α transitions to the state 11645 and receives an instruction to use dynamic cost calculation from the setting interface 118 or the GVRP transceiver 1162, the main controller 1164α sets the cost reference timer 1167. (State 11645)
When the main controller 1164α receives a timer expiration notification from the cost reference timer 1167, the main controller 1164α receives information on the accumulated number of bytes passed from the resource monitor 119, and at the same time issues a counter reset notification to reduce the accumulated number of bytes passed through the resource monitor 119 to 0. To reset. Further, the cost is calculated from the cumulative number of passing bytes, the number of TCP flows, or the number of HTTP requests, and the result is passed to the function calculator 1168.

  The function calculator 1168 evaluates the value input from the main controller 1164α using the set function, and returns the result to the main controller 1164α. Here, a proportional function is set, and an example in which the output value for a certain input value is the same will be described.

  When the main controller 1164α receives the cost evaluation result from the function calculator 1168, the main controller 1164α notifies the smoothing circuit 1169 of the value.

  The smoothing circuit 1169 smoothes the input value with a low-pass filter or the like according to the setting, and returns the result to the main controller 1164α.

  When the main controller 1164α receives the cost value after completion of smoothing from the smoothing circuit 1169, the main controller 1164α notifies the standby tree manager 1151 of the cost value. The tree manager 1151 recalculates the spanning tree based on this cost information, and notifies the main controller 1164α whether the topology is changed or not as a result of the calculation. (State 1164A in FIG. 18)

  In the state 1164A, the main controller 1164α transitions to the state 11645 and resets the cost reference timer when the recalculated tree is the same as the pre-calculation tree or the degree of change is smaller than a preset change. To do. In the state transition diagram of FIG. 18, as an example, it is described that the setting does not transit to the state 11645 when there is even a slight change. (State 1164A)

  When the recalculated tree is different from the pre-calculated tree in the state 1164A and the change is greater than the preset change, the main controller 1164α sets the stability timer 1165 and the timer expires. Later, the state transitions to the state 1164B. In the state transition diagram of FIG. 18, it is assumed that the setting is changed to 1164B when there is even a slight change. (State 1164A)

  When the main controller 1164α receives the expiration notice of the stability timer 1165, the main controller 1164α replaces the tree manager 1152 registered for the current use with the tree manager 1151 registered for the backup, and newly uses the tree manager 1151 for the current use and the tree manager 1151 for the new use. The manager 1152 is assumed. Further, the usage tag group change notification is transmitted to the root node of the new tree through the GVRP transceiver 1162. The contents of the usage tag group change notification are reflected in the BPDU and transmitted to all nodes. Thereafter, the stability timer 1165 is activated and the state 11646 is entered. (State 1164B)

  When the main controller 1164α receives the expiration notification of the stability timer 1165, the main controller 1164α notifies the tree manager 1152 of a new cost, and recalculates the spanning tree based on the newly calculated cost information. In addition, the cost reference timer 1167 is scheduled. (State 11646)

  Next, a spanning tree switching operation when the transfer route from the node 15 to the node 13 is changed in the present embodiment will be described in detail using a specific example with reference to FIG.

  Referring to FIG. 7, this operation example includes nodes 11 to 16, clients 91 to 96, bidirectional links 81 to 86, and bidirectional links 21 to 28.

  The node 11 has a client 91 via a link 81, the node 12 has a client 92 by a link 82, the node 13 has a client 93 by a link 83, the node 14 has a client 94 by a link 84, and the node 15 has a link 85. The client 95 is connected to the node 16, and the client 96 is connected to the node 16 via the link 26.

  The link between the node 11 and the node 12 is a link 21, the link between the node 12 and the node 13 is a link 22, the link between the node 13 and the node 14 is a link 23, the link between the node 11 and the node 15 is a link 24, and the node 15 And node 16 are connected by link 25, nodes 16 and 14 are connected by link 26, nodes 12 and 15 are connected by link 27, and nodes 13 and 16 are connected by link 28. Yes.

  Two tag groups to which all the ports of the nodes 11 to 16 belong are already set. The first tag group is called a tag group 41 and the second tag group is called a tag group 42.

  The nodes 11 to 16 have two spanning tree circuits that operate independently. The spanning tree that operates on the tag group 41 is referred to as a tree 51, and the spanning tree that operates on the tag group 42 is referred to as a tree 52.

  In FIG. 7, assume that the tree 41 is already stable with the initial cost set equal to 10 for all links, with node 13 as the root node. The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52.

  Currently, a tag of the tag group 41 which is the active system is added to data transmitted from the client to the nodes 11 to 16. Data to which this tag is added is transferred along the tree 51.

  In the initial state, none of the clients 91 to 96 performs data transfer.

  Each node confirms the state of the BPDU by transmitting and receiving the BPDU at a period defined by the Hello Time at regular time intervals. In this frame, there are identification flags for the active system and the standby system. Currently, only the BPDU to which the tag of the tag group 41 that is the active system is added has the flag of the active system, and the tag group 42 that is the standby system. A flag indicating the active system is not added to the BPDU to which the tag is added.

  Since each node has already received a specification to use dynamic cost calculation from the setting interface 118 or the GVRP transceiver 1162, every time the cost reference timer expires, the frame that has flowed through the link since the previous timer expired. The flow is recalculated and the spanning tree is recalculated using the standby tree.

  Here, it is assumed that data transfer is started between the client 95 and the client 93 and between the client 96 and the client 93.

  In the initial transfer, data from the client 95 to the client 93 is transferred via the links 85, 25, 28, 83 using the tree 51. Data from the client 96 to the client 93 is also transferred via the links 86, 28 and 83 using the tree 51.

  When a certain time elapses from the data transfer (when the cost reference timer 1167 of each node expires), the cost of the links 21 to 28 in the tree 52 is recalculated based on the free capacity of the links 21 to 28, and the function calculator And applied to a smoothing circuit. Here, since the free bandwidth of the link 28 has decreased, the cost of the link 28 of the tree 52 is reduced to 15 by the node 16, and the free bandwidth of the link 25 has decreased to a lesser extent than the link 28. As a result, the cost of the link 25 of the tree 52 is changed to 12, respectively. At this time, the cost of the tree 51 currently used is not changed.

  The node 16 detects the cost change, and sends the BPDU created by using the changed cost (the value of the Root path Cost 22056 in FIG. 5 is changed) to the adjacent nodes 13, 14, and 15. Respectively. A tag of the tag group 42 is added to this BPDU.

  Similarly, the node 15 detects the cost change and transmits the BPDU created using the changed cost to the adjacent node 11, node 12, and node 16.

  When the node 15 receives the BPDU with the cost 15 added from the node 16, the node 15 also adds the cost of the link 25 and recognizes that the cost 27 is required to reach the node 13 via the link 25 and the link 28.

  Thereafter, the node 15 receives the BPDU with the cost 10 added from the node 12 (the node 12 periodically transmits the cost up to the root node (node 13) to the adjacent nodes (node 11 and node 15). In addition, the cost of the link 27 is also added, and it is recognized that the cost 20 is required to reach the node 13 via the link 27 and the link 22. Since this cost is smaller than the cost via the link 25, the node 15 switches the stop port from the link 27 side to the link 25 side, and uses the link 27 and the link 22 to form a tree that reaches the link 13. The spanning tree protocol is also used when creating a tree via the links 27 and 22. Then, the stability timer 1165 is started and waits for the tree to stabilize.

  Here, it is assumed that the tree 52 is stable with the node 13 as a root node.

  When the stability timer 1165 in FIG. 13 expires, the node 13 that is the root node of the tree 51 determines that the tree formation is stable, and uses it for transfer to the node 13 that is the root node of the tree 52 that becomes the new tree. A tag group change message is transmitted to instruct the active tree to be switched from the tree 51 to the tree 52. Thereafter, the stability timer 1165 is activated.

  When the node 13 that is the root node of the tree 52 receives the tree switching message, the node 51 that is the root of the tree 41 that is the old active tree is transferred from the tree 51 to the tree 52 for the transfer. Send a tag group change message, instructing to switch. The contents of the usage tag group change notification are reflected in the BPDUs of the tree 51 and the tree 52 transmitted from the node 13 and transmitted to all nodes.

  When the node 13 completes the transmission of the tag group change message to the old working root node, the node 13 has used the tree received by the client 93 from the client 93 and used to forward the frame to be transmitted within the network. Switch from the existing tree 51 to the tree 52. When the switching is completed, the frame transmitted from the client 93 to the client 95 is transferred to the client 95 via the link 83, the link 22, the link 27, and the link 85.

  In this way, the transfer paths of frames transmitted from the client 93 to the client 95 and from the client 93 to the client 96 are distributed, and the congestion of the link 28 is eliminated.

  Thereafter, every time the cost reference timer expires, the spanning tree is recalculated by cost calculation based on the free bandwidth of the link, and dynamic path change that periodically reflects the free bandwidth in the cost is performed. As a result, the traffic of each link is distributed, the load on the link is distributed, and congestion can be prevented.

  Next, the effect of this embodiment will be described.

  Conventionally, since the cost is calculated by using the link capacity and is used for route selection at the time of spanning tree construction, the route cannot be changed for dynamic load distribution according to traffic.

  In this embodiment, the traffic load can be distributed by calculating the link cost based on dynamic information such as a free bandwidth and a server load.

  Conventionally, when trying to dynamically change the cost according to the traffic situation, it is being rebuilt because the span of the data frame is stopped locally and the entire network is stopped and the spanning tree is rebuilt to change the path. In some cases, the network stopped.

In this embodiment, the network after the cost change is generated while the tree before the change is operated, and the tree used after the stabilization of the new tree is switched to reconfigure the spanning tree along with the route change. It is possible to distribute the load without stopping.
As a result, it is possible to reduce the possibility of congestion.

(Third embodiment)
Hereinafter, a third embodiment of the present invention will be described in detail with reference to the drawings.

  The third embodiment of the present invention is different from the second embodiment in that a transition is made between the active system and the standby system regardless of whether the cost has been changed or not. In the following, the case where the capacity of the free bandwidth is used as the cost will be described, but the number of passing TCP flows and the number of HTTP requests can be similarly realized unless otherwise specified.

  Referring to FIG. 19, in the third embodiment of the present invention, in FIG. 18 in the second embodiment, the transition between the state 1164A and the state 1164B does not occur, and the state 11443 and the state 11644 And the transition between the state 11647 and the state 11648 is not caused by the expiration of the stability timer 1165 but is caused by detection of switching of the working flag in the BPDU.

  When the main controller 1164β in the third embodiment receives a notification from the resource monitor 119 that it is newly connected to the network, the main controller 1164β waits for the arrival of an active system notification from the tree manager 1151 or the tree manager 1152. When the active system notification included in the BPDU is received from the tree manager 1151 or the tree manager 1152, the tree manager 1151 or the tree manager 1152 specified in the notification is used for the current use, and the tree manager 1151 or 1152 specified for the backup is used for the standby. To state 11642 or state 11649. Here, the case where the state transitions to the state 11642 will be described as an example, but the following description is the same when the state transitions to the state 11649. (State 11641 in FIG. 19)

  The main controller 1164β sets the tree manager 1151 as active and the tree manager 1152 as standby. Further, a BPDU transmission stop command is issued to each of the tree managers 1151 and 1152. (State 11642 in FIG. 19)

  If the main controller 1164β receives a node addition request from the setting interface 118 in the state 11642, the main controller 1164β transitions to the state 11443. If the active system notification is received from the tree manager 1151 or the tree manager 1152 and the relationship between the active system and the standby system has changed, the state 11649 is entered. (State 11642 in FIG. 19)

The main controller 1164β notifies the tree manager 1152 of the link up, and at the same time, permits the tree manager 1152 to transmit BPDUs. Further, the stability timer 1165 is activated. (State 11643 in FIG. 19)
In the state 11643, the main controller 1164β receives the active system notification contained in the BPDU from the tree manager 1151 or the tree manager 1152, and if there is a change in the relationship between the active system and the standby system, it is registered as active. The existing tree manager 1151 and the tree manager 1152 registered for backup are replaced with each other, and a new current tree manager 1152 and a spare tree manager 1151 are used. Thereafter, the stability timer 1165 is activated. (State 11644 in FIG. 19)

  When the main controller 1164β receives the timer expiration notification from the stability timer 1165, the main controller 1164β notifies the tree manager 1151 of the link up, and at the same time, permits the tree manager 1152 to transmit the BPDU. Usually, it is stable in this state. (State 11645 in FIG. 19)

  In the state 11645, if the current tree manager 1152 is the root node, the main controller 1164β of this node activates the stability timer 1165, and when the timer expiration notification is received from the stability timer 1165, it is registered as active. The tree manager 1152 registered for spare and the tree manager 1151 registered for spare are replaced with each other, and the new working is designated as the tree manager 1151 and the spare is designated as the tree manager 1152. Further, the usage tag group change notification is transmitted to the root node of the new tree through the GVRP transceiver 1162. The contents of the usage tag group change notification are reflected in the BPDU and transmitted to all nodes. (State 11645 in FIG. 19)

  The replacement of the tree manager may be performed by a spare tree manager 1151. In this case, in the state 11645, if the spare tree manager 1151 is the root node, the main controller 1164 of the node operates the stability timer 1165 and receives the timer expiration notification from the stability timer 1165. The tree manager 1152 registered in FIG. 1 and the tree manager 1151 registered for backup are exchanged, and the new working is designated as the tree manager 1151 and the spare is designated as the tree manager 1152. Further, the usage tag group change notification is transmitted to the root node of the old working tree through the GVRP transceiver 1162. The contents of the usage tag group change notification are reflected in the BPDU and transmitted to all nodes. (State 11645 in FIG. 19)

  In the state 11645, the main controller 1164β receives the active system notification included in the BPDU from the tree manager 1151 or the tree manager 1152, and transitions to the state 11646 if the relationship between the active system and the standby system changes. . Then, the tree manager 1152 registered for the current use is replaced with the tree manager 1151 registered for the backup, and the new work is set as the tree manager 1151 and the backup is set as the tree manager 1152. (State 11645 in FIG. 19)

  When the main controller 1164β receives a node deletion request from the setting interface 118 in the state 11645, the main controller 1164β transitions to the state 11644. (State 11645 in FIG. 19)

  The main controller 1164β notifies the tree manager 1151 of all connected link links down, and simultaneously issues a BPDU transmission stop command to the tree manager 1151. (State 11644 in FIG. 19)

  In the state 11644, the main controller 1164β receives the active system notification included in the BPDU from the tree manager 1151 or the tree manager 1152, and if there is a change in the relationship between the active system and the standby system, the main controller 1164β transitions to the state 11443. The tree manager 1152 registered for the current use and the tree manager 1151 registered for the backup are exchanged, and the new work is set as the tree manager 1151 and the spare is used as the tree manager 1152. Thereafter, the stability timer 1165 is activated. (State 11643 in FIG. 19)

  When the main controller 1164β receives a timer expiration notification from the stability timer 1165, the main controller 1164β notifies the tree manager 1151 of link down, and simultaneously issues a BPDU reception stop command to the tree manager 1151. Furthermore, the state transits to the state 11641 unconditionally and waits until the node is disconnected. (State 11642 in FIG. 19)

  When the main controller 1164β receives a designation to use dynamic cost calculation from the setting interface 118 or the GVRP transceiver 1162, the cost reference is made so that it expires after a certain period of time has elapsed since the use tag group change notification was received. Timer 1167 is set. (State 11645 in FIG. 19)

  When the main controller 1164β receives a timer expiration notification from the cost reference timer 1167, the main controller 1164β receives information on the accumulated number of bytes passed from the resource monitor 119, and at the same time issues a counter reset notification to set the accumulated number of bytes passed through the resource monitor 119 to 0. To reset. Further, the cost is calculated from the cumulative number of passing bytes, the number of TCP flows, or the number of HTTP requests, and a notification is sent to the tree manager 1151 which is a standby system. The tree manager 1151 recalculates the spanning tree based on the newly calculated cost information, and notifies the main controller 1164β whether the topology is changed or not as a result of the calculation. And it changes to the state 11645 unconditionally. (State 1164A in FIG. 19)

  With reference to FIGS. 14 and 15, the operation when adding node 17 in the present embodiment will be described in detail using a specific example.

  Referring to FIG. 14, this operation example includes nodes 11 to 17 and links 21 to 30. However, the node 17, the link 29, and the link 30 are not connected in the initial state.

  Referring to FIG. 15, this operation example includes nodes 11 to 17 and links 21 to 30.

  Two tag groups to which all nodes and ports of the nodes 11 to 16 belong are already set in the initial state, and the first tag group is called a tag group 41 and the second tag group is called a tag group 42.

  Basically, all nodes and all ports are joined to two tag groups, but a tag group composed of only some ports or nodes may be created. In the following description, it is assumed that all nodes and all ports join two tag groups.

  The nodes 11 to 16 have two spanning tree circuits that operate independently. The spanning tree that operates on the tag group 41 is referred to as a tree 51, and the spanning tree that operates on the tag group 42 is referred to as a tree 52.

  Although two spanning trees are always created, it is not always necessary to create two tag groups. Only the tag group 41 is set and the tag group 42 is not used. A spanning tree operating on the tag group 41 can be called a tree 51, and a spanning tree operating without belonging to the tag group can be called a tree 52. On the other hand, only the tag group 42 is set, the tag group 41 is not used, the spanning tree that operates without belonging to the tag group is called the tree 51, and the spanning tree that operates on the tag group 42 is called the tree 52. it can.

  Here, the case where both the tag group 41 and the tag group 42 are used will be described, but the operation when both the tag group 41 and the tag group 42 are used is only the tag group 41 or only the tag group 42. The same applies to the case where is used.

  In FIG. 14, it is assumed that the tree 51 is represented by a bold line and is stable with the node 11 as a root node.

  The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52.

  Currently, a tag of the tag group 41 is added to data transmitted from the client to the nodes 11 to 16. Data to which this tag is added is transferred along the tree 51.

  When node 17 is connected to link 29 and link 30, node 17 does not join any tag group and starts receiving BPDUs.

  When the node 17 receives the BPDU of each tag group, the node 17 confirms that the current active system is the tag group 41 and the current standby system is the tag group 42. Then, the own node is allowed to participate only in the tag group 42 to perform transmission / reception of BPDU, and the tag group 41 is set to perform only reception of BPDU and not perform transmission.

  When the node 17 is added, a change occurs in the members of the tag group 42. Therefore, an operation of updating the tree 52 is started by the spanning tree protocol. Since there is no change in the members of the tag group 41, the tree 51 is not updated.

  The tag transmitted from the client is added with the tag of the tag group 41 as before, and continues to be transferred along the tree 51.

  When the stability timer expires, the node 11 that is the root node of the tree 51 that is the active tree transmits a usage tag group change notification to the node 11 that is the root node of the tree 51, and the tree 51 is set as the standby system. Command to transition to. Then, the flag of the tag group 42 is attached, and the working system flag is set in the BPDU transmitted for the tree 52. The BPDU propagates to all nodes while being transferred at each node.

  The node 11 senses a change in the status of addition of the working system flag, transitions the tree 51 to the protection system, attaches the flag of the tag group 41, and sets the protection system flag in the BPDU transmitted for the tree 51. The BPDU propagates to all nodes while being transferred at each node.

  The nodes 11 to 17 confirm that the working flag is added to the BPDU with the tag of the tag group 42, and switch the tag to be added to the frame transferred from the client from the tag group 41 to the tag group 42. . The frame to which this tag is added is transferred along the tree 52.

  After a while after the above switching is completed, no frames flow through the tree 51.

  The node 17 determines that there is no longer a node to which the tag of the tag group 41 is added after a lapse of a certain time from the transmission of the usage tag group change notification, and joins the own node to the tag group 41 to prepare for the next topology change. .

  As described above, the node 17 can be added without stopping the network. Thereafter, the same operation is repeated when adding a node. However, the tag group 41 and the tag group 42 in the above description are interchanged.

  Next, a spanning tree switching operation when the transfer route from the node 15 to the node 13 is changed in the present embodiment will be described in detail using a specific example with reference to FIG.

  Referring to FIG. 7, this operation example includes nodes 11 to 16, clients 91 to 96, bidirectional links 81 to 86, and bidirectional links 21 to 28.

  The node 11 has a client 91 via a link 81, the node 12 has a client 92 by a link 82, the node 13 has a client 93 by a link 83, the node 14 has a client 94 by a link 84, and the node 15 has a link 85. The client 95 is connected to the node 16, and the client 96 is connected to the node 16 via the link 26.

  The link between the node 11 and the node 12 is a link 21, the link between the node 12 and the node 13 is a link 22, the link between the node 13 and the node 14 is a link 23, the link between the node 11 and the node 15 is a link 24, and the node 15 And node 16 are connected by link 25, nodes 16 and 14 are connected by link 26, nodes 12 and 15 are connected by link 27, and nodes 13 and 16 are connected by link 28. Yes.

  Two tag groups to which all the ports of the nodes 11 to 16 belong are already set. The first tag group is called a tag group 41 and the second tag group is called a tag group 42.

  The nodes 11 to 16 have two spanning tree circuits that operate independently. The spanning tree that operates on the tag group 41 is referred to as a tree 51, and the spanning tree that operates on the tag group 42 is referred to as a tree 52.

  It is assumed that the tree 41 is represented by a bold line, and that the node 13 is the root node and is already stable with an initial cost set equal to 10 for all links. The tag of the tag group 41 is attached to the BPDU of the tree 51, and the tag of the tag group 42 is attached to the BPDU of the tree 52.

  Currently, a tag of the tag group 41 which is the active system is added to data transmitted from the client to the nodes 11 to 16. Data to which this tag is added is transferred along the tree 51.

  In the initial state, none of the clients 91 to 96 performs data transfer.

  Each node transmits a HELLO frame at regular time intervals to check the state of the BPDU. In this frame, there are identification flags for the active system and the standby system. Currently, only the BPDU to which the tag of the tag group 41 that is the active system is added has the flag of the active system, and the tag group 42 that is the standby system. A flag indicating the active system is not added to the BPDU to which the tag is added.

  Since each node has already received a specification to use dynamic cost calculation from the setting interface 118 or the GVRP transceiver 1162, every time the cost reference timer expires, the frame that has flowed through the link since the previous timer expired. The flow is recalculated and the spanning tree is recalculated using the standby tree.

  Here, it is assumed that data transfer is started between the client 95 and the client 93 and between the client 96 and the client 93.

  In the initial transfer, data from the client 95 to the client 93 is transferred via the links 85, 25, 28, 83 using the tree 51. Data from the client 96 to the client 93 is also transferred via the links 86, 28 and 83 using the tree 51.

  When a certain time has elapsed since the data transfer, the costs of the links 21 to 28 in the tree 52 are recalculated one by one based on the free capacity of the links 21 to 28. Here, since the free bandwidth of the link 28 has decreased, the cost of the link 28 of the tree 52 is reduced to 15 by the node 16, and the free bandwidth of the link 25 has decreased to a lesser extent than the link 28. As a result, the cost of the link 25 of the tree 52 is changed to 12, respectively. At this time, the cost of the tree 51 currently used is not changed.

  The node 16 detects the cost change and transmits the BPDU created using the changed cost to the adjacent node 13, node 14, and node 15.

  Similarly, the node 15 detects the cost change and transmits the BPDU created using the changed cost to the adjacent node 11, node 12, and node 16.

  When the node 15 receives the BPDU with the cost 15 added from the node 16, the node 15 also adds the cost of the link 25 and recognizes that the cost 27 is required to reach the node 13 via the link 25 and the link 28.

  After that, when receiving the BPDU with the cost 10 added from the node 12, the node 15 adds the cost of the link 27 and recognizes that the cost 20 is required to reach the node 13 via the link 27 and the link 22. To do. Since this cost is smaller than the cost via the link 25, the node 15 switches the stop port from the link 27 side to the link 25 side, and uses the link 27 and the link 22 to form a tree that reaches the link 13.

  When the topology stabilization timer that was started in the node 13 that is the root node of the tree 51 expires, the active tree that is used for transfer is transferred from the tree 51 to the node 13 that is the root of the tree 41 that is the old active tree. A tag group change message is sent that instructs switching to the tree 52. The contents of the usage tag group change notification are reflected in the BPDUs of the tree 51 and the tree 52 transmitted from the node 13 and transmitted to all nodes.

  When the node 13 completes the transmission of the tag group change message to the old working root node, the node 13 has used the tree received by the client 93 from the client 93 and used to forward the frame to be transmitted within the network. Switch from the existing tree 51 to the tree 52. When the switching is completed, the frame transmitted from the client 93 to the client 95 is transferred to the client 95 via the link 83, the link 22, the link 27, and the link 85.

  In this way, the transfer paths of frames transmitted from the client 93 to the client 95 and from the client 93 to the client 96 are distributed, and the congestion of the link 28 is eliminated.

  Thereafter, every time the cost reference timer expires, the spanning tree is recalculated by cost calculation based on the free bandwidth of the link, and dynamic path change that periodically reflects the free bandwidth in the cost is performed. As a result, the traffic of each link is distributed, the load on the link is distributed, and congestion can be prevented.

  Next, the effect of this embodiment will be described.

  Conventionally, since the cost is calculated by using the link capacity and is used for route selection at the time of spanning tree construction, the route cannot be changed for dynamic load distribution according to traffic.

  In this embodiment, the traffic load can be distributed by calculating the link cost based on dynamic information such as a free bandwidth and a server load.

  Conventionally, every time the cost is dynamically changed according to traffic conditions, the data frame forwarding is stopped locally or the entire network, and the spanning tree is rebuilt to change the path. The network might stop.

In this embodiment, the network after the cost change is generated while the tree before the change is operated, and the tree used after the stabilization of the new tree is switched to reconfigure the spanning tree along with the route change. It is possible to distribute the load without stopping.
As a result, it is possible to reduce the possibility of congestion.

(Fourth embodiment)
Hereinafter, the fourth embodiment of the present invention will be described in detail with reference to the drawings.

  The fourth embodiment of the present invention corresponds to the case where the tag and spanning tree used for each destination node are switched and the destination node is set as the root node in the first embodiment.

  When a frame is transmitted in a network in which IEEE 802.1D and IEEE 802.1w are operating, the lowest cost route to the destination is not always selected, an unused link appears, and the load is concentrated on the root node There was a problem that the network stopped for a long time when the root node failed.

  In this embodiment, frames are transferred using a tree whose destination is the root node, so that frame transmission at the lowest cost to the destination is improved, link utilization is improved, and resistance to root node failures is enhanced. I do.

  Referring to FIG. 20, the fourth embodiment of the present invention is different from that of FIG. 8 in the first embodiment in that tree managers 1151 are installed by the number of nodes existing in the network.

  The tree manager 1151 has the same function as the tree manager 1151 shown in FIG. 8 in the first embodiment of the present invention.

  The tree manager 1152 and the tree manager 1153 are tree managers similar to the tree manager 1151. Thereafter, the tree manager 1151 is used to describe the tree manager 1151 to the tree manager 1153, but the description of the tree manager 1151 can be similarly applied to the tree manager 1152 to the tree manager 1153 unless otherwise specified. It is.

  The tree manager 116 creates the number of nodes existing in the network by the tree selector 116, or the number of nodes existing in the same hierarchy when the subnets are divided into hierarchies. Therefore, one or more tree managers may increase to infinity. In FIG. 20, the tree managers are collectively expressed as tree managers 1151 to 1153.

  In addition to the function of the tree selector 116 shown in FIG. 8 in the first embodiment of the present invention, the tree selector 116 in FIG. 20 displays a new tree manager when the presence of a new node in the network or hierarchy is detected. It has a function to generate. Further, it has a function of notifying other nodes of the detection of the new node, and a function of receiving a notification of detecting a new node from another node and generating a tree manager. Further, a function for detecting deletion of a node and deleting a tree manager, a function for notifying other nodes of the detection of the deleted node, and a node deletion notification from another node are received. Has a function to delete.

  FIG. 21 is a configuration example of the forwarding table 114 that determines an output port using a tag as a key in FIG. 20 of the present embodiment.

  The tag field 1141 is a field serving as a search index, and it is checked whether or not the information in this field matches the content written in the tag of the received frame.

  The output port 1142 is a field for describing to which port the frame should be transferred when the contents written in the tag of the received frame and the contents of the field 1141 match the tag.

  Note that this embodiment determines not only the tag forwarding that determines the transfer destination port according to the contents of the tag, but also the transfer destination based on the MAC address that has been conventionally used, as shown in this operation example. The same applies to normal MAC address transfer. In this case, a plurality of ports corresponding to the corresponding MAC address may be described in the output port field 1142.

  FIG. 22 is a configuration example of the tag table 117 that determines a tag to be inserted using the destination MAC address as a key in FIG. 20 of the present embodiment.

  The destination MAC address 1171 is a field that becomes an index for search, and it is checked whether the information in this field matches the contents written in the destination MAC address field of the received frame, that is, the MAC DA field. In this case, the tag described in the insertion tag field 1172 is inserted into the received frame.

  The insertion tag field 1172 is a field in which a tag to be inserted with respect to the destination MAC address field 1171 is described. In this embodiment, a destination node ID is described, and this ID is inserted as a tag into the frame.

  FIG. 23 is a configuration diagram of a tree 61 that is a configuration diagram of a spanning tree having the node 11 as a root node. The tree 61 is created by setting the priority value of the node 11 to a value smaller than each of the nodes 12 to 16. The tree 61 is used when transmitting a frame toward the node 11 and when transmitting a broadcast frame from the node 11 to each of the nodes 12 to 16.

  FIG. 24 is a configuration diagram of a tree 62 that is a configuration diagram of a spanning tree having the node 12 as a root node. The tree 62 is created by setting the priority value of the node 12 to a value smaller than each of the nodes 11 and 13 to 16. The tree 62 is used when transmitting a frame toward the node 12 and when transmitting a broadcast frame from the node 12 to each of the nodes 11 and 13 to 16.

  FIG. 25 is a configuration diagram of a tree 63 that is a configuration diagram of a spanning tree having the node 13 as a root node. The tree 63 is created by setting the priority value of the node 13 to a value smaller than each of the nodes 11 to 12 and the nodes 14 to 16. The tree 63 is used when transmitting a frame toward the node 13 and when transmitting a broadcast frame from the node 13 to each of the nodes 11 to 12 and the nodes 14 to 16.

  FIG. 26 is a configuration diagram of a tree 64 that is a configuration diagram of a spanning tree having the node 14 as a root node. The tree 64 is created by setting the priority value of the node 14 to a value smaller than each of the nodes 11 to 13 and the nodes 15 to 16. The tree 64 is used when transmitting a frame toward the node 14 and transmitting a broadcast frame from the node 14 to each of the nodes 11 to 13 and the nodes 15 to 16.

  FIG. 27 is a configuration diagram of a tree 65, which is a configuration diagram of a spanning tree having the node 15 as a root node. The tree 65 is created by setting the priority value of the node 15 to a value smaller than each of the nodes 11 to 14 and the node 16. The tree 65 is used when transmitting a frame toward the node 15 and when transmitting a broadcast frame from the node 15 to each of the nodes 11 to 14 and the node 16.

  FIG. 28 is a configuration diagram of a tree 66 that is a configuration diagram of a spanning tree having the node 16 as a root node. The tree 66 is created by setting the priority value of the node 16 to a value smaller than those of the nodes 11 to 15. The tree 66 is used when transmitting a frame toward the node 16 and when transmitting a broadcast frame from the node 16 to each of the nodes 11 to 15.

  Next, with reference to FIG. 23 to FIG. 28, regarding the creation operation of the tree 63 when the node 13 is newly added to the network already configured by the nodes 11 to 12 and the nodes 14 to 16. explain.

  When a node 13 is added to the network, the node 13 receives a BPDU frame transmitted from an adjacent node and generates a tree manager for each newly detected identification tag. In this example, five tree managers with the nodes 11 to 12 and the nodes 14 to 16 as root nodes are generated.

  Next, the node 13 generates a tag ID from the node ID, generates a tree manager in which the priority value of the node is set low, and adds the tag ID to the BPDU frame output from the tree manager. To send. Here, it is assumed that the tag ID is 43.

  The node 12 and the node 16 newly receive the BPDU with the tag ID 43, generate a tree manager, and then transmit the BPDU with the tag ID 43 added to the adjacent nodes.

  The tree 63 is completed by repeating the above BPDU transmission operation.

  Next, with reference to FIG. 23 to FIG. 28, a procedure when each of the nodes 11 to 16 in the above drawings transmits a frame to each of the nodes 11 to 16 will be described. Is delivered to the destination by the lowest cost route, and the load distribution of the link resource is performed. It is assumed that the cost of each link is the same, and that each of the trees 61 to 66 in each figure has already been configured and the topology is stable.

  When a frame is transmitted from each of the nodes 12 to 16 to the node 11, the tree 61 is used. For example, when transmitting a frame from the node 15 to the node 11, the node 15 adds a tag ID 41 that is an identification tag of the tree 61 to the data frame and transmits the data frame.

  When a frame is transmitted from the node 11 and each of the nodes 13 to 16 to the node 12, the tree 62 is used. For example, when transmitting a frame from the node 14 to the node 12, the node 14 adds a tag ID 42 that is an identification tag of the tree 62 to the data frame and transmits the data frame.

  When a frame is transmitted from each of the nodes 11 to 12 and the nodes 14 to 16 to the node 13, the tree 63 is used. For example, when a frame is transmitted from the node 11 to the node 13, the node 11 adds a tag ID 43 that is an identification tag of the tree 63 to the data frame and transmits the data frame.

  When a frame is transmitted from each of the nodes 11 to 13 and the nodes 15 to 16 to the node 14, the tree 64 is used. For example, when transmitting a frame from the node 12 to the node 14, the node 12 adds a tag ID 44 that is an identification tag of the tree 64 to the data frame and transmits it.

  When a frame is transmitted from each of the nodes 11 to 14 and the node 16 to the node 15, the tree 65 is used. For example, when a frame is transmitted from the node 16 to the node 15, the node 16 adds a tag ID 45 that is an identification tag of the tree 65 to the data frame and transmits the data frame.

  When a frame is transmitted from each of the nodes 11 to 15 to the node 16, the tree 66 is used. For example, when transmitting a frame from the node 14 to the node 16, the node 14 adds a tag ID 46 that is an identification tag of the tree 66 to the data frame and transmits the data frame.

  The tree 61 is used when the node 11 broadcasts a frame to all the nodes in the network. For example, the node 11 adds a tag ID 41, which is an identification tag of the tree 61, to a data frame whose destination is broadcast, and transmits the data frame.

  The tree 62 is used when the node 12 broadcasts a frame to all nodes in the network. For example, the node 12 adds a tag ID 42 that is an identification tag of the tree 62 and transmits the data frame whose destination is broadcast.

  When the node 13 broadcasts the frame to all the nodes in the network, the tree 63 is used. For example, the node 13 adds a tag ID 43 that is an identification tag of the tree 63 to the data frame whose destination is broadcast, and transmits the data frame.

  The tree 64 is used when the node 14 broadcasts a frame to all nodes in the network. For example, the node 14 adds a tag ID 44 that is an identification tag of the tree 64 to the data frame whose destination is broadcast, and transmits the data frame.

  The tree 65 is used when the node 15 broadcasts a frame to all the nodes in the network. For example, the node 15 adds a tag ID 45 that is an identification tag of the tree 65 to a data frame whose destination is broadcast, and transmits the data frame.

  The tree 66 is used when the node 16 broadcasts a frame to all nodes in the network. For example, the node 16 adds a tag ID 46 that is an identification tag of the tree 66 to a data frame whose destination is broadcast, and transmits the data frame.

  The data frame can be transferred through the lowest cost path by adding the transmission tag with the method described above and transferring the data frame. Further, since frames are transferred using a plurality of trees having different root nodes, as the spanning trees shown in the prior art 1 and the prior art 2 are concentrated, traffic is concentrated in the vicinity of the root node and the link is further away from the root node. It can be seen that the added traffic can be distributed without causing the phenomenon that the usage rate of the traffic decreases.

  Next, with reference to FIGS. 23 to 28, the operation when a failure occurs in a node will be described by taking the case where a failure occurs in the node 12 as an example. In the initial state, it is assumed that the trees 61 to 66 are already constructed and stable.

  When the node 12 stops due to a failure, the tree 61 is routed via the link 25, the link 27, the link 26, and the link 23 as a route from the node 13 to the node 11 by the rapid spanning tree method defined in IEEE802.1w. Is selected, and the transfer of the frame to the node 11 and the frame broadcast from the node 11 to each node is continued.

  If the node 12 stops due to a failure, the tree 62 must be reconfigured because the node 12 is the root node. Another node other than the node 12 becomes a root node, and the tree 62 is reconfigured before the node 12 is restored. This reconfiguration takes several tens of seconds in IEEE 802.1D and several seconds in IEEE 802.1w, but the tree 62 is originally sent from each node to the route 12 and from the route 12 to each node. Therefore, even if it takes a long time to reconfigure, it does not affect communications performed between other nodes other than the node 12.

  When the node 12 stops due to a failure, the tree 63 is a route from the node 11 to the node 13 as a route from the node 11 to the node 13 via the link 23, the link 26, the link 27, and the link 25 by the rapid spanning tree method defined in IEEE802.1w. Is selected, and the transfer of the frame to the node 13 and the frame broadcast from the node 13 to each node is continued.

  When the node 12 stops due to a failure, the tree 64 is reconstructed by a rapid spanning tree method defined in IEEE802.1w, and a frame transmitted from each node to the node 14 and a broadcast from the node 14 to each node are broadcast. Frame transfer continues.

  When the node 12 stops due to a failure, the tree 65 is reconstructed by a rapid spanning tree method defined in IEEE802.1w, and a frame from each node to the node 15 and broadcast from the node 15 to each node are broadcast. The frame transfer continues.

  In the tree 66, when the node 12 is stopped due to a failure, the route via the link 23, the link 26, and the link 27 is selected as the route from the node 11 to the node 16 by the rapid spanning tree method defined in IEEE802.1w. The transfer of the frame to the node 16 and the frame broadcast from the node 16 to each node is continued.

  Next, referring to FIG. 29 and FIG. 30, a method of configuring a spanning tree when a part of the client in FIG. 7 of the first embodiment is connected to a plurality of nodes by dual homing. State.

  In FIG. 29, a client 97 is a set of one or more clients, and has a function of transmitting and receiving a frame between the node 11 and the node 16 through a link 87 and a link 88.

  The link 87 is a bidirectional link that connects the client 97 to the node 15 and the node 15 to the client 97.

  The link 88 is a bidirectional link connecting the client 97 to the node 16 and the node 16 to the client 97.

  When there is a client group connected to a plurality of nodes like the client 97 in FIG. 29, the client is regarded as a virtual node and a spanning tree is set.

  FIG. 30 is a network configuration diagram when the client 97 is regarded as the virtual node 18 in FIG.

  The spanning tree 74 is a spanning tree having the node 18 as a root node. A frame transmitted from each of the nodes 11 to 16 to the node 18 arrives at the node 18, that is, the client 97 using the spanning tree 74. A broadcast frame transmitted from the client 97, that is, the node 18 is also broadcast to each of the nodes 11 to 16 along the spanning tree 74.

  Note that the node 18 is a virtual node, and the actual operation of the node 18 is performed on behalf of the node 15 or 16. Which of the node 15 and the node 16 takes over the operation of the node 18 is determined not only by manual setting by the setting interface 118 but also by a method such as automatic setting for a smaller node ID and automatic setting for a larger node ID. The

  Next, a method for performing communication without installing a virtual node when the client 97 is connected to the node 15 and the node 16 through the link 87 and the link 88 with reference to FIG. 29 will be described. .

  In FIG. 29, the node 15 and the node 16 detect that the client 97 is connected to a plurality of nodes by setting on the setting interface or learning. The node 15 detects that the client 97 is connected to the node 16. The node 16 detects that the client 97 is connected to the node 15.

  Node 15 and node 16 exchange control messages with each other and determine which of the nodes 15 or 16 forwards the frame to client 97. This forwarding node is determined as a node having a small node ID, a node having a large node ID, or a node determined by setting.

  When the forwarding node is determined, the client 97 is considered to be connected only to the node 16 and frame forwarding is started. The nodes 11 to 16 recognize that the client 97 is connected to the node 16 by learning or the like, and transmit a frame addressed to the client 97 with a tree identification tag having the node 16 as a root node.

  The nodes 15 and 16 always monitor each other's situation by Keep Alive or the like. If the node 15 cannot confirm the operation of the node 16, the node 15 transfers the frame from the client 97 to the nodes 11 to 16. Then, the nodes 11 to 16 learn that 15 is added as a node ID to the frame transmitted from the client 97, and transmit the frame addressed to the client 97 toward the node 15.

  Through the above operation, the client 97 can transmit and receive a frame. The above operation can be similarly applied even when the node 15 and the node 16 are switched.

  Next, referring to FIG. 29, when the client 97 is connected to the node 15 and the node 16 through the link 87 and the link 88 in a dual homing connection, communication is performed without installing a virtual node. A method will be described in which a detection node transmits a switching notification to all the nodes in the network so that the connection destination of the client 97 is changed at high speed.

  In FIG. 29, the node 15 and the node 16 detect that the client 97 is connected to a plurality of nodes by setting in the setting interface 118 or learning. The node 15 detects that the client 97 is connected to the node 16. The node 16 detects that the client 97 is connected to the node 15.

  Node 15 and node 16 exchange control messages with each other and determine which of the nodes 15 or 16 forwards the frame to client 97. This forwarding node is determined as a node having a small node ID, a node having a large node ID, or a node determined by setting.

  When the forwarding node is determined, the client 97 is considered to be connected only to the node 16 and frame forwarding is started. The nodes 11 to 16 recognize that the client 97 is connected to the node 16 by learning or the like, and transmit a frame addressed to the client 97 with an identification tag of a tree having the node 16 as a root node.

  The nodes 15 and 16 always monitor each other's situation by Keep Alive or the like. If the node 15 cannot confirm the operation of the node 16, the node 15 transfers the frame from the client 97 to the nodes 11 to 16. Further, the node 15 notifies all the nodes in the network that the node 15 is now in charge of the frame transfer addressed to the client 97 instead of the node 16.

  The nodes 11 to 16 receive the notification, insert a tag addressed to the node 15 into the frame addressed to the client 97, and transmit the frame addressed to the client 97 toward the node 15.

  Through the above operation, the client 97 can transmit and receive a frame. The above operation can be similarly applied even when the node 15 and the node 16 are switched.

  Next, the effect of this embodiment will be described.

  Conventionally, the lowest cost route to a destination has not always been selected.

  In this embodiment, the lowest cost route to the destination can be selected by transferring the frame using a tree whose destination is the root node.

  Conventionally, the link utilization rate is low, but the load is concentrated near the root node.

  In the present embodiment, by setting a plurality of spanning trees having different root nodes, it is possible to increase the link utilization rate and distribute the load without concentrating the load near the root node.

  Further, conventionally, there has been a problem that it takes time to construct a tree when a root node fails, and the network stops during that time.

  In this embodiment, by transferring a frame using a tree whose destination is the root node, frames other than the frame whose destination is the root node are not disabled for a long time due to the influence of the root node failure. Therefore, it is possible to avoid a network stop due to a root node failure.

  As a result, it is possible to reduce the possibility of congestion.

(Fifth embodiment)
Hereinafter, a fifth embodiment of the present invention will be described in detail with reference to the drawings.

  The fifth embodiment of the present invention identifies the version of the BPDU in the first embodiment, sets a large cost for the section using the low-speed IEEE 802.1D, that is, the prior art 1, and sets the high-speed This corresponds to a case where a spanning tree with a small cost is generated for a section using IEEE802.1w, that is, prior art 2.

  In the section using IEEE802.1D, the route switching at the time of failure is slow, and it takes time to reconfigure the spanning tree. Therefore, if a tree that passes through this section is set, it will take time to generate a failure or change the route. There is a problem that congestion occurs and frames are lost.

  In the present embodiment, the cost of the IEEE 802.1D usage section is set to be large so that the spanning tree is prevented from being set through the IEEE 802.1D usage section, and the switching and route change at the time of failure are accelerated. Prevent congestion and frame loss.

  Referring to FIG. 31, the fifth embodiment of the present invention is different in that a cost operator 11516 is added to FIG. 10 in the first embodiment.

  In addition to the operation of the tree controller 11514 in the first embodiment of the present invention, the tree controller 11514 determines the version of the received BPDU, and if a BPDU of a version lower than the set version arrives, The cost of the link to which the node that transmitted the BPDU is connected is reset by the cost operator 11516 and written in the tree table 11515. This operation is performed only once each time a cost change notification is received from the tree selector 116.

  The cost operator 11516 adds a preset value to the value input from the tree controller 11514 and returns it to the tree controller 11514.

  When receiving a BPDU reception notification from the BPDU transceiver 11512, the tree controller 11514 sets a value in the tree table 11515 according to the content of the notification. The PDU reception notification includes information regarding the received BPDU version and reception port, and the information is also held in the tree table 11515.

  When cost information is notified from the tree selector 116, the tree controller 11514 sets a cost value in the table according to the notified information. At this time, when a cost is set for a port that has received a BPDU of a version older than a preset version, the cost notified from the tree selector 116 is notified to the cost operator 11516.

  The cost operator 11516 adds a preset value to the value input from the tree controller 11514 and returns it to the tree controller 11514.

  The tree controller 11514 notifies the tree table 11515 of the cost returned from the cost operator 11516 as the cost of the corresponding port.

  The tree controller 11514 reconstructs the tree according to the spanning tree algorithm after the cost update for all ports is completed.

  Next, the spanning tree creation operation in the present embodiment will be described with reference to FIGS.

  In FIGS. 32 to 34, it is assumed that the node 12 is a node that does not support IEEE 802.1w, and only supports IEEE 802.1D. Nodes other than the node 12, that is, the node 11, the node 13, the node 14, the node 15, and the node 16 are assumed to support IEEE 802.1w.

  Each of the nodes 11, 15, and 13 recognizes that the node 12 is a node that supports only 802.1D based on the version information or protocol ID in the BPDU frame transmitted from the node 12.

  The nodes 11, 15, and 13 set the costs of the links 21, 22, and 24 to be sufficiently higher than the costs of the other links. Here, it is assumed that the costs of the link 21, the link 22, and the link 24 are set to 10, and the costs of the other links, that is, the link 23, the link 26, the link 27, and the link 25 are set to 1.

  FIG. 32 is a configuration diagram of the spanning tree when the node 11 or the node 14 is the root node in the cost setting state.

  FIG. 33 is a configuration diagram of the spanning tree when the node 15, the node 16, or the node 14 is the root node in the cost setting state.

  FIG. 34 is a configuration diagram of the spanning tree when the node 13 or the node 16 is a root node in the cost setting state.

  As shown in FIGS. 32 to 34, in this embodiment, it is possible to configure the tree so as to avoid the IEEE802.1D usage section that takes time to recover from the failure. It can be made smaller and can be recovered quickly.

  Next, the effect of this embodiment will be described.

  Conventionally, in the section using IEEE802.1D, route switching at the time of failure is slow, and it takes time to reconfigure the spanning tree.

  In the present embodiment, by setting the cost of the IEEE 802.1D usage section to be large, it is possible to prevent the spanning tree from being set through the IEEE 802.1D usage section, speed up switching and route change at the time of failure, It is possible to reduce the possibility of congestion and the possibility of missing frames.

(Sixth embodiment)
Hereinafter, a sixth embodiment of the present invention will be described in detail with reference to the drawings.

  In the sixth embodiment of the present invention, in the first embodiment, the discriminator identifies the version of BPDU, and through the tree selector, creates tree managers by the number of low-speed IEEE 802.1D usage sections. This corresponds to a case where a route that bypasses the section is provided at high speed when a failure occurs in the section using IEEE 802.1D.

  In the section using IEEE802.1D, the route switching at the time of failure is slow, and it takes time to reconfigure the spanning tree. Therefore, if a failure occurs in this section, it takes time to change the route or change the route, and congestion occurs. There was a problem that a frame was missing.

  In the present embodiment, tree managers are created as many as the number of IEEE 802.1D usage sections, and a tree in which the cost of one different IEEE 802.1D usage section is set large is created for the IEEE 802.1D usage section, due to a failure or the like. When it becomes necessary to detour the section, a detour operation can be performed at high speed by using a tree with a large cost for the section, and congestion and frame loss can be prevented.

  Referring to FIG. 35, the sixth embodiment of the present invention is different in that there are as many tree managers 1151 as the number of IEEE802.1D usage sections in FIG. 8 in the first embodiment of the present invention.

  In addition to the function of the tree manager 1151 shown in FIG. 8 in the first embodiment of the present invention, the tree manager 1151 is a BPDU frame in which the received BPDU frame conforms to IEEE802.1D by a version field or other means. Is confirmed, an 802.1D frame reception notification is transmitted to the tree selector 116 for the BPDU frame. The node ID of the node that has transmitted the 802.1D-compliant BPDU frame is written in the 802.1D frame reception notification.

  The tree manager 1152 and the tree manager 1153 are tree managers similar to the tree manager 1151. Thereafter, the tree manager 1151 is used to describe the tree manager 1151 to the tree manager 1153, but the description of the tree manager 1151 can be similarly applied to the tree manager 1152 to the tree manager 1153 unless otherwise specified. It is.

  Tree managers are created by the tree selector 116 by the number of IEEE802.1D usage sections. Accordingly, there is a possibility that one or more tree managers may increase to infinity. In FIG. 35, the tree managers are collectively expressed as tree managers 1151 to 1153.

  The tree selector 116 receives an 802.1D frame reception notification from any of the tree managers 1151 to 1153 in addition to the function of the tree selector 116 shown in FIG. 8 in the first embodiment of the present invention. A new tree manager, a function for notifying other nodes in the network of the 802.11D usage node, and a notification of the 802.1D usage node transmitted from the other node. It has a function to generate a manager.

  In addition to the above function, the tree selector 116 detects that the 802.1D usage node has become available for some reason, such as version upgrade, and deletes the tree manager, and information related to the deletion Is transmitted to other nodes in the network, and the tree manager is deleted based on the deletion information notified from the other nodes.

  FIG. 36 is a configuration diagram of a tree 67 that is created by a normal procedure of IEEE 802.1w with the node 11 as a root node.

  FIG. 37 is a configuration diagram of a tree 68 that is created with the node 11 as a root node and the cost of the tree 21 being increased. This tree is also used when a failure occurs in the link 21.

  FIG. 38 is a configuration diagram of a tree 69 that is created by setting the node 11 as a root node and increasing the cost of the tree 22. This tree is also used when a failure occurs in the link 22.

  FIG. 39 is a configuration diagram of a tree 70 that is created by setting the node 11 as a root node and increasing the cost of the tree 24. This tree is also used when a failure occurs in the link 24.

  Next, referring to FIGS. 36 to 39, the node 12 in FIGS. 36 to 39 is a node that does not support IEEE 802.1w, and IEEE 802.1D is used in the link 21, the link 22, and the link 24. The operation of the case will be described. The root node is assumed to be node 11.

  First, a spanning tree 67 shown in FIG. 36 is formed by a normal procedure according to IEEE 802.1w. At this time, since the node 12 is a node that does not support 802.1w, the node 12 transmits a BPDU frame to which a protocol ID of IEEE 802.1D is added.

  When the node 11 receives the BPDU with the IEEE802.1D protocol ID added from the node 12, the node 11 generates a new tree manager, and the unique tag calculated from the link ID, the node ID, and the like is generated for the tree manager. An ID is assigned, and the creation of a new group is broadcast to all nodes by a GVRP frame or other frames. Here, it is assumed that a tag ID 48 is assigned as a new tag ID. At this time, the cost of the link 21 is set large.

  The nodes 12 to 16 receive and transfer the new group creation notification transmitted from the node 11, generate a tree manager, and start exchanging BPDUs. A tag of tag ID 48 is added to a BPDU exchanged between newly created tree managers. The spanning tree created here is a tree 68.

  When the node 13 receives the BPDU to which the IEEE 802.1D protocol ID is added from the node 12, the node 13 generates a new tree manager, and the unique tag calculated from the link ID, the node ID, and the like for the tree manager An ID is assigned, and the creation of a new group is broadcast to all nodes by a GVRP frame or other frames. Here, it is assumed that a tag ID 49 is assigned as a new tag ID. At this time, the cost of the link 22 is set large.

  The nodes 11 to 12 and the nodes 14 to 16 receive and transfer the new group creation notification transmitted from the node 13, generate a tree manager, and start exchanging BPDUs. A tag of tag ID 49 is added to the BPDU exchanged between the newly created tree managers. The spanning tree created here is a tree 69.

  The node 15 receives the BPDU to which the IEEE 802.1D protocol ID is added from the node 12, generates a new tree manager, and generates a unique tag calculated from the link ID, the node ID, and the like for the tree manager. An ID is assigned, and the creation of a new group is broadcast to all nodes by a GVRP frame or other frames. Here, it is assumed that a tag ID 50 is assigned as a new tag ID. At this time, the cost of the link 24 is set large.

  The nodes 11 to 14 and the node 16 receive and transfer the new group creation notification transmitted from the node 15, generate a tree manager, and start exchanging BPDUs. A tag with tag ID 50 is added to the BPDU exchanged between the newly created tree managers. The spanning tree created here is a tree 70.

  In normal times, communication between the nodes is performed using the tree 67, and the trees 68 to 70 are not used.

  Here, if a failure occurs in the link 21, the node 11 detects the failure of the link 21 and immediately switches the tree used for transfer from the tree 67 to the tree 68. Also, a usage tag group change notification is broadcast to all nodes, and a tag used for transfer is notified to be switched to the tag ID 48.

  Each of the nodes 12 to 16 receives the usage tag group change notification transmitted from the node 11, inserts a tag with a tag ID 48 into the frame transmitted from the own node, and defines a tree used for transfer as a tree. Switch from 67 to tree 68.

  Note that switching to the tree 68 when a failure occurs in the link 21 may be performed by the node 12 instead of the node 11. If the node 12 detects a failure of the link 21, the tree used for forwarding is immediately switched from the tree 67 to the tree 68. Also, a usage tag group change notification is broadcast to all nodes, and a tag used for transfer is notified to be switched to the tag ID 48. The subsequent operation is the same as when the node 11 detects a failure.

  The tree 67 is reconfigured due to the failure of the link 21, but since reconfiguration according to 802.1D is performed on the link 21, there is a possibility that it takes time to complete the reconfiguration.

  In the present embodiment, when the link 21 fails, the tree used for transfer is immediately switched from the tree 67 to the tree 68, so that frame transfer can be continued without waiting for reconfiguration of the tree 67. it can.

  The operation in the case where a failure has occurred in the link 21 has been described above, but the above operation can be similarly applied to the case in which a failure occurs in the link 22 or the link 24.

  Next, the effect of this embodiment will be described.

  Conventionally, in the section using IEEE802.1D, route switching at the time of failure is slow, and it takes time to reconfigure the spanning tree.

  In the present embodiment, tree managers are created as many as the number of IEEE 802.1D usage sections, and a tree in which the cost of one different IEEE 802.1D usage section is set large is created for the IEEE 802.1D usage section, and the above-described tree is created due to a failure or the like. If there is a need to detour the section, switching to use a tree with a large cost for the section makes it possible to perform a detour operation at a high speed, thereby reducing the possibility of congestion and frame loss. Can be reduced.

(Seventh embodiment)
Next, a seventh embodiment of the present invention will be described in detail with reference to the drawings.

  In the seventh embodiment of the present invention, in the first embodiment, the failure detection frame is periodically transmitted at a short interval using the failure detector, and the failure detection frame does not arrive. Thus, failure detection is performed, and failure information is notified to the tree manager through the resource monitor and the tree selector.

  In IEEE802.1D and IEEE802.1w, a failure is detected by the arrival of a periodically transmitted HELLO frame. However, since the transmission interval of HELLO frames is long, it takes a long time to detect a failure.

  In the present embodiment, failure detection frames are periodically transmitted from the failure detector at short intervals, and high-speed failure detection is performed when a certain number of failure detection frames or a certain number of frames have not arrived.

  Referring to FIG. 40, in the seventh embodiment of the present invention, in FIG. 4 in the first embodiment, a failure detector 120 for detecting a failure by transmitting and receiving a failure detection frame is added. Is different.

  The failure detector 120 periodically transmits a failure detection frame to the adjacent node through the frame transfer unit 111, and receives the failure detection frame transmitted from the adjacent node from the frame transfer unit 111, A failure detection notification is transmitted to the resource monitor 119 when failure detection frames sent from a node cannot be received for a certain period of time or when a failure detection frame of a certain number or more has not been received.

  In addition to the function of the resource monitor 119 in the first embodiment, the resource monitor 119 has a function of receiving a failure notification from the failure detector 120 and transferring the failure notification to the tree selector 116.

  In addition to the function of the tree selector 116 in the first embodiment, the tree selector 116 has a function of receiving a failure notification from the resource monitor 119 and transferring it to the tree managers 1151 and 1152.

  In addition to the functions of the tree managers 1151 and 1152 in the first embodiment of the present invention, the tree managers 1151 and 1152 receive a failure notification from the tree selector 116 and re-span the spanning tree according to IEEE 802.1w or IEEE 802.1D. Make the configuration.

  Next, with reference to FIG. 40, an operation example when the node 11 detects a failure of the link 21 in the present embodiment will be described.

  The failure detector 120 periodically transmits a failure detection frame to the adjacent nodes 12 and 15 via the frame transfer device 111 and the link 21 or the link 24.

  The failure detector 120 also receives a failure detection frame transmitted from the adjacent nodes 12 and 15 via the link 21 or link 24 and the frame transfer device 111. At this time, the failure detector 120 can also identify the ID of the port where the failure detection frame has arrived.

  When a failure detection frame arrives, the failure detector 120 operates a timer of the port where the failure detection frame has arrived, and sets a notification to be transmitted when a certain time has elapsed.

  When the failure detector 120 receives the failure detection frame, the failure detector 120 resets a timer set for each port. For example, when a failure detection frame is received from the link 21, the timer of the port to which the link 21 is connected is reset. When a failure detection frame is received from the link 24, the timer of the port to which the link 24 is connected is reset.

  Here, in the failure detector 120, if a failure detection frame does not arrive for a certain time or more due to a link failure or the like, a timeout occurs because the timer is not reset. When a timeout occurs, the failure detector 120 recognizes that some failure has occurred, and notifies the tree manager 1151 and the tree manager 1152 of the occurrence of shogi via the resource monitor 119 and the tree selector 116.

  Receiving the notification, the tree manager 1151 and the tree manager 1152 immediately reconfigure the spanning tree by assuming that the failed port is unusable, and avoiding the failure.

  Next, the effect of this embodiment will be described.

  Conventionally, since the transmission interval of the HELLO frame used in the spanning tree protocol is long, high-speed failure detection cannot be performed.

  In the present embodiment, by adding a failure detector that transmits and receives a failure detection frame at a short interval, it is possible to detect a failure faster than the HELLO frame.

  As a result, it is possible to reduce the possibility of congestion and frame loss.

(Embodiment 8)
Hereinafter, an eighth embodiment of the present invention will be described in detail with reference to the drawings.

  The eighth embodiment of the present invention is configured such that, in the first embodiment, the tag and spanning tree used for each destination node can be switched, and the destination node can be set as the root node. .

  When a frame is transmitted in a network in which IEEE 802.1D and IEEE 802.1w are operating, the lowest cost route to the destination is not always selected, an unused link appears, and the load is concentrated on the root node There was a problem that the network stopped for a long time when the root node failed.

  In this embodiment, frames are transferred using a tree whose destination is the root node, so that frame transmission at the lowest cost to the destination is improved, link utilization is improved, and resistance to root node failures is enhanced. Is realized.

  Referring to FIG. 41, in the eighth embodiment of the present invention, in FIG. 8 in the first embodiment, the frame forwarder 111 becomes the frame forwarder 111γ, the forwarding table 114 becomes the forwarding table 114γ, The difference is that tree managers 1151 are installed by the number of nodes existing in the network after changing the operation, and become tree managers 1151γ to 1153γ.

  FIG. 41 shows a case where the present embodiment is applied to the node 11 shown in FIG.

  The frame transfer unit 111γ receives the frame received from the link 21 or link 23 and the tag insertion unit 112 according to the description of the output port 1142 of the forwarding table 114γ, and the tag deletion unit 113 or the tree selector 116. Forward to. At this time, if the description of the output port 1142 is an initial value, the received frame is discarded.

  The forwarding table 114γ is a forwarding table similar to the forwarding table 114.

  The tree manager 1151γ has the same configuration as the tree manager 1151, but is different in function and operation. Thereafter, the tree manager 1151γ to the tree manager 1153γ will be used to describe the tree manager 1151γ. However, the description of the tree manager 1151γ applies similarly to the tree manager 1152γ to the tree manager 1153γ unless otherwise specified. Is possible.

  FIG. 42 is a configuration example of the forwarding table 114γ that determines an output port using a forwarding tag as a key.

  The tag field 1141 is a tag field similar to the tag field in the forwarding table 114.

  The output port 1142 is an output port similar to the output port in the forwarding table 114, and is referred to by the frame transfer device 111γ when transferring a unicast frame.

  In FIG. 42, “END” of the output port 1142 is an identifier indicating that the output port entry corresponding to the destination node ID is the own node in the edge node. For example, the output port field for the node identifier entry equal to the own node ID Described in

  FIG. 43 is a block diagram showing the configuration of the tree manager 1151γ.

  When a spanning tree is configured as shown in FIG. 23, the port of each node is a root port (root port) or a designated port (designated port) as shown in FIG. 54 according to the configuration of the spanning tree. Alternatively, it is determined as an alternate port. In FIG. 54, the root port is indicated by (R), the designated port is indicated by (D), and the alternate port is indicated by (ALT). Of course, the type of the port varies depending on the configuration of the spanning tree (the position of the root node).

  When the tag of the tag group 41 is added to the BPDU frame transmitted / received by the tree manager 1151γ, the tree controller 11514γ determines the root port (Root Port) by the IEEE802.1w or IEEE802.1D protocol. In the present embodiment, the root port determined as the output port 1142 is set in the entry of the tag group 41 in the forwarding table 114γ. In this case, at least six BPDU frames are transmitted and received.

  Here, if the root port does not exist (node 11 in FIG. 54), the output port 1142 column is set as the port addressed to the own node, and the tagged frame of the tag group 41 input to the frame transfer unit is set. , Transfer to the tag deleter 113

Next, the unicast frame transfer operation will be described using unicast frame transfer from the node 13 to the node 11 as an example with reference to FIGS.
FIG. 44 is a table showing the status of port settings and forwarding table settings for each node in the spanning tree 61. Note that FIG. 44 does not show the actual table contents, but merely shows the setting status.

  In the state shown in FIG. 23, the tree 61 which is the tree of the tag group 41 has already been constructed and is stable, and the root ports in the nodes 11 to 16 are as indicated by the root port 6102 in FIG. Based on this result, it is assumed that the output port in the forwarding table 114 of each node is determined as indicated by the output port 1142 in FIG.

  Here, the tree 61 represents a tree having the node 11 as a root node. A tag 41 represents an ID (value) of a tag indicating the tree 61. That is, adding the tag 41 to the frame means transferring the frame using the tree 61. Thus, each of the tree number, the node number, and the tag number indicates that the first digit corresponds. For example, the tree 62 indicates a tree having the node 12 as a root, and the tag 42 indicates an ID (value) of a tag indicating the tree 62. This is the same in the following description.

  In addition, here, as an example, as described above, a case has been described in which different numbers are assigned to trees, nodes, and tags, but in addition, the same number (ID) is assigned to each corresponding tree, node, and tag. At the same time, it is also possible to manage each correspondence relationship by clarifying it.

  First, the node 13 adds a tag of the tag group 41 to the unicast frame addressed to the node 11 and transmits the frame. At this time, the output port of the frame of the tag group 41 in the node 13 is designated as the port on the link 22 side which is the root port of the tree 61. Therefore, the frame is output to the link 22 side.

  When the node 12 receives a frame from the link 22, the node 12 searches the forwarding table using the tag group 41 as a key, and obtains a port on the link 21 side as an output port. Then, the received frame is output to the link 21 side.

  When the node 11 receives the frame from the link 21, the node 11 confirms that the node 11 is addressed to the node 11 and transfers the frame to the tag deletion unit 113.

  With the above operation, it becomes possible to transfer a unicast frame from the node 13 to the node 11 by the lowest cost path using the tag of the tag group 41 and the spanning tree 61.

  Next, effects of the above-described eighth embodiment will be described.

  Conventionally, the lowest cost route to a destination has not always been selected, but in this embodiment, the lowest cost route to a destination is determined by transferring a frame using a tree in which the destination is a root node. You can choose.

  Conventionally, while the link utilization rate is low, there is a problem that the load is concentrated near the root node. In this embodiment, the link utilization rate is increased by setting multiple systems of spanning trees with different root nodes. It is possible to distribute the load around the root node without concentrating the load.

Furthermore, conventionally, there has been a problem that the tree construction at the time of the root node failure takes time and the network stops during that time, but in this embodiment, by transferring the frame using the tree whose destination is the root node, Since frames other than the frame destined for the root node are not disabled for a long time due to the root node failure, it is possible to avoid a network stop due to the root node failure. As a result, it is possible to reduce the possibility of congestion.
(Embodiment 9)
Hereinafter, a ninth embodiment of the present invention will be described in detail with reference to the drawings.

  In the ninth embodiment of the present invention, in the eighth embodiment, a plurality of broadcast output ports are described in the forwarding table in addition to a normal unicast output port, and a broadcast frame is transferred. It has a configuration that can.

  In this embodiment, by transmitting a broadcast frame using a tree in which the source node of the broadcast frame is a root node, the broadcast frame can be transmitted to each node through the shortest path, and high-speed transfer is possible become.

  Referring to FIG. 45, in the ninth embodiment of the present invention, in FIG. 41 in the eighth embodiment, the frame forwarder 111 is the frame forwarder 111β, the forwarding table 114 is the forwarding table 114β, and the tree manager. The difference is that 1151-1153 are tree managers 1151β-1153β.

  FIG. 45 shows a case where the present embodiment is applied to the node 11 shown in FIG.

  The frame transfer unit 111β transfers the frame received from the link 21 or the link 24 and the tag insertion unit 112 to the link 21 or the link 24 and the tag deletion unit 113 or the tree selector 116 in accordance with the description of the forwarding table 114β.

  At this time, if the input frame is a unicast frame, the received frame is transferred to the port described in the output port 1142 of the forwarding table 114β.

  If the input frame is a broadcast frame, the received frame is copied and transferred to a plurality of ports and tag deleters described in the broadcast output port 1144 of the forwarding table 114β. If the initial value is set for the broadcast output port, the received frame is transferred only to the tag deleter.

  The broadcast frame and the unicast frame are discriminated based on the transmission destination MAC address 3201, the extension tag priority 5003, or the extension tag information field 5004.

  The forwarding table 114β is a forwarding table in which the broadcast output port 1144 column is added to the forwarding table 114. The broadcast output port 1144 column indicates a port that becomes a transfer destination when a broadcast signal transmitted from the node indicated by the corresponding tag ID is received. An example of the forwarding table in this embodiment is shown in FIG. In FIG. 46, a link name is used to indicate the transfer destination port. FIG. 46 illustrates the forwarding table 114β of the node 11 in the network having the physical topology illustrated in FIGS. As described above, in the present invention, the tree used for frame transfer differs for each destination node. For example, a frame addressed to the node 11 is transferred using the tree shown in FIG. 23, and a frame addressed to the node 12 is transferred using the tree shown in FIG.

  Here, the way of viewing FIG. 46 will be described below by taking as an example the case of broadcasting a frame to which a tag 42 is added. The addition of the tag 42 indicates that it is a broadcast frame transmitted from the node 12. In the present invention, the broadcast frame transmitted from the node 12 is transferred using the tree shown in FIG. Therefore, when receiving the broadcast frame, the node 11 needs to transfer it to the link 23 side. Based on such a concept, FIG. 46 is created.

  The tree manager 1151β has the same configuration as the tree manager 1151, but differs in its function and operation. Hereinafter, the tree manager 1151β to the tree manager 1153β will be representatively described using the tree manager 1151β, but the description of the tree manager 1151β applies similarly to the tree manager 1152β to the tree manager 1153β unless otherwise specified. Is possible.

  FIG. 46 is a configuration example of the forwarding table 114β that determines an output port using a tag as a key.

  The tag field 1141 is a tag field similar to the tag field in the forwarding table 114.

  The output port 1142 is an output port similar to the output port in the forwarding table 114, and is referred to by the frame transfer unit 111β when transferring a unicast frame.

  The broadcast output port 1144 is an output port that is referred to by the frame transfer unit 111β when transferring a broadcast frame. In this column, a plurality of ports are described. If two or more ports are described, the frame is copied after the number of ports described is transferred. If the value set in this field remains the initial value when referenced by the frame transfer unit 111β, the frame transfer unit 111β transfers the frame only to the tag deletion unit.

  Note that (END) in FIG. 46 indicates that the identifier “END” is described when the node is an edge node.

  FIG. 47 is a block diagram showing the configuration of the tree manager 1151β.

  When the tag of the tag group 41 is added to the BPDU frame transmitted / received by the tree manager 1151β, the tree controller 11514β has the root port (Root Port) and the designated port (Designated Port) by the IEEE802.1w or IEEE802.1D protocol. ). In the present embodiment, the root port determined as the output port 1142 and one or more determined designated ports as the broadcast output port 1144 are set in the entry of the tag group 41 of the forwarding table 114β. .

  If the root port does not exist, the output port 1142 column is set as the port addressed to the own node, and the tagged frame of the tag group 41 input to the frame transfer unit is transferred to the tag deletion unit 113. To do.

  If there is no designated port, the broadcast output port 1144 column is set to the initial value of the table.

  Next, the broadcast frame transfer operation will be described with reference to FIGS.

  FIG. 48 is a table showing port settings and forwarding table settings for each node in the spanning tree 61.

  In the initial state, the tree 61 that is the tree of the tag group 41 has already been constructed and is stable, and the root port and the designated port in each of the nodes 11 to 16 are the root port 6102 in FIG. As shown in the designated port 6104, the output port and the spare output port in the forwarding table 114β of each node are determined as shown in the output port 1142 and the broadcast output port 1144 in FIG. It shall be.

  First, the node 11 adds the tag of the tag group 41 to the broadcast frame and transmits it. At this time, the output ports of the broadcast frame of the tag group 41 in the node 11 are designated as the ports on the link 21 side and the link 23 side which are designated ports of the tree 61. Therefore, the frame is output after being copied to the link 21 side and the link 23 side and further to the tag deleter.

  When the node 12 receives the broadcast frame from the link 21, the node 12 searches the forwarding table using the tag group 41 as a key, and obtains the port on the link 22 side as the broadcast output port. Then, the received broadcast frame is output to the link 22 side and the tag remover.

  When the node 13 receives the broadcast frame from the link 22, the node 13 searches the forwarding table using the tag group 41 as a key, and obtains an initial value as the broadcast output port. Then, the received frame is output to the tag remover.

  When the node 14 receives the broadcast frame from the link 23, the node 14 searches the forwarding table using the tag group 41 as a key, and obtains the port on the link 26 side as the broadcast output port. Then, the received frame is output to the link 26 side and the tag deleter.

  When the node 15 receives the broadcast frame from the link 26, the node 15 searches the forwarding table using the tag group 41 as a key, and obtains the link 27 side port as the broadcast output port. Then, the received frame is output to the link 27 side and the tag deleter.

  When the node 16 receives the broadcast frame from the link 27, the node 16 searches the forwarding table using the tag group 41 as a key, and obtains an initial value as the broadcast output port. Then, the received frame is output to the tag remover.

  With the above operation, the broadcast frame output from the node 11 is transferred to each node on the network through the lowest cost path.

  Next, effects of the ninth embodiment described above will be described.

  Conventionally, the lowest cost route to a destination is not always selected at the time of broadcasting, but in this embodiment, by transmitting a broadcast frame using a tree in which a transmission source node is a root node, The broadcast frame can be transferred by selecting the lowest cost route to the node.

  Conventionally, while the link utilization rate is low, there is a problem that the load is concentrated near the root node. In this embodiment, the link utilization rate is increased by setting multiple systems of spanning trees with different root nodes. It is possible to distribute the load around the root node without concentrating the load.

  Further, conventionally, there has been a problem that it takes time to construct a tree when a root node fails, and the network stops during that time. In this embodiment, a broadcast frame is transferred using a tree in which the transmission source node is the root node. As a result, a broadcast frame other than a frame whose root node is a transmission source node is not disabled for a long time due to the influence of the root node failure, so that a network stop due to the root node failure can be avoided. As a result, it is possible to reduce the possibility of congestion.

(Embodiment 10)
Hereinafter, a tenth embodiment of the present invention will be described in detail with reference to the drawings.

  In the tenth embodiment of the present invention, in the eighth embodiment, two output ports are described in the forwarding table, and when one output port cannot be used due to a failure or the like, the other port can be used. In addition, the failure detector according to the seventh embodiment is used to perform high-speed failure detection.

  In this embodiment, high-speed failure recovery is performed by transferring a unicast frame using a tree whose destination is a root node and registering an alternative output port determined by the spanning tree in the forwarding table in advance. .

  Referring to FIG. 49, in the eighth embodiment of the present invention, in FIG. 41 in the fourth embodiment, the frame forwarder 111 is the frame forwarder 111α, the forwarding table 114 is the forwarding table 114α, and the tree manager. 1151 to 1153 are tree managers 1151α to 1153α, and the point that the failure detector 120 described in the seventh embodiment is added is different.

  FIG. 49 shows a case where the present embodiment is applied to the node 11 shown in FIG.

  The frame transfer unit 111α transfers the frame received from the link 21 or the link 23 and the tag insertion unit 112 to the link 21 or the link 23 and the tag deletion unit 113 or the tree selector 116 in accordance with the description of the forwarding table 114α.

  At this time, if the resource monitor 119 detects that there is a failure in the port described in the output port 1142 of the forwarding table 114α, the received frame is transferred to the port described in the spare output port 1143. . If a failure of the port described in the output port 1142 is detected, but the description of the spare output port is an initial value (or not set), the received frame is discarded.

  The forwarding table 114α is a forwarding table in which a spare output port 1143 column is added to the forwarding table 114.

  The tree manager 1151α has the same configuration as the tree manager 1151, but is different in function and operation. Hereinafter, the tree manager 1151α to the tree manager 1153α will be used to describe the tree manager 1151α, but the description of the tree manager 1151α is similarly applied to the tree manager 1152α to the tree manager 1153α unless otherwise specified. Is possible.

  FIG. 50 is a configuration example of the forwarding table 114α of the node 12 in FIG. 23 that determines an output port using a tag as a key.

  The tag field 1141 is a tag field similar to the tag field in the forwarding table 114.

  The output port 1142 is an output port similar to the output port in the forwarding table 114.

  The spare output port 1143 is a field for describing an output destination port to be used when the port described in the output port 1142 becomes unavailable. When the frame transfer unit 111α detects that the port described in the output port 1142 is unavailable, the frame transfer unit 111 transfers the frame to the port described in the spare output port 1143 of the entry.

  FIG. 51 is a block diagram showing the configuration of the tree manager 1151α.

  When the tag of the tag group 41 is added to the BPDU frame transmitted and received by the tree manager 1151α, the tree controller 11514α determines the root port (Root Port) and the alternate port (Alternate Port) by the IEEE802.1w protocol. In the present embodiment, the root port determined as the output port 1142 and the alternate port determined as the backup output port 1143 are set in the entry of the tag group 41 of the forwarding table 114α.

  Here, if there is no root port, the output port 1142 column is set as the local node destination port, and the tagged frame of the tag group 41 input to the frame transfer unit is transferred to the tag deletion unit 113. To do.

  If there is no alternate port, the spare output port 1143 column is set to the initial value of the table.

Next, with reference to FIG. 23 and FIG. 52, the unicast frame transfer operation when a failure occurs in the link will be described by taking the case where a failure occurs in the link 21 as an example.
FIG. 52 is a table showing the port settings and forwarding table settings of each node in the spanning tree 61.

  In the initial state, the tree 61 which is the tree of the tag group 41 is already constructed and stable, and the root port and the alternate port in each of the nodes 11 to 16 are the root port 6102 and the alternate port in FIG. 6103, and as a result, the output ports and spare output ports in the forwarding table 114α of each node are decided as shown in the output port 1142 and spare output port 1143 in FIG. To do.

  First, the node 13 adds a tag of the tag group 41 to the unicast frame addressed to the node 11 and transmits the frame. At this time, the output port of the frame of the tag group 41 in the node 13 is designated as the port on the link 22 side which is the root port of the tree 61. Therefore, the frame is output to the link 22 side.

  Assume that a failure has occurred in the link 21 in this state.

  When receiving a frame from the link 22, the node 12 searches the forwarding table using the tag group 41 as a key, and obtains a port on the link 21 side as an output port and a port on the link 24 side as a spare output port. Then, the reception frame is output to the link 21 side, but since the failure detection information of the link 21 is received from the resource monitor, the reception frame is output to the link 24 side which is a spare output port.

  When receiving a frame from the link 24, the node 15 searches the forwarding table using the tag group 41 as a key, and obtains a port on the link 26 side as an output port and a port on the link 24 side as a spare output port. Then, it is confirmed that no failure has occurred on the link 26 side, and the received frame is output to the link 26 side.

  When receiving a frame from the link 26, the node 14 searches the forwarding table using the tag group 41 as a key, and obtains an initial value as a port on the link 23 side as an output port and a spare output port. Then, it is confirmed that no failure has occurred on the link 23 side, and the received frame is output to the link 23 side.

  When the node 11 receives the frame from the link 23, the node 11 confirms that it is addressed to itself and transfers the frame to the tag deleting unit 113.

  As a result of the above operation, when a failure occurs in the link 21, referring to a preset standby output port, as a route for transferring a unicast frame from the node 13 to the node 11, the link 22, the link 23, The route via the link 24, the link 26, and the link 23 is immediately selected, and the frame transfer to the node 11 is continued. Therefore, it is possible to bypass at a high speed, and avoid network congestion.

  Next, effects of the tenth embodiment described above will be described.

  Conventionally, the lowest cost route to the destination has not always been selected, but in this embodiment, the lowest cost to the destination is transferred by transferring a unicast frame using a tree in which the destination is the root node. A route can be selected.

  Conventionally, while the link utilization rate is low, there is a problem that the load is concentrated near the root node. In this embodiment, the link utilization rate is increased by setting multiple systems of spanning trees with different root nodes. It is possible to distribute the load around the root node without concentrating the load.

  Furthermore, conventionally, there has been a problem that the tree construction at the time of the root node failure takes time and the network stops during that time, but in this embodiment, by transferring the frame using the tree whose destination is the root node, Since frames other than the frame destined for the root node are not disabled for a long time due to the root node failure, it is possible to avoid a network stop due to the root node failure. As a result, it is possible to reduce the possibility of congestion.

  Furthermore, conventionally, there has been a problem that it takes time to switch the output destination port at the time of the root port side link failure, and during that time frame transfer stops. However, in this embodiment, preliminary output for the output link failure beforehand By setting the link in the forwarding table, it is possible to change the route at high speed when the root port side link, that is, the output link fails. As a result, it is possible to reduce the possibility of congestion.

  The function of each means, which is a component of the spanning tree constituent node in the network of the present invention, is realized not only by hardware, but also by a spanning tree reconfiguration program (application) that executes the function of each means described above. Program) 950 is loaded into the memory of the computer processing apparatus and the computer processing apparatus is controlled. The spanning tree reconfiguration program 950 is stored in a magnetic disk, a semiconductor memory, or other recording medium, loaded from the recording medium to a computer processing device, and controls the operation of the computer processing device, thereby realizing each function described above. To do.

  Although the present invention has been described with reference to the preferred embodiments and examples, the present invention is not necessarily limited to the above embodiments and examples, and various modifications can be made within the scope of the technical idea. Can be implemented.

It is a figure which shows the structural example of the Ethernet (R) frame with the conventional VLAN tag. It is a figure which shows the structural example of the Ethernet (R) frame with an extension tag of this invention. It is a figure which shows the other structural example of the Ethernet (R) frame with an extension tag of this invention. It is a figure which shows the structural example of the expansion tag storage area of this invention. It is a format figure which shows the frame structure of Configuration BPDU frame in this invention. It is a format figure which shows the frame structure of Topology Change Notification BPDU frame in this invention. It is a block diagram which shows the structure of the 1st Embodiment of this invention. It is a block diagram which shows the structure of the node 11 in the 1st Embodiment of this invention. It is a table | surface which shows the structural example of the forwarding table 114 in the 1st Embodiment of this invention. It is a block diagram which shows the structure of the tree manager 1151 in the 1st Embodiment of this invention. It is a block diagram which shows the structure of the tree selector 116 in the 1st Embodiment of this invention. 5 is a flowchart showing the operation of the main controller 1164 in the first embodiment of the present invention. It is a table | surface which shows the structural example of the tag table 117 in the 1st Embodiment of this invention. It is a block diagram which shows the structure of the spanning tree 51 before the node 700 is added in the 1st Embodiment of this invention. It is a block diagram which shows the structure of the spanning tree 52 after the node 700 is added in the 1st Embodiment of this invention. It is a sequence diagram which shows exchange of the control frame in the 1st Embodiment of this invention. It is a block diagram which shows the structure of the tree selector 116 in the 2nd Embodiment of this invention. 12 is a flowchart showing the operation of the main controller 1164 in the second embodiment of the present invention. It is a flowchart which shows operation | movement of the main controller 1164 in the 3rd Embodiment of this invention. It is a block diagram which shows the structure of the node 11 in the 4th Embodiment of this invention. It is a table | surface which shows the structural example of the forwarding table 114 in the 4th Embodiment of this invention. It is a table | surface which shows the structural example of the tag table 117 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree 61 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree 62 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree 63 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree 64 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree 65 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree 66 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree 74 in the 4th Embodiment of this invention. It is a block diagram which shows the structure of the tree manager 1151 in the 5th Embodiment of this invention. It is a block diagram which shows the structure of the tree 71 in the 5th Embodiment of this invention. It is a block diagram which shows the structure of the tree 72 in the 5th Embodiment of this invention. It is a block diagram which shows the structure of the tree 73 in the 5th Embodiment of this invention. It is a block diagram which shows the structure of the node 11 in the 6th Embodiment of this invention. It is a block diagram which shows the structure of the tree 67 in the 6th Embodiment of this invention. It is a block diagram which shows the structure of the tree 68 in the 6th Embodiment of this invention. It is a block diagram which shows the structure of the tree 69 in the 6th Embodiment of this invention. It is a block diagram which shows the structure of the tree 70 in the 6th Embodiment of this invention. It is a block diagram which shows the structure of the node 11 in the 7th Embodiment of this invention. It is a block diagram which shows the structure of the node 11 in the 8th Embodiment of this invention. It is a table | surface which shows the structural example of the forwarding table 114γ in the 8th Embodiment of this invention. It is a block diagram which shows the structure of the tree manager 1151 (gamma) in the 8th Embodiment of this invention. It is a table | surface which shows the example of the setting condition of the tree 61 in the 8th Embodiment of this invention. It is a block diagram which shows the structure of the node 11 in the 9th Embodiment of this invention. It is a table | surface which shows the structural example of the forwarding table 114 (beta) in the 9th Embodiment of this invention. It is a block diagram which shows the structure of the tree manager 1151 (beta) in the 9th Embodiment of this invention. It is a table | surface which shows the example of the setting condition of the tree 61 in the 9th Embodiment of this invention. It is a block diagram which shows the structure of the node 11 in the 10th Embodiment of this invention. It is a table | surface which shows the structural example of the forwarding table 114 (alpha) in the 10th Embodiment of this invention. It is a block diagram which shows the structure of the tree manager 1151 (alpha) in the 10th Embodiment of this invention. It is a table | surface which shows the example of a setting condition of the tree 61 in the 10th Embodiment of this invention. It is a figure which shows the other structural example of the expansion frame in this invention. It is a figure explaining the port state of each node in the structure of the spanning tree shown in FIG.

Explanation of symbols

11-18 Node 21-30 Bidirectional Link 31 Active BPDU (Tag Group 41)
32 Backup BPDU (Tag Group 42)
33 Active BPDU (Tag Group 42)
34 Backup BPDU (Tag Group 41)
35 Use tag group change notification (GVRP, etc.)
41-50 Tag Group 51-52 Spanning Tree 61-73 Spanning Tree]
81-88 Bidirectional link 91-97 Client 111 Frame forwarder 111α Frame forwarder 111β Frame forwarder 111γ Frame forwarder 111 Tag inserter 113 Tag deleter 114 Forwarding table 114α Forwarding table 114β Forwarding table 114γ Forwarding table 116 Tree selector 117 Tag table 118 Setting interface 119 Resource monitor 1143 Spare output port 1144 Broadcast output port 1150 Sorter 1151 Tree manager 1151α Tree manager 1151β Tree manager 1151γ Tree manager 1152 Tree manager 1153 Tree manager 1161 Tag deleter 1162 GVRP transmitter 116 Tag inserter 1164 Main controller 1165 Stability timer 1166 Arrival interval timer 1167 Cost reference timer 1168 Function calculator 1169 Smoothing circuit 11511 Tag deleter 11512 BPDU transmitter 11513 Tag inserter 11514 Tree controller 11514α Tree controller 11514β Tree controller 11514γ Tree controller 11515 Tree table 11516 Cost operator 11641-11649 State 1164A State 1164B State 120 Fault detector 2800 TPID
2801 TCI
2802 Priority
2803 CFI
2804 VLAN ID
3508 Broadcast tag (extended tag)
5001 Extended tag identification area 5002 Extended tag information area 5003 Priority, tag type field 5004 Extended tag information field 6101 Node 6102 Root port 6103 Alternate port 6104 Designated port

Claims (18)

In a node that constitutes a spanning tree on a network in which multiple nodes are connected,
A node that generates a spanning tree in which each node in the network is the root node, and forwards the frame using the spanning tree in which the destination is the root node. In a node that constitutes a spanning tree on a network in which multiple nodes are connected,
Multiple tree managers that generate multiple spanning trees that operate independently;
A tag table that returns tags corresponding to the transfer tree;
A tag inserter that inserts a tag with a response from the tag table into the frame;
A tree selector that generates the same number of tree managers as the number of links that are using the slow protocol present in the network;
A forwarding table that records the frame transfer output destination for each destination;
A frame transfer device for transferring a frame to a transfer output destination specified in the forwarding table;
And a classifier that determines a transfer destination tree manager according to the tag. The tree selector is
The main controller in the tree selector that creates or deletes the tree manager,
A tag remover that removes the tag attached to the frame;
A GVRP transceiver for transmitting control frames;
The node according to claim 2 , further comprising a tag inserter for adding a tag to the frame. The tree manager
A tag remover that removes the tag attached to the frame;
A BPDU transceiver for transmitting and receiving BPDUs;
A tag inserter for adding tags to the frame;
A tree controller that creates a spanning tree according to the spanning tree protocol;
The node according to claim 2, further comprising: a tree table that holds parameters used in the spanning tree protocol. In a spanning tree configuration program that operates on each node that constitutes a spanning tree on a network in which a plurality of nodes are connected,
A spanning tree configuration program characterized by generating a spanning tree in which each node in a network is a root node and transferring a frame using a tree in which a destination is a root node. In a spanning tree configuration program that operates on each node that constitutes a spanning tree on a network in which a plurality of nodes are connected,
Multiple tree manager processes that generate multiple spanning trees that operate independently;
Tag table processing that returns a tag corresponding to the spanning tree used for forwarding;
Tag insertion processing for inserting a tag that has been answered from the tag table into the frame;
A tree selector process for generating the same number of tree managers as the number of root nodes existing in the network;
Forwarding table processing that records the frame transfer output destination for each destination,
Frame transfer processing for transferring a frame to a transfer output destination specified in the forwarding table;
And a classification process for determining a tree manager to which the frame is transferred according to the tag. The tree selector process is
The main controller process in the tree selector that creates or deletes the tree manager,
Tag deletion processing to delete the tag attached to the frame;
GVRP transmission / reception processing for transmitting a control frame for spanning tree switching;
The spanning tree configuration program according to claim 6 , wherein tag insertion processing for adding a tag to a frame is executed. The tree manager process is
Tag deletion processing to delete the tag attached to the frame;
BPDU transmission / reception processing for transmitting / receiving BPDUs;
Tag insertion processing for adding a tag to the frame;
Tree controller processing to create a spanning tree according to the spanning tree protocol;
The spanning tree configuration program according to claim 6 or 7 , wherein tree table processing for holding parameters used in the spanning tree protocol is executed. In the spanning tree configuration program that operates on each node that constitutes the spanning tree on a network in which a plurality of nodes are connected,
Multiple tree manager processes that generate multiple spanning trees that operate independently;
Tag table processing that returns tags corresponding to the transfer tree;
Tag insertion processing for inserting a tag that has been answered from the tag table into the frame;
A tree selector process that generates the same number of tree managers as the number of links that are using a slow protocol existing in the network;
Forwarding table processing that records the forwarding output destination of the frame for each destination,
Frame transfer processing for transferring a frame to a transfer output destination specified in the forwarding table;
And a sorting process for determining a tree manager of a frame transfer destination according to the tag. The tree selector process is
The main controller process in the tree selector that creates or deletes the tree manager,
Tag deletion processing to delete the tag attached to the frame;
GVRP transmission / reception processing for transmitting a control frame;
The spanning tree configuration program according to claim 9 , further comprising: a tag insertion process for adding a tag to the frame. The tree manager process is
Tag deletion processing to delete the tag attached to the frame;
BPDU transmission / reception processing for transmitting / receiving BPDUs;
Tag insertion processing for adding a tag to the frame;
Tree controller processing to create a spanning tree according to the spanning tree protocol;
11. The spanning tree configuration program according to claim 9, further comprising: a tree table process that holds parameters used in the spanning tree protocol. In a network system in which a forwarding path is set by spanning tree on a network in which multiple nodes are connected,
A network system characterized by generating a spanning tree in which each node in the network is a root node, and transferring a frame using the tree in which a destination is a root node. In a network system in which a forwarding path is set by spanning tree on a network in which multiple nodes are connected,
Multiple tree managers that generate multiple spanning trees that operate independently;
A tag table that returns tags corresponding to the transfer tree;
A tag inserter that inserts a tag with a response from the tag table into the frame;
A tree selector that generates as many tree managers as there are nodes in the network;
A forwarding table that records the frame transfer output destination for each destination;
A frame transfer device for transferring a frame to a transfer output destination specified in the forwarding table;
And a classifier that determines a tree manager of a transfer destination of the frame according to the tag. In a network system in which a forwarding path is set by spanning tree on a network in which multiple nodes are connected,
Multiple tree managers that generate multiple spanning trees that operate independently;
A tag table that returns tags corresponding to the transfer tree;
A tag inserter that inserts a tag with a response from the tag table into the frame;
A tree selector that generates the same number of tree managers as the number of links that are using the slow protocol present in the network;
A forwarding table that records the frame transfer output destination for each destination;
A frame transfer device for transferring a frame to a transfer output destination specified in the forwarding table;
And a classifier that determines a tree manager of a transfer destination of the frame according to the tag. The network system according to claim 13 or 14 , wherein the forwarding table has a broadcast output port field. The network system according to claim 13 or 14 , wherein the forwarding table has a spare output port field. 15. The network system according to claim 13 , wherein an output destination port is determined using a port type determined by a spanning tree. The network system according to claim 17 , wherein the port type determined by the spanning tree is one of a root port and a designated port.

Images