JP4611863B2 - Loop detection method and loop detection apparatus - Google Patents

Description

  The present invention relates to a loop detection method and a loop detection device, and more particularly to a loop detection method and a loop detection device for detecting a loop when continuity is confirmed in a computer network.

  As a loop detection technique in a computer network, a method of generating a loop detection message at a node on the computer network and transferring it to an adjacent node as described in Patent Document 1 or Patent Document 2 is known. ing. In this method, the node examines the contents of the loop detection message received from the adjacent node, and determines that a loop has occurred if its own name is included. On the other hand, if the loop detection message does not include its own name, the receiving node adds its own name to the loop detection message and further transfers it.

  Patent Document 3 and Patent Document 4 describe an invention for detecting a loop for each port.

US Pat. No. 5,321,812 JP 05-134993 A JP 2004-320248 A JP 2005-033360 A

  The technique described in Patent Document 1 or Patent Document 2 requires a dedicated protocol for transferring and processing the loop detection message. In order to exchange information using a dedicated protocol, it is necessary to recognize where the same protocol is implemented on all nodes, or where the node that implements the protocol is located in some way if not implemented. . In addition, since the name included in the loop detection message is assigned to each node, it is impossible to know which interface has detected the loop when there are multiple interfaces in the node. . Furthermore, there is no description of what to do after detecting the loop.

In the technique described in Patent Document 3 or Patent Document 4, since the port number is stored in the apparatus, there is a possibility that processing of the apparatus may be delayed. Patent Document 4 describes that a port that detects a loop is shifted to a blocking state.
An object of the present invention is to provide a method and a loop detection apparatus that enable loop detection without being aware of the adjacent relationship.

  The above-described problem is that a step of transmitting a first loop detection frame including a first passage address describing an address of the first interface from the first interface, and a first passage address or a second passage address. Receiving a second loop detection frame containing from the second interface, checking whether the first pass address or the second pass address includes the address of its own interface, the first pass address or When the address of its own interface is included in the second passage address, it can be solved by a loop detection method including the step of blocking the first interface or the second interface.

  In addition, a first loop detection frame including a first passage address describing the address of the first interface is transmitted from the first interface, and a second loop including the first passage address or the second passage address is transmitted. Resolved by the loop detection device that receives the detection frame from the second interface and blocks the first interface or the second interface when the address of the own interface is included in the first passage address or the second passage address can do.

  Even if a new protocol is not implemented in the device on the network, the loop detection frame can be distributed over the entire path where the loop may occur.

  Therefore, a desired object can be achieved by providing a function for generating, detecting, and transferring a loop detection message only on a specific portion.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings using examples.

  A first embodiment will be described with reference to FIGS. Here, FIG. 1 is a block diagram illustrating the configuration of a layer 2 (L2) network. FIG. 2 is a block diagram illustrating the internal configuration of a switch having a loop detection function. FIG. 3 is a processing flowchart for explaining the operation of the loop detection frame handling block. FIG. 4 is a flowchart for explaining fault handling.

  In FIG. 1, the layer 2 (L2) network is configured by L2 switches 101 to 105. Each L2 switch is assigned a MAC address for each interface (port). For example, in the L2 switch 101, it is assumed that the MAC address 101-1 is assigned to the interface 1 and 101-2 is assigned to the interface 2.

  The loop detection frame is generated first by any L2 switch in the network. In this embodiment, the switch 101 generates the first loop detection frame. The trigger for sending the loop detection frame may be performed by a maintenance person or a timer.

  The loop detection frames 111 to 115 are transferred through the network. In particular, the loop detection frame 111 is sent from the L2 switch 101 toward the L2 switch 102. Similarly, the loop detection frame 112 is directed from the L2 switch 102 to the L2 switch 104, the loop detection frame 113 is directed from the L2 switch 101 to the L2 switch 103, and the loop detection frame 114 is directed from the L2 switch 104 to the L2 switch 105. The loop detection frame 115 is transmitted from the L2 switch 105 toward the L2 switch 102.

  The configuration inside the frame will be described using the loop detection frame 111. The frame includes a destination MAC address 121, a source MAC address 122, a type field 123, a pointer field 124, and a passing MAC address 125 that sequentially describes the transmission interface that has passed. Here, the passing MAC address 125 is generally a set of a plurality of addresses. In the loop detection frame, the broadcast address is registered in the destination MAC address 121, the MAC address of the L2 switch as the starting point is registered in the transmission source MAC address 122, and the type value indicating that it is a loop detection frame is inserted in the type field 123 Has been. The pointer field 124 indicates how far the MAC address is currently registered. In FIG. 1, an arrow with the pointer field as a base point indicates the current value of the registered MAC address.

  Here, the loop notification frame 131 will be described. Although the data structure is the same as that of the loop detection frame, the destination address 121 is rewritten with the MAC address of the switch that is the starting point of the loop detection frame, and the source address 122 is rewritten with the MAC address of the switch that detected the loop. Is rewritten to a value indicating a loop notification frame. Unlike the loop detection frame, the address of the own interface that has received the loop detection frame is added to the passing MAC address 125.

  The header configuration of the loop detection frame and the loop notification frame is the same as that of a normal Ethernet (registered trademark) frame header, and the L2 switch can transfer as a normal Ethernet frame.

  In FIG. 1, a solid line connection indicates an actual network physical connection, and a dotted arrow indicates a loop detection frame transfer path. The loop detection frames 111 and 113 are transmitted by broadcast to the L2 switches 102 and 103, starting from the L2 switch 101. The loop detection frame 112 is broadcasted from the L2 switch 102 receiving the loop detection frame 111 to the L2 switch 104. Thereafter, the L2 switches 104 and 105 operate similarly. As a result, the L2 switch 102 that has received the loop detection frame 115 detects the MAC address of its own transmission interface 102-2 in the passing MAC address 125. The presence of the MAC address of its own transmission interface in the passing MAC address means that a loop is formed. Therefore, the L2 switch 102 transmits the loop notification frame 131 to the L2 switch 101. The L2 switch 101 can grasp the loop path from the passing MAC address sandwiched between the transmission address and the reception address of the L2 switch 102 included in the loop notification frame 131.

  In the configuration of FIG. 1, the L2 switch 102 broadcasts a loop detection frame to the L2 switch 105 as indicated by a broken line arrow. However, the frame configuration is omitted for simplicity of illustration. This is common for counterclockwise transfers. This counterclockwise path also forms a loop.

  In FIG. 2, a switch having a loop detection function excluding an interface includes a type field identification block 201, a loop detection frame handling block 202, an L2 switching processing unit 203, and an output frame selection block 204. The L2 switch processing unit 203 is a switching processing unit provided in a normal packet transfer device. In order to realize loop detection, a type field identification block 201, a loop detection frame handling block 202, and an output frame selection block 204 are newly added.

  The type field identification block 201 checks the type field in the header part of the Ethernet frame. Here, since a dedicated type value is assigned to the loop detection frame, only the detected loop detection frame is delivered to the loop detection frame handling block 202. The remaining packets are processed by the L2 switching processing unit 203. The loop detection frame handling block 202 actually analyzes, updates, and generates a loop detection frame. The output frame selection block 204 performs output control of the Ethernet frame and the loop detection frame switched by the L2 switching processing unit 203.

  The operation of the loop detection frame handling block in the L2 switch having the loop detection function will be described below with reference to FIGS.

  Returning to FIG. 1, loop detection frames 111 and 113 are output from the L2 switch 101. The loop detection frame 111 is received by the L2 switch 102, identified as a loop detection frame by the type field identification block 201 in FIG. 2, and delivered to the loop detection frame handling block 202.

  In FIG. 3, the frame handling block 202 recognizes the reception of the loop detection frame (S301), and confirms whether or not the MAC address assigned to the own switch is registered in the passing MAC address 125 (S302). Since the MAC address assigned to the switch 102 is not registered in the frame 111, the process proceeds to step 307, where the frame handling block 202 assigns its own transmission port address to the passing MAC address 125 of the received loop detection frame. To do. The frame handling block 202 performs the process of sending the loop detection frame that has been subjected to the adding process (S308), and ends. Specifically, the frame handling block 202 outputs the loop detection frame 112 newly assigned with the MAC address 102-2 and the loop detection frame assigned the MAC address 102-3 from the corresponding interfaces. If only the interface that received the loop detection frame has a port, the loop detection frame is discarded.

  The loop detection frame 112 is then received by the L2 switch 104 and output as the loop detection frame 114. Further, the loop detection frame 114 received by the L2 switch 105 becomes a loop detection frame 115 and is output toward the L2 switch 102.

  After receiving the loop detection frame 115, the L2 switch 102 similarly checks whether the MAC address assigned to its own switch is registered in step 302 according to the flow of FIG. As a result, it is detected that the MAC address 102-2 is registered, and it is recognized that the self-switch transmission frame has returned to the self-switch again, that is, that a loop has occurred. As a result, the process proceeds to the failure handling process in step 303. The fault handling process will be described later with reference to FIG.

  When it is determined that the loop notification frame needs to be transmitted in the fault handling process, the process returns to step 304 in FIG. In step 304, the frame handling block 202 assigns the MAC address of its own receiving interface to the passing MAC address 125 of the loop detection frame. The frame handling block 202 rewrites the header part so that the source MAC address becomes the MAC address of the own switch and the destination MAC address becomes the MAC address of the switch at the starting point of the loop detection frame (in this case, indicates the L2 switch 101). The value is updated to the loop notification frame (S305). The frame handling block 202 transfers this loop notification frame to the output frame selection block 204, performs transmission processing (S306), and ends.

  The L2 switch 101 receives the loop notification frame 131, and a loop is formed with the L2 switches 104 and 105 sandwiched between the interfaces to which the MAC addresses 102-2 and 102-3 of the L2 switch 102 are assigned. Can be confirmed. The L2 switch 102 tentatively avoids loops (described later), but eventually the operator connects the loop notification frame 131 information returned to the L2 switch 101 so that an optimal network configuration is obtained. It is desirable to control the relationship.

  In the fault handling process shown in FIG. 4, the frame handling block 202 first checks whether or not the loop is detected for the first time in step 322. Specifically, there is a register that holds the history of whether or not a loop has been detected in its own switch. This register is held when a loop is detected, and the history can be cleared periodically or by an instruction such as a command. It is like that. In step 322, the frame handling block 202 checks this register to determine whether it is the first detection. In the first case, the process proceeds to step 323. In step 323, the frame handling block 202 blocks the interface that has received the loop detection frame, and discards all frames that arrive thereafter. This is in order to avoid the loop provisionally because frame transfer in the network cannot be performed correctly in a state where the loop is formed.

  Next, the frame handling block 202 determines whether or not to clear the MAC address learning table (switching table) (S324). If it is set to be cleared, the frame handling block 202 performs processing for notifying the transmission source of the table clearing and clearing request (S325). This is because if the reception interface is cut off in step 323, the network topology is changed, so that there is a case where the frame is not transferred on the route learned so far. The detailed description of the management of the MAC address learning table is omitted in this embodiment. Thereafter, the process proceeds to step 304 in FIG. When it is set in step 235 that the MAC learning table is not required to be cleared, the frame handling block 202 jumps to step 325 and proceeds to step 304 in FIG.

  If it was not the first time in step 322, the received frame may have been forwarded in the same loop as the previously detected loop. Therefore, in order to prevent double detection, the L2 switch 102 determines whether to discard or transfer the received frame (S326). The transfer is selected when there is a possibility that a loop different from the loop detected this time exists. Otherwise, discard is selected. If discard is selected, the L2 switch 102 discards the received frame (S327) and ends. If transfer is selected, the L2 switch 102 sends a loop detection frame to an interface other than the interface that detected the loop (S328), and ends the process.

  If the operator of the L2 switch 102 does not need to collect the occurrence state of the loop, the processing from step 304 to step 306 in FIG. 3 may be skipped and the processing may be terminated.

  In this embodiment, even if the L2 switch 104 and the L2 switch 105 are L2 switches that do not understand the loop detection frame, the broadcast message 112 is not added with the MAC address of the transmission interface of the switches 104 and 105, but the switch 102 The MAC address 102-2 of the transmission interface is finally sent to the switch 102 via the switches 104 and 105. Even in this case, the switch 102 similarly detects the presence of a loop between the interfaces to which 102-2 and 102-3 are assigned as MAC addresses. In this case, the switch 102 receives the loop detection frame including the pass address generated by itself.

  Although the first embodiment has been described using the L2 switch, a router may be used. Switches and routers are path transfer devices. Switches and routers having a loop detection function are also called loop detection devices. Further, the above-described modification can be similarly applied to other embodiments.

  In the loop detection frame, the broadcast address is described in the destination MAC address, but the present invention is not limited to this, and a MAC address that no one knows (not used) may be described.

  A second embodiment of the present invention will be described with reference to FIGS. Here, FIG. 5 is a block diagram illustrating a network configuration. FIG. 6 is a flowchart for explaining processing of the loop detection frame handling block. In the second embodiment, it is assumed that the manager monitors the state of the managed network, such as a manager of SNMP (Simple Network Management Protocol).

  In FIG. 5, the management target network 401 is the same as that in FIG. 1 except that the L2 switch 103 is not provided therein. The error collection node 402 is deployed independently of the management target network 401. The error collection node 402 corresponds to an SNMP manager or the like.

  The L2 switch 101 sends out the loop detection frame in response to an instruction from the error collection node 402. At this time, not the MAC address of the L2 switch 101 but the MAC address of the error collection node 402 is registered and transmitted as the source MAC address 122.

  In FIG. 6, steps 501, 502, 504, 506, 507, and 508 are the same as steps 301, 302, 304, 306, 307, and 308 in FIG. The transmitted loop detection frame is transferred in the same manner as in the first embodiment, and the loop is detected by the L2 switch 102 in step 502 of FIG. In the present embodiment, the error collection node 402 is intended to grasp the loop state, so that it is not necessary to perform the post-detection fault handling process (step 303 in FIG. 3). Also, in step 505, the loop notification frame is returned to the error collection node 402 instead of to the L2 switch 101 that first output the loop notification frame.

  In the second embodiment, an example is given in which the error collection node is located at a location that can be reached from the managed network by layer 2 switching. However, if the error collection node is not reachable, an error may occur by encapsulating a loop notification frame with an IP header. What is necessary is just to reply to a collection node.

  By using this embodiment, a plurality of managed networks can be managed collectively using one error collection node.

  A third embodiment of the present invention will be described with reference to FIGS. Here, FIG. 7 is a block diagram illustrating a network configuration. FIG. 8 is a flowchart for explaining fault handling. In this embodiment, when an interface is shut off at the time of detecting a loop, an interface to be cut off is selected in consideration of the priority of each interface.

  FIG. 7 shows the same network configuration as that of FIG. 1, but defines the physical line speed of each interface. In this embodiment, an interface with a higher line speed has a higher priority, and an interface with a lower line speed has a lower priority. In FIG. 7, three types of line speeds of 1 Gbit / s, 100 Mbit / s, and 10 Mbit / s are defined, and the priority is 1 Gbit / s> 100 Mbit / s> 10 Mbit / s.

  FIG. 8 shows a processing flow of the loop detection frame handling block after loop detection. Loop detection frame transmission / reception, processing up to loop detection and after the fault handling processing (part described with reference to FIG. 3) and when the detected frame is not the first time, it is the same as in the first embodiment, so loop detection is performed for the first time. Only the operation after the operation is described.

  If loop detection is the first time in step 722 of FIG. 8, the priority of the interface that detected the loop is compared in step 731. Here, the priority between the interface to which the MAC address 102-2 registered in the passing MAC address 125 of the loop detection frame is assigned and the interface to which the MAC address 102-3 that has received the loop detection frame is assigned. For comparison, the interface with 102-2 has a line speed of 10 Mbit / s, and the interface with 102-3 has a line speed of 1 Gbit / s, so the interface with 102-3 has higher priority. .

  Therefore, the L2 switch 102 selects the step 733 side for the branch of step 731 and blocks the interface No. 2 having the MAC address 102-2. On the other hand, in FIG. 7, when the line speed between the L2 switches 102 and 104 is also 1 Gbit / s, the branch of step 731 selects the step 732 side, and the MAC address 102-3 which is the interface on the receiving side is selected. Block interface # 3 you have.

  Specifically, as a method of comparing priorities, a register for setting priority is provided for each interface, and it is possible to determine which priority is higher by comparing the values of the registers. In this embodiment, if the priority level 3 is set for the 1 Gbit / s line interface, the priority level 2 is set for the 100 Mbit / s line interface, and the priority level 1 is set for the 10 Mbit / s line interface, the L2 switch 102 uses the interface. As for the priority of No. 3 and No. 2, it can be determined that the interface No. 3 having a large value set in the register has a high priority.

The definition of the interface priority based on the line speed shown in this embodiment is merely an example, and the priority may be determined by other factors.
According to the present embodiment, an interface to be blocked when a loop is detected can be selected according to priority. If the priority is specified by the line speed, not only can the network bandwidth after being blocked be used more efficiently, but also the high-priority interface that should not be blocked by the method of defining the priority is protected. It is possible to avoid loops in the form.

It is a block diagram explaining the structure of a layer 2 (L2) network. It is a block diagram explaining the internal structure of the switch provided with the loop detection function. It is a processing flowchart explaining operation | movement of a loop detection frame handling block. It is a flowchart explaining fault handling. It is a block diagram explaining the structure of a network. It is a flowchart explaining the process of a loop detection frame handling block. It is a block diagram explaining the structure of a network. It is a flowchart explaining fault handling.

Explanation of symbols

  101 ... L2 switch, 102 ... L2 switch, 103 ... L2 switch, 104 ... L2 switch, 105 ... L2 switch, 111 ... Loop detection frame, 112 ... Loop detection frame, 113 ... Loop detection frame, 114 ... Loop detection frame, 115 ... Loop detection frame, 131 ... Loop notification frame, 121 ... Destination address, 122 ... Transmission source address, 123 ... Type value, 124 ... Pointer field, 125 ... Passing MAC address, 401 ... Managed network, 402 ... Error collection node.

Claims (4)

A loop detection method in a packet transfer device, comprising:
Receiving a first loop detection frame including one or more passage addresses indicating an interface address of a packet transfer apparatus that has passed through the first interface;
Check before Symbol communication over the address containing the address of its own interface,
When there is not any address of the interface of the own prior Symbol communication over address,
Create a second loop detection frame by adding the address of a different second interface and the first interface before Symbol passing over address, transmitted from different ones of the second interface and the first interface ,
When the address of the own interface is included in the passing address, the occurrence of a loop is detected,
Adding the address of the first interface that has received the first loop detection frame to the passing address included in the first loop detection frame to notify another packet transfer apparatus of the occurrence of the loop A loop detection method comprising: creating a loop notification frame and transmitting the frame to another packet transfer apparatus . The loop detection method according to claim 1,
When it contains the address of the interface of the own prior Symbol communication over address, loop detection method characterized by blocking the first interface. The loop detection method according to claim 1 or 2,
A broadcast detection method or an unlearned MAC address is used as a destination MAC address included in the first loop detection frame and the second loop detection frame . A loop detection apparatus including a first interface and a second interface,
Receiving a first loop detection frame including one or more passage addresses indicating an interface address of a packet transfer apparatus that has passed through the first interface;
Check if the transit address includes the address of its own interface,
When the passing address does not include the address of the own interface,
Adding the address of the second interface to the passing address to create a second loop detection frame, and transmitting from the second interface;
When the address of the own interface is included in the passing address, the occurrence of a loop is detected,
Adding the address of the first interface that has received the first loop detection frame to the passing address included in the first loop detection frame to notify another loop detection device of the occurrence of the loop A loop detection device that creates a loop notification frame and transmits it to the other loop detection device.

Images