JP2018007094A - Physical asymmetric routing prevention mechanism in redundant configuration of relay device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method of preventing the occurrence of asymmetric routing in redundant relay devices including interfaces with different routing priority, and a redundant relay device using the method.SOLUTION: A first relay device operating as a main relay device and a second relay device operating as a backup relay device sets an inter-relay device route between them, and the second relay device sets a LAN side port of itself to a frame transmission/reception suppression state, and cancels the frame transmission/reception suppression state when failure is detected at the first relay device itself or a LAN side of the first relay device.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a technique for preventing asymmetric routing when a network configuration is made redundant.

  When normal routing is used in a network redundancy configuration, asymmetric routing in which the path of transmission traffic and reception traffic are different may occur. The occurrence of asymmetric routing is known to cause an increase in network load and a decrease in network performance, and various methods for solving such asymmetric routing problems have been proposed.

  For example, in Patent Document 1, when asymmetric routing via a server proxy in which transmission traffic and reception traffic are different is detected, aging of the MAC table and the ARP table is performed by switching the master right between redundant relay apparatuses. A control method for preventing unicast flooding due to a time difference is disclosed. Patent Document 2 discloses a method of exchanging connection information with neighboring proxies and redirecting traffic so that communication between the client and the server passes through the same proxy.

JP 2014-183549 A Special table 2008-536369

  However, any of the background arts described above is a response process after the occurrence of asymmetric routing is detected, and is not a technique for preventing the occurrence of asymmetric routing itself.

  Further, in any background art, a multi-stage configuration such as a proxy and an L3 switch is assumed, and the reliability of a routing protocol (hereinafter referred to as a routing priority) in a relay device such as a proxy and an L3 switch is considered. It has not been. The routing priority is also called AD (Administrative Distance) or preference value (Preference value). Usually, the directly connected route has the highest priority, and is based on static route, OSFP (Open Shortest Path First), etc. It is defined that the routing priority decreases in the order of dynamic routes.

  The method of determining the output interface for each of the directly connected route, the static route, and the dynamic route is described in RFC (Request for Comments) 2328 (page 167; “16.1.1 The next hop calculation”). As a specific example, the default AD value in a router manufactured by Cisco Systems, which is the de facto standard, is set such that direct connection is 0, static is 1, dynamic (OSPF) is 110, and Juniper Networks The default preference values for routers made by the manufacturer are set such that direct connection is 0, static is 20, and dynamic (OSPF) is 60. In any case, the smaller the value, the higher the routing priority.

  In a relay device that does not consider such a difference in routing priority, there is a case where asymmetric routing cannot be prevented as described below. Hereinafter, an example of a redundant configuration in which asymmetric routing occurs will be briefly described with reference to FIG.

  As illustrated in FIG. 1, the redundant routers RT1 and RT2 are customer edge routers that connect a LAN (Local Area Network) segment A to a main line and a backup line on the WAN (Wide Area Network) side, respectively. (CE router). The respective ports P11 and P21 of the routers RT1 and RT2 are directly connected to the hub HUB on the LAN side, and the respective ports P13 and P23 are connected to each other through a static or dynamic route (crossover route). Furthermore, it is assumed that the ports P12 and P22 are connected to the WAN side as dynamic routes. When a failure has not occurred, the host computer in the LAN segment A can send and receive IP packets to and from the host computer at the opposite base B through the main line using the main router RT1 as a default gateway. Note that a network device that operates dynamic routing is installed at the opposite base B, and the routing information is directly exchanged or indirectly shared with the routers RT1 and RT2 on the LAN segment A side. To do.

  In the above network configuration, it is assumed that a failure has occurred in the main line connecting the LAN segment A and the opposite base B while both the routers RT1 and RT2 are operating normally. In this case, the router RT1 transfers the packet from the LAN segment A to the router RT2 through the route according to the dynamic routing protocol such as OSPF, and the router RT2 also transfers the packet to the opposite base B through the backup line by the dynamic routing ( Backup path (bound) R10a).

  However, when the router RT2 receives an IP packet from the opposite site B through the backup line, the router RT2 performs routing in preference to the directly connected LAN side route having a higher routing priority than the static route or the dynamic route. That is, even if the router RT2 is designed so as to pass through a static route or a dynamic route, the router RT2 ignores this and transfers the received packet directly to the LAN segment A through the LAN side route (backup route (return) R20a ). In this way, an asymmetric routing is generated in which the route R10a goes through the crossover route and the return route R20a does not go through the crossover route but through the directly connected LAN side route.

  As described above, when the directly connected route having the highest routing priority is included in the redundant router, it is not possible to prevent the occurrence of asymmetric routing by simply redirecting traffic.

  Next, a case where master right switching control is applied to the redundant configuration of the routers RT1 and RT2 described above will be described. That is, redundancy of the default gateway is performed between the routers RT1 and RT2 by VRRP (Virtual Router Redundancy Protocol). Here, it is assumed that the port P11 of the router RT1 has the master right. In this case, the router RT1 having the master right monitors one counter base B, and when a failure is detected, the master right is failed over from the router RT1 to the RT2. When the switching control of the master right adopted between such single bases is applied to a network configuration having a plurality of opposing bases B and C, asymmetric routing may occur as described below.

  As shown in FIG. 2, it is assumed that segment A can communicate independently with a plurality of opposing bases B and C through a redundant configuration of routers RT1 and RT2. If both routers RT1 and RT2 are operating normally and a failure occurs on the main line connecting LAN segment A and opposite base B, the master right is switched from router RT1 to RT2, and router RT2 is the default gateway. Become. Since the router RT2 has become the default gateway, the router RT2 transfers the packet from the segment A to the router RT1 through the cross route according to the dynamic routing between the segment A and the opposite base C, and the router RT1 dynamically transfers the packet. It is transferred to the opposite base C through the main line by routing (backup route (bound) R10a).

  However, when the router RT1 receives an IP packet from the opposite base C through the main line, the router RT1 receives the received packet in order to give priority to the directly connected LAN side route having a higher routing priority than the static route. Transfer directly to LAN segment A (backup path (return) R20a). That is, the outgoing route R10a goes through the crossover route, and the return route R20a goes through the directly connected LAN side route instead of the crossover route, and the master right is switched due to the relationship with the opposite base B. As a result, asymmetric routing occurs in the IP packet transmitted / received to / from the opposite base C that is not related to the opposite base B.

  As described above, when the directly connected route having the highest routing priority is included in the redundant router, the switching of the master right and the traffic redirection cannot completely prevent the occurrence of asymmetric routing.

  Accordingly, an object of the present invention is to provide a method for preventing the occurrence of asymmetric routing in a redundant configuration of a relay device in which interfaces having different routing priorities coexist, and a redundant configuration and a relay device using the method.

The asymmetric routing prevention method according to the present invention prevents asymmetric routing from occurring in a redundant configuration of a plurality of relay devices that are directly connected to a local area network (LAN) and provided between the LAN and a wide area network (WAN). A first relay device operating as a main relay device and a second relay device operating as a backup relay device set a route between them, and the second relay device is connected to the LAN side of its own station The port is set to a frame transmission / reception suppression state, and the frame transmission / reception suppression state is canceled when a failure is detected on the first relay device itself or on the LAN side of the first relay device.
A redundant configuration according to the present invention is a redundant configuration of a plurality of relay devices that are directly connected to a local area network (LAN) and provided between the LAN and a wide area network (WAN), and are used as a main relay device. A first relay device that operates, and a second relay device that operates as a backup relay device and is connected to the first relay device via a crossover route, wherein the second relay device is the first relay device itself or The monitoring means for monitoring the occurrence of a failure on the LAN side of the first relay device and the LAN side port of the second relay device are set in a frame transmission / reception inhibited state, and the occurrence of the failure is detected in the frame transmission / reception inhibition state. And a control means that is released when the
A relay device according to the present invention is a relay device that is directly connected to a local area network (LAN) and is provided between the LAN and a wide area network (WAN). Redundant configuration control means that operates as a main relay device or backup relay device in the configuration, and when the relay device operates as a backup relay device, the adjacent relay device itself operating as the main relay device or the LAN of the adjacent relay device When the relay device operates as a backup relay device, the monitoring unit for monitoring the occurrence of a failure on the side sets the LAN side port of the relay device to the frame transmission / reception inhibited state, and sets the frame transmission / reception inhibited state to the occurrence of the failure. Control means for releasing when detected.

  As described above, according to the present invention, it is possible to prevent the occurrence of asymmetric routing in a redundant configuration of a relay apparatus in which interfaces having different routing priorities are mixed.

FIG. 1 is a network configuration diagram schematically showing an example of a redundant configuration for explaining a problem to be solved by the present invention. FIG. 2 is a network configuration diagram schematically showing another example of a redundant configuration for explaining the problem to be solved by the present invention. FIG. 3 is a network configuration diagram schematically showing a redundant configuration of a relay device according to an embodiment of the present invention. FIG. 4 is a schematic network configuration diagram for explaining cost adjustment. FIG. 5 is a schematic network configuration diagram for explaining cost adjustment in the present embodiment. FIG. 6 is a network configuration diagram schematically showing a redundant router configuration according to the first embodiment of the present invention. FIG. 7 is a sequence diagram showing an example of a routing control operation when a failure occurs on the main line in the redundant configuration according to the first embodiment. FIG. 8 is a network configuration diagram schematically showing the routing control operation shown in FIG. FIG. 9 is a sequence diagram showing an example of a routing control operation when a failover occurs due to a failure in the redundant configuration according to the first embodiment. FIG. 10 is a network configuration diagram schematically showing the routing control operation shown in FIG. FIG. 11 is a sequence diagram showing an example of a routing control operation when a failure occurs in the main router LAN in the redundant configuration according to the first embodiment. FIG. 12 is a network configuration diagram schematically showing the routing control operation shown in FIG. FIG. 13 is a block diagram showing a schematic function of the router in the first example of the redundant configuration according to the second embodiment of the present invention. FIG. 14 is a network configuration diagram schematically showing a first example of the routing operation in the redundant configuration shown in FIG. FIG. 15 is a network configuration diagram schematically showing a second example of the routing operation in the redundant configuration shown in FIG. FIG. 16 is a network configuration diagram schematically showing a third example of the routing operation in the redundant configuration shown in FIG. FIG. 17 is a network configuration diagram schematically showing a fourth example of the routing operation in the redundant configuration shown in FIG. FIG. 18 is a block diagram showing the configuration of a redundant router according to the third embodiment of the present invention. FIG. 19 is a network configuration diagram schematically showing a first example of a network configuration to which the redundant router according to the third embodiment is applied. FIG. 20 is a network configuration diagram schematically showing a first example of the routing operation in the network configuration shown in FIG. FIG. 21 is a network configuration diagram schematically showing a second example of the routing operation in the network configuration shown in FIG. FIG. 22 is a network configuration diagram schematically showing a third example of the routing operation in the network configuration shown in FIG. FIG. 23 is a network configuration diagram schematically showing a fourth example of the routing operation in the network configuration shown in FIG. FIG. 24 is a network configuration diagram schematically showing a second example of a network configuration to which the redundant router according to the third embodiment is applied. FIG. 25 is a network configuration diagram schematically showing a first example of the routing operation in the network configuration shown in FIG. FIG. 26 is a network configuration diagram schematically showing a second example of the routing operation in the network configuration shown in FIG. FIG. 27 is a network configuration diagram schematically showing a third example of the routing operation in the network configuration shown in FIG. FIG. 28 is a network configuration diagram schematically showing a fourth example of the routing operation in the network configuration shown in FIG.

<Outline of Embodiment>
According to the embodiment of the present invention, a relay device directly connected to a user-side or customer-side LAN is made redundant, and between the redundant main-side first relay device and backup-side second relay device. The route is set. When the directly connected LAN side port has the highest routing priority in the first relay device and the second relay device, frame transmission / reception between the second relay device and the LAN through the LAN side port is suppressed. . The frame transmission / reception suppression state of the LAN side port is canceled when a failure is detected on the first relay device itself or on the LAN side of the first relay device. By such frame transmission / reception suppression / cancellation control through the LAN side port, it is possible to prevent the occurrence of physical asymmetric routing in a redundant configuration of a relay apparatus in which interfaces having different routing priorities coexist.

  Various methods can be used to suppress frame transmission / reception of the LAN side port. As a first example, the LAN side port of the second relay device can be set to a MAC address learning suppression state, and packet transmission / reception between the second relay device and the LAN through the LAN side port can be suppressed. This MAC address learning suppression state is canceled when a failure is detected on the first relay device itself or on the LAN side of the first relay device. As a second example, by using STP (Spanning Tree Protocol) blocking and performing blocking setting / release control of the LAN side port of the second relay device, the second relay device passing through the LAN side port and the LAN It is also possible to control packet transmission / reception in between. It is also possible to set and cancel frame transmission / reception suppression using Syslog or the like as a trigger. By using such a configuration, it is possible to prevent the occurrence of physical asymmetric routing in a redundant configuration of a relay apparatus in which interfaces having different routing priorities are mixed.

  As will be described in detail later, in a redundant configuration in which the first relay device and the second relay device are connected by a route, a difference occurs in addition of cost values performed in dynamic routing, and asymmetric routing occurs on the WAN side. The possibility of making it arise. The difference in the cost value in the dynamic routing can be uniformly adjusted by a transfer segment separately created for the transfer route.

  The term “relay device” represents a network node having a routing function and a plurality of interfaces having different routing priorities, and includes a router, a layer 3 switch, and the like. Hereinafter, embodiments and examples of the present invention will be described in detail with reference to the drawings.

1. In one embodiment of the present invention described below, an edge relay device on the LAN side is redundant in a virtual dedicated network that connects geographically separated LANs using an IP-based closed network provided by the WAN side. An example will be described. By providing redundancy, if the main relay device fails, for example, the role of the default gateway can be automatically transferred from the main side to the backup relay device, and communication between the base LANs can be maintained. . As a redundancy protocol used in the present embodiment, VRRP (Virtual Router Redundancy Protocol), HSRP (Hot Standby Router Protocol), and the like are known, but are not limited thereto.

1.1) Configuration As illustrated in FIG. 3, it is assumed that the relay devices 10 and 20 as edge relay devices on the LAN side are made redundant. Here, the relay device 10 is the main and the relay device 20 is the backup. Here, as an example, the relay device 10 has three physical ports P11, P12, and P13, the relay device 20 has three physical ports P21, P22, and P23, and the ports P12 and P22 can be used as routed ports. Yes, ports P11, P13, P21 and P23 can be used as switching ports.

  The port P11 of the relay device 10 is directly connected to the hub 11 on the LAN side, the port P12 is connected to the main line 12m provided by the WAN side, and the port P13 is connected to the port P23 of the backup relay device 20. These ports P11, P12, and P13 are connected to a LAN side IF (interface), a WAN side IF, and a transition IF in the relay apparatus 10, respectively. Similarly, the port P21 of the relay device 20 is directly connected to the LAN-side hub 11, the port P22 is connected to the backup line 12b provided by the WAN side, and the port P23 is connected to the port P13 of the main relay device 10. ing. These ports P21, P22, and P23 are connected to the LAN side IF, the WAN side IF, and the transition IF in the relay apparatus 20, respectively.

  The main relay device 10 and the backup relay device 20 have a routing function and a network monitoring function based on normal routing priority. In normal routing priority, a directly-connected route has the highest priority (minimum AD value or preference value), followed by static route (walking route S) and dynamic route in this order. The routing priority has dropped.

  Further, the backup relay device 20 has a control function for setting or canceling the direct connection port P21 to the frame transmission / reception inhibited state. The frame transmission / reception suppression control function normally sets the port P21 to a data transmission / reception suppression state, and automatically releases the frame transmission / reception suppression state when a failure occurrence is detected on the main relay device 10 or its LAN side. When recovering from the failure, the port P21 automatically returns to the frame transmission / reception inhibited state. The purpose of this frame transmission / reception inhibition is to inhibit frame 2 transmission / reception.

  As described above, the LAN side ports P11 and P21 of the relay apparatuses 10 and 20 are connected to the hub 11, and the ports 13 and P23 are connected by the crossing route S, so that the relay apparatus 10, the relay apparatus 20, and the hub 11 are connected. Can form a loop in a layer 2 network. The fact that the port P21 of the backup relay apparatus 20 is in the normal frame reception inhibited state prevents a traffic loop in the layer 2 network.

1.2) Cost adjustment However, when the above-described configuration is adopted, the LAN side segment is set to cross with the LAN side of the relay device. Therefore, when the cost value of dynamic routing is added, the cost value between the LAN side and crossing is set. Are the same. As a result, as described below, when a failure occurs in the main line, there is a possibility that asymmetric routing on a site basis occurs on the WAN side.

  As illustrated in FIG. 4, the LAN side segment via the main relay device 10 is designated as cost C10, the LAN side segment via the backup relay device 20 is designated as cost C100, and the LAN side segment via the main line and the main relay device 10 is normally used. It is assumed that the network device sharing the dynamic routing protocol of the other base is notified to access the network.

  In this state, when a failure occurs in the main line 12m and the packet is transferred through the route of the main relay device 10, the transfer route S, and the backup relay device 20, the main relay device 10 passes through the transfer route S from its own station. When the cost value of “10” is added to the packet at this time, it is calculated. Conversely, when the network device at the communication target base transmits a packet from the backup line 12b to the backup relay device 20, it calculates that a cost value of “100” is added when the packet passes through the crossing route S. Therefore, a difference occurs between the cost value from the local site to the other site and the cost value from the other site to the local site when a failure occurs in the main line 12m. If the calculated cost values are different between the bases, there is a possibility that it is determined that the cost is lower through other bases depending on the cost design of each base. For example, as shown in FIG. 4, asymmetric routing occurs in units of bases where the route R10a going to the WAN and the return route R20a are different from each other on the WAN side, which may cause a routing loop. In the case of communication between single sites, it may be possible to cope with construction means such as defining only with static routes, but when there are multiple sites and dynamic communication control by dynamic routing is performed, the frame transmission / reception suppression port It is difficult to prevent asymmetric routing by itself.

  Therefore, as illustrated in FIG. 5, a transition segment is separately created, and cost adjustment is performed to set a cost value (high priority) lower than the set cost value of the main relay apparatus 10 for the transition segment.

  In FIG. 5, the LAN side segment via the main relay device 10 has a cost value “11” (C11), the LAN side segment via the backup relay device 20 has a cost value “100” (C100), and the transition segment is a main relay. Both the device 10 and the backup relay device 20 are set to the cost value “10” (C10). Thereby, the cost value of the transition segment is set lower than the cost value of the LAN side segment via each relay device. It is assumed that the network device sharing the dynamic routing protocol of another base is notified so that the LAN side segment is accessed via the main line and the main relay device 10 in normal times.

  Consider a case where a failure occurs in the main line in this state, and a packet is transferred through the route of the main relay device 10, the transition route S, and the backup relay device 20. In the main relay device 10, when a packet entering from the LAN side port P11 tries to pass through the crossing route S, the route having the lowest cost value (high priority) as the route for the other base is the crossing of the crossing route S. Segment, and its cost value is “10”. Accordingly, the main relay apparatus 10 calculates that a cost value of “10” is added to the packet transferred through the transition segment. Conversely, when the network device at the communication target base transmits a packet to the backup relay device 20 through the backup line 12b, the route with the lowest cost value (high priority) as the LAN-side segment route is also the same as the transit route S. Since it is a transition segment, it is calculated that a cost value of “10” is also added to this packet.

  However, in actuality, when communication for the LAN side segment arrives at the backup relay device 20, the direct connection route with the highest priority is selected, so the transition route S is not the transition segment of the transition route S. Communication is performed via the LAN side segment. However, since the network device at the other site calculates the route from the other site to the own site through the backup line 12b, the route to the backup relay device 20 is calculated based on the cost value of dynamic routing. Has no effect.

  In this way, packets transmitted from the local site to the other site via the backup line 12b pass through the transfer segment of the transfer route S, and packets received from the other site pass through the LAN side segment of the transfer route S. In other words, some asymmetry is included on the logical path, but asymmetry on the physical path is eliminated.

  The failure range of the main line 12m includes the WAN IF of the main relay device 10, but by monitoring the WAN IF of the main relay device 10 and failing over the default gateway to the backup relay device 20 when a failure is detected. The width of occurrence of logical asymmetric routing can be reduced.

  Specifically, a case where a failure is detected in the WAN side IF of the main relay apparatus 10 and the default gateway fails over to the backup relay apparatus 20 will be considered. Since the LAN side IF of the backup relay device 20 is maintained in a frame transmission / reception suppression state, when a packet addressed to another site arrives from the LAN side segment to the main relay device 10, the main relay device 10 crosses the arrival packet and passes through the route S. The data is transferred to the backup relay device 20 via the LAN side segment. This is because the LAN side segment of the transfer route S is a direct connection route having a higher routing priority and the MAC address of the default gateway is registered. The packet addressed to the other site that has reached the backup relay device 20 is transferred to the backup line 12b according to the dynamic routing by referring to the presence or absence of the destination address in the routing table. Packets from other bases are transferred by dynamic routing until they arrive at the backup relay device 20 as described above, and after arrival, they pass through the LAN side segment of the transit route S, which is a direct connection route with higher priority. To the main relay device 10. In this way, the occurrence of asymmetric routing can be avoided both physically and logically.

2. First Embodiment Hereinafter, a mechanism for preventing asymmetric routing in a redundant configuration using routers will be described as a first embodiment of the present invention. As described above, a network node such as a layer 3 switch having the same function is described. It is also applicable when using. According to the present embodiment, by setting the LAN side port of the redundant router to the MAC address learning inhibition state, the LAN side port can be set to the frame transmission / reception inhibition state.

2.1) Configuration As illustrated in FIG. 6, the backup router 20 in the redundant configuration suppresses the MAC address learning function of the port P21, so that the layer 2 or higher of the port P21 according to the ARP table and the MAC address table is used. Communication is not possible. For this reason, the backup router 20 recognizes the MAC address to be communicated through the route S. Hereinafter, the routing operation of the redundant configuration illustrated in FIG. 6 will be described.

2.2) Operation <Normal operation>
As shown in FIG. 7, first, assume that the router 10 and the router 20 are made redundant by the redundancy protocol, and the router 10 is determined as the main and the router 20 is determined as the backup (operation S101). The backup router 20 sets the port P21 directly connected to the LAN side to the MAC address learning inhibited state (Operation S102). Subsequently, the routers 10 and 20 monitor the network (operations S103 and S104). At that time, the backup router 20 transmits a packet for monitoring the LAN side IF of the main router 10 to the main router 10 at a predetermined cycle, and a failure of the LAN side IF of the main router 10 occurs based on the response packet or the presence / absence of a response. May be detected. Alternatively, as described later, the tree state can be monitored using a BPDU (Bridge Protocol Data Unit) in STP (Spanning Tree Protocol).

  If no failure is detected in the network, the LAN segment can perform normal communication with the opposite base through the main router 10 and the main line 12m (operation S105). That is, the main router 10 transfers the IP packet entered from the LAN segment side port P11 to the WAN side port P12 by dynamic routing, and transmits it to the opposite site through the main line 12m. Conversely, the main router 10 transfers the IP packet that has entered from the WAN side port P12 to the LAN side port P11 by directly-connected, and outputs it to the LAN segment. Therefore, in a state where no failure has occurred, the LAN segment can transmit and receive IP packets with the opposite base through the main router 10 and the main line 12m, and asymmetric routing occurs between the transmission traffic and the reception traffic. Absent.

<Operation when main line fails>
As shown in FIG. 7, when the main router 10 detects a failure on the main line 12m on the WAN side (Operation S106), the LAN segment passes through the main router 10, the transfer route S, the backup router 20, and the backup line 12b. Backup communication can be performed with the base (operation S107). This backup communication will be described with reference to FIG.

  In FIG. 8, when the main router 10 detects that a failure has occurred in the main line 12m, the main router 10 transfers the IP packet received from the port P11 on the LAN segment side to the port P13 by dynamic routing, and passes through the transfer route S to the backup router. 20 output. The backup router 20 transfers the IP packet entered from the port P23 of the crossover route S to the WAN side port P22 by dynamic routing, and transmits it to the opposite site through the backup line 12b.

  In contrast, when an IP packet enters from the WAN side port P22, the backup router 20 tries to transfer the packet to the port P21 of the direct connection route having the highest routing priority. The MAC address learning is inhibited and the MAC address to be communicated is recognized via the crossover route S. Therefore, the IP packet that has entered from the port P22 is transferred to the port P23 on the transfer route S side, and is output to the main router 10 through the transfer route S. The main router 10 transfers the IP packet that has entered from the port P13 of the transit route S to the LAN side port P11 that has the highest routing priority and is linked to the MAC address to be communicated, and outputs it to the LAN segment through the hub 11. To do. Thus, when a failure occurs in the main line 12m, the LAN segment can transmit and receive IP packets to and from the opposite base through the main router 10, the transfer route S, the backup router 20, and the backup line 12b. There is no physical asymmetric routing.

<Operation when failover occurs due to failure>
As shown in FIG. 9, when performing normal operation through the main router 10, the main router 10 has a fault in its own station (a casing fault of the main router 10, or a duplicate fault of the local LAN side IF and the crossover IF). Is detected (operation S201), and the default gateway fails over to the router 20 (operation S202). The backup router 20 cancels the inhibition of MAC address learning of its own LAN side port P21 triggered by a LAN failure separately detected in parallel with the operation of taking over the role of the default gateway (operation S203). As described above, the backup router 20 may detect the presence or absence of the response of the monitoring packet as to how the LAN router IF failure occurs in the main router 10, but the output is performed according to the failure of the main router 10. The backup router 20 may detect it by using a syslog message or the like as a trigger. Thus, the LAN segment can communicate with the opposite base through the backup route through the backup router 20 and the backup line 12b (operation S204). This backup communication will be described with reference to FIG.

  As shown in FIG. 10, when a failure occurs in the main router 10, the backup router 20 takes over the role of the default gateway and cancels the MAC address learning suppression of the LAN side port P21. As a result, the backup router 20 transfers the IP packet received from the port P21 on the LAN segment side to the port P22 by dynamic routing, and transmits it to the opposite site through the backup line 12b.

  Conversely, when an IP packet enters from the WAN side port P22, the backup router 20 cancels the MAC address learning suppression of the LAN side port P21 having the highest routing priority. It recognizes via the side port P21, transfers the incoming IP packet to the port P21, and outputs it to the LAN segment through the hub 11. Thus, when a failover occurs due to a failure in the main router 10, the MAC address learning suppression port P21 of the backup router 20 is released, whereby the LAN segment transmits an IP packet between the opposite base through the backup router 20 and the backup line 12b. Thus, the physical and logical asymmetric routing as described above does not occur.

<Operation when a failure occurs on the main router LAN>
As shown in FIG. 11, when a normal operation is performed through the main router 10, a failure is detected on the LAN side of the main router 10 (operation S301), and no failover to the backup side of the default gateway is caused thereby. Shall. Therefore, the main router 10 is still operating as a default gateway. In this case, when the occurrence of a failure on the LAN side of the main router 10 is detected, the backup router 20 cancels the MAC address learning inhibition of the LAN side port P21 of the local station (operation S302). In this case, since the main router 10 is the default gateway, the LAN segment can communicate with the opposite base through the backup router 20, the transfer route S, the main router 10, and the main line 12m (operation S303). The main communication via this transfer route will be described with reference to FIG.

  As shown in FIG. 12, when a failure occurs on the LAN side of the main router 10 and the default gateway is maintained on the main router 10, the backup router 20 cancels the MAC address learning suppression of the LAN side port P21, and the LAN The IP packet (frame) entered from the segment side port P21 is transferred to the main router 10 operating as a default gateway by L2 switching. The main router 10 transfers the IP packet received from the port P13 to the port P12 by dynamic routing, and transmits it to the opposite site through the main line 12m.

  On the other hand, when an IP packet enters from the WAN side port P12, the main router 10 transfers the IP packet from the port 13 of the route S to the backup router 20 because a failure has occurred on the LAN side. . The backup router 20 transfers the IP packet that has entered from the port P23 to the port P21 for which the MAC address learning suppression is canceled, and outputs it to the LAN segment through the hub 11. Thus, when a failure occurs on the LAN side of the main router 10 and the default gateway is maintained by the main router 10, the MAC address learning suppression port P21 of the backup router 20 is released. As a result, the LAN segment can send and receive IP packets to and from the opposite base through the backup router 20, the transfer route S, the main router 10 and the main line 12m, and the physical and logical asymmetry as described above. Routing does not occur.

2.3) Effect As described above, according to the first embodiment of the present invention, a transit route is set between the main router and the backup router, and the LAN side port of the backup router is normally in a MAC address learning inhibited state. Is set to When a failure is detected on the main router itself or on the LAN side of the main router, the MAC address learning suppression of the LAN side port of the backup router is canceled, and packets are transmitted and received between the backup router and the LAN through the LAN side port. It becomes possible. By such a MAC address learning suppression control of the LAN side port in the backup router, it is possible to prevent physical asymmetric routing of transmission and reception traffic in a configuration in which the customer side edge router is made redundant.

3. Second Embodiment In the redundant configuration according to the second embodiment of the present invention, the setting / release control of the blocking port on the backup router side is performed using STP blocking. Hereinafter, a router according to the present embodiment and its control operation will be described in detail with reference to the drawings. However, the same components as those in FIG.

3.1) Redundant Configuration As illustrated in FIG. 13, each of the routers 10 and 20 has a functional configuration including a routing unit RT and a virtual interface vIF corresponding to each port. The virtual interface vIF is, for example, SVI (Switched Virtual Interface). In the router 10, the virtual interfaces vIF11, vIF12, and vIF13 correspond to the LAN side port P11, the WAN side port P12, and the transition port P13, respectively. In the router 20, virtual interfaces vIF21, vIF22, and vIF23 correspond to the LAN side port P21, the WAN side port P22, and the transition port P23, respectively.

  As already described, since the router 10 and the router 20 connected by the crossover router S are connected to the hub 11 by the respective LAN side ports P11 and P21, they can form a loop in the layer 2. Therefore, by applying the STP protocol to this loop formation, it is possible to realize the setting / release control of the blocking port. More specifically, according to the STP protocol, the port P21 of the router 20 is automatically set to the blocking state at the layer 2, and when the loop is not formed due to the occurrence of the failure, the blocking is automatically released. Thus, by using STP, blocking control can be easily performed, and as a result, occurrence of physical asymmetric routing can be prevented.

  As described above, the difference in the cost value added to the packet becomes a problem by adopting the loop configuration, but the virtual interface vIF13 and vIF23 are set to a lower cost value (higher priority) than vIF11 and vIF21. By doing so, the difference can be adjusted.

3.2) Routing operation <When WAN side port is not down due to main line failure>
As shown in FIG. 14, it is assumed that the main router 10 detects a failure in the WAN main line 12m in a state where the LAN side port P21 of the backup router 20 is blocked by the STP protocol described above. In this case, in the main router 10 operating as the default gateway, the IP packet that has entered from the port P11 on the LAN segment side is transferred to the port P13 by dynamic routing, and is output to the backup router 20 through the transfer route S. At this time, there are two paths, vIF11 and vIF13, from the communication to the destination facing base. Since the cost adjustment of the vIF 13 is lower (higher priority) than the vIF 11 by performing the cost adjustment described above, the IP packet is transferred to the port P13 via the vIF 13 and output to the backup router 20. . The backup router 20 transfers the IP packet entered from the port P23 of the crossover route S to the WAN side port P22 by dynamic routing, and transmits it to the opposite site through the backup line 12b.

  On the contrary, when an IP packet enters from the WAN side port P22, the backup router 20 transfers the packet to the directly connected virtual interface vIF 21 having the highest routing priority. However, since the port P21 is in the blocking state, the MAC address of the destination IP is registered via the transfer route S of the vIF 21. Therefore, the backup router 20 transfers the IP packet that has entered from the WAN side port P22 to the port P23 on the crossover route S side, and outputs it to the main router 10 through the crossover route S. The main router 10 transfers the IP packet that has entered from the port P13 of the crossover route S to the LAN side port P11 that has the highest routing priority and is linked to the MAC address to be communicated, and passes through the hub 11 to the LAN segment. Output. Thus, when a failure occurs in the main line 12m, the LAN segment can transmit and receive IP packets to and from the opposite base through the main router 10, the transfer route S, the backup router 20, and the backup line 12b. There is no physical asymmetric routing.

<When WAN side port goes down due to main line failure>
As shown in FIG. 15, it is assumed that the WAN side port P12 of the main router 10 is down while the LAN side port P21 of the backup router 20 is blocked by the STP protocol described above. In this case, the default gateway fails over to the backup router from the main router that monitors the local station WAN IF. In the main router 10, the IP packet entered from the port P11 on the LAN segment side is transferred to the port P13 via the vIF 11 by direct connection. Since dynamic routing is not used and passes through L2, as described in FIG. 14, even if the cost value of vIF 13 is lower (priority is higher) than vIF 11, it does not go through vIF 13 and passes through the transit route S. Output to the vIF 21 of the backup router 20. The backup router 20 transfers the IP packet that has entered the vIF 21 from the port P23 of the transition route S to the WAN side port P22 by dynamic routing, and transmits it to the opposite site through the backup line 12b.

  On the contrary, when an IP packet enters from the WAN side port P22, the backup router 20 transfers the packet to the directly connected virtual interface vIF 21 having the highest routing priority. Since the port P21 is in the blocking state, the MAC address of the destination IP is registered via the transfer route S of the vIF 21. Thereby, the backup router 20 transfers the incoming IP packet to the port P23 on the crossing route S side, and outputs it to the main router 10 through the crossing route S. The main router 10 transfers the IP packet that has entered from the port P13 of the crossover route S to the LAN side port P11 that has the highest routing priority and is linked to the MAC address to be communicated, and passes through the hub 11 to the LAN segment. Output. Thus, when a failure occurs in the WAN side port of the main router 10 connected to the main line 12m, the LAN segment is connected to the opposite base through the main router 10, the transfer route S, the backup router 20, and the backup line 12b. IP packets can be transmitted and received, and physical and logical asymmetric routing as described above does not occur.

<When a failover occurs due to a failure of the main router>
As shown in FIG. 16, when the main router 10 detects a failure in its own station, the default gateway fails over to the router 20. Thereby, the backup router 20 takes over the role of the default gateway, and the STP blocking of the LAN side port P21 of its own station is automatically released. As a result, the backup router 20 transfers the IP packet received from the port P21 on the LAN segment side to the port P22 by dynamic routing, and transmits it to the opposite site through the backup line 12b.

  Conversely, when an IP packet enters from the WAN side port P22, the backup router 20 transfers the IP packet to the LAN side port P21 from which blocking has been released, and outputs it to the LAN segment through the hub 11. Thus, failover occurs due to a failure in the main router 10, and the blocking port P21 of the backup router 20 is released, so that the LAN segment transmits and receives IP packets to and from the opposite base through the backup router 20 and the backup line 12b. And the physical and logical asymmetric routing as already mentioned does not occur.

<When a failure occurs on the main router LAN>
As shown in FIG. 17, when a failure occurrence is detected on the LAN side of the main router 10 and the main router 10 is operating as a default gateway, blocking of the LAN side port P21 of the backup router 20 is released. As a result, the IP packet that has entered from the port P21 on the LAN segment side is transferred to the main router 10 that operates as a default gateway, is transferred to the port P12 by dynamic routing by the main router 10, and is transmitted to the opposite site through the main line 12m. The The process up to the main router 10, which is the default gateway, is the same as that in the case where the WAN side port goes down due to the main line failure already described (FIG. 15), and thus description thereof is omitted.

  Conversely, when an IP packet enters from the WAN side port P12, the main router 10 transfers the IP packet from the port 13 of the route S to the backup router 20. The backup router 20 transfers the IP packet that has entered from the port P23 to the port P21 that has been deblocked, and outputs it to the LAN segment through the hub 11. Thus, when a failure occurs on the LAN side of the main router 10 and the default gateway is maintained by the main router 10, the blocking port P21 of the backup router 20 is released. As a result, the LAN segment can send and receive IP packets to and from the opposite base through the backup router 20, the transfer route S, the main router 10 and the main line 12m, and the physical and logical asymmetry as described above. Routing does not occur. The process up to output from the backup router to the hub 11 is the same as that in the case where the WAN side port is down due to the main line failure already described (FIG. 15), and the description is omitted.

3.3) Effect As described above, according to the present embodiment, a transit route is set between the main router and the backup router, and the LAN side port of the backup router is normally set in a blocking state. When a failure is detected on the main router itself or on the LAN side of the main router, blocking of the LAN port of the backup router is released, and packets can be transmitted and received between the backup router and the LAN through the LAN side port. . By such blocking control of the LAN side port in the backup router, it is possible to prevent physical asymmetric routing of transmission and reception traffic in a configuration in which the customer side edge router is made redundant. In addition, by setting a transition segment and setting a value lower than the dynamic routing cost value of the LAN side port of the main router and backup router for the transition segment, between the network devices performing transmission and reception There is no difference in the calculated cost value. Accordingly, in the network device in which route information is shared by the dynamic routing protocol, there is no unintentional error in route calculation by adopting this embodiment.

4). Third Example 4.1) Configuration of Router As illustrated in FIG. 18, a router 300 according to this example is connected to a LAN interface unit 301 including a plurality of LAN interfaces # 1, # 2,. A WAN-side interface 302 and a memory 303. In this embodiment, in the LAN side interface unit 301, the physical port P1 of the interface # 1 is directly connected to the hub of the LAN segment A, and the physical port P2 of the interface # 2 is directly connected to the hub of the LAN segment B And Further, the physical port P3 of the interface # 3 is connected to a redundant neighboring router, and a transition route S by dynamic routing is set as will be described later.

  The memory 303 stores a routing table 303R, an ARP table 303A, and a MAC address table 303M. In addition to the dynamic route, a direct route is set in the routing table 303R, and the correspondence relationship between the destination IP address, the MAC address, and the port number is stored in the ARP table 303A according to the ARP protocol. The MAC address table stores the correspondence between the MAC address and the port number, and the switching port is mainly used in combination with the ARP table.

  The router control unit 304 controls the STP control unit 305, the routing control unit 306, and the redundant configuration control unit 307 according to the program stored in the program memory 308, thereby realizing the router routing function according to this embodiment. The functions of the router control unit 304, the STP control unit 305, the routing control unit 306, and the redundant configuration control unit 307 are executed by executing respective programs on a central processing unit (CPU) of the router. It can also be realized in terms of wear.

  The STP control unit 305 executes the STP protocol and blocks the port at the determined position. The STP control unit 305 uses a BPDU (Bridge Protocol Data Unit) to monitor a tree state including the routers 10 and 20 and the hub 11, and when a failure occurs, enables a blocking port to transmit and receive traffic. To.

  The routing control unit 306 performs normal routing control using the MAC address table 303M, the ARP table 303A, and the routing table 303R. As described above, the redundant configuration control unit 307 controls the redundant configuration with the adjacent routers according to a redundancy protocol such as VRRP or HSRP. As will be described later, in the case of a router in which a plurality of LAN segments are directly connected, the router control unit 304 performs the above STP control, routing control, and redundant configuration control for each LAN segment. Further, in the routing table 303R, in addition to the destination network address, the next hop, the output port, and the routing information source, normal routing priority (AD value or preference value) set for each type of routing information source is provided. Is stored.

  Hereinafter, a configuration example in which the router according to the second embodiment is made redundant as a CE router and its routing operation will be described in detail with reference to the drawings. The router according to the first embodiment can also be used.

5. Network configuration example 1
5.1) Configuration As illustrated in FIG. 19, the redundant configuration including the router 10 and the router 20 functions as a redundant CE router for each of the LAN segments A and B. That is, the router 10 is the main and the router 20 is the backup for the LAN segment A, and the router 10 is the backup and the router 20 is the main for the LAN segment B. The redundant configuration, STP blocking control, and routing operation of the routers 10 and 20 for one LAN segment are basically the same as the configuration and operation described in the above-described embodiment (FIG. 3).

  19, the router 10 has four physical ports P11, P12, P13, and P14, and the ports P11, P12, and P13 are connected to the LAN side interfaces # 1 to # 3 in the router 300 shown in FIG. , Port P14 is connected to the WAN side interface 302. Similarly, the router 20 has four physical ports P21, P22, P23, and P24, and the ports P21, P22, and P23 are connected to the LAN side interfaces # 1 to # 3 in the router 300 shown in FIG. The port P24 is connected to the WAN side interface 302.

  The redundant CE router for LAN segment A is configured as follows. In the router 10 which is the main router related to the LAN segment A, the port P11 is directly connected to the hub 11 of the LAN segment A, the port P13 is connected to the port P23 of the router 20, and the port P14 is provided by the WAN side 12m. It is connected to the. In the router 20 that is a backup router for the LAN segment A, the port P22 is directly connected to the hub 11 of the LAN segment A, the port P23 is connected to the port P13 of the router 10, and the port P24 is provided by the WAN side. It is connected to the. The LAN side port P22 of the backup router 20 related to the LAN segment A is in a blocking state due to STP.

  The redundant CE router for LAN segment B is configured as follows. In the router 20, which is the main router for the LAN segment B, the port P21 is directly connected to the hub 21 of the LAN segment B, the port P23 is connected to the port P13 of the router 10, and the port P24 is provided by the WAN side 22m. It is connected to the. In the router 10 which is a backup router for the LAN segment B, the port P12 is directly connected to the hub 21 of the LAN segment B, the port P13 is connected to the port P23 of the router 20, and the backup line 12b provided by the WAN side is the port P14. It is connected to the. The LAN side port P12 of the backup router 10 related to the LAN segment B is in a blocking state due to STP.

  The port P13 of the router 10 and the port P23 of the router 20 are set by dynamic routing. In this way, between the router 10 and the router 20, the crossing routes S (A) and S (B) are set for the LAN segments A and B, respectively.

  Ports P14 and P24 of routers 10 and 20 are used as routed ports or switching ports associated with SVI, and P11, P12, P13, P21, P22, and P23 also associated with SVI are used as switching ports. It is only necessary to connect the cable to the hub 11 in the segment A and to the hub 21 in the LAN segment B. In this way, by using a single segment hub on the LAN side, the user does not need to be aware of the connection port, and it is possible to prevent connection mistakes, occurrence of loops, etc. during replacement work at the time of failure.

  As described above, the redundant CE router related to the LAN segment A and the redundant CE router related to the LAN segment B perform substantially the same operations as those in the second embodiment (FIGS. 13 to 17) described above, and various faults occur. The occurrence of physical asymmetric routing can also be avoided with respect to the occurrence. Therefore, in the following description, the routing operation of the redundant CE router related to the LAN segment A is exemplified, but the same applies to the redundant CE router related to the LAN segment B.

5.2) Routing operation <First example>
As shown in FIG. 20, if there is no failure in the line between the LAN segment A and the main router 10 and the main line 12m between the router 10 and the opposite base C, the LAN segment A is connected to the main router 10 and the main line 12m. Normal communication can be performed with the opposite site C through (route Ra-c). Similarly, if there is no failure in the line between the LAN segment A and the main router 10 and the main line 12m between the main router 10 and the opposing base D, the LAN segment A passes through the main router 10 and the main line 12m. Normal communication can be performed with D. That is, the main router 10 transfers the IP packet entered from the port P11 of the LAN segment A to the WAN side port P14 by dynamic routing, and transmits it to the opposite site C or D through the main line 12m. Conversely, the main router 10 transfers the IP packet that has entered from the WAN side port P14 to the LAN side port P11 by directly-connected, and outputs it to the LAN segment A. Therefore, in a state where no failure has occurred, the LAN segment A can transmit and receive IP packets between the opposite bases C and D through the main router 10 and the main line 12m, and between the transmission traffic and the reception traffic. Asymmetric routing does not occur. Regarding LAN segment A, the LAN side port P22 of the backup router 20 is in the STP blocking state, and for LAN segment B, the LAN side port P12 of the backup router 10 is in the STP blocking state. Formation is prevented.

  Assume that a failure has occurred on the main line 12m side of the opposite site D in the normal communication state. When the main router 10 detects the occurrence of a failure on the WAN main line 12m, the communication between the LAN segment A and the opposite base D is a backup through the main router 10, the transfer route S, the backup router 20, and the backup line 22b. Switch to communication. Specifically, when the main router 10 detects the occurrence of a failure on the main line 12m, the main router 10 transfers the IP packet received from the port P11 of the LAN segment A to the port P13 by dynamic routing, and outputs it to the backup router 20 through the crossing route S. To do. The backup router 20 transfers the IP packet that has entered from the port P23 of the crossover route S to the WAN side port P24 by dynamic routing, and transmits it to the opposite site D through the backup line 22b (path Ra-d).

  Conversely, when an IP packet enters from the WAN side port P24, the backup router 20 is directly connected to the LAN segment A, and the port P22 having the highest routing priority is in a blocking state. Since the MAC address of the destination IP that is directly connected is registered via the transfer route S, the packet is transferred to the port P23 on the transfer route S side, and is output to the main router 10 through the transfer route S. The main router 10 transfers the IP packet that has entered from the port P13 of the transit route S to the LAN side port P11 having the highest routing priority, and outputs it to the LAN segment A through the hub 11 (path Ra-d).

  Thus, when a failure occurs on the main line 12m side of the opposite site D, the LAN segment A communicates with the opposite site D through the route Ra-d of the main router 10, the transfer route S, the backup router 20, and the backup line 22b. Can continue to transmit and receive IP packets, and the physical asymmetric routing as described above does not occur.

<Second example>
As shown in FIG. 21, when normal communication is performed between the LAN segment A and the opposite site D through the main router 10 and the main line 12m, a failure occurs on the main line 12m side of the router 10. As in the first example, the communication between the LAN segment A and the opposite site D is switched to the backup communication through the main router 10, the transfer route S, the backup router 20, and the backup line 22b. In detail, since it is the same as that of a 1st example, description is abbreviate | omitted.

  However, if the failure location is the WAN side IF of the main router 10, the communication path is as shown in FIG. 21, but the router 10 passes through L2 when the default gateway of the LAN segment A fails over to the router 20. The data is transferred from the router 20 to the backup line by dynamic routing. In contrast, when an IP packet enters the WAN side port P24, the backup router is directly connected to the LAN segment A and the port P22 having the highest routing priority is in a blocking state, and is also logically directly connected to the LAN segment A. Since the MAC address of the destination IP is also registered via the transfer route S, the packet is transferred to the port P23 on the transfer route S side, and is output to the main router 10 through the transfer route S. This prevents asymmetric routing, both physically and logically.

<Third example>
As shown in FIG. 22, when a failure occurs in the main router 10 during normal operation through the main router 10, the backup router 20 takes over the role of the default gateway of the LAN segment A, and the LAN side for the LAN segment A The STP blocking of the port P22 is released. Thus, the LAN segment A can continue communication with the opposite base D through the backup router 20 and the path Ra-d of the backup line 22b. More specifically, the backup router 20 plays the role of the default gateway of the LAN segment A from the router 10 when the keep alive (Keepalive) for determining the master / slave relationship of the default gateway using the multicast packet from the main router 10 is interrupted. take over. As a result, the IP packet entered from the port P22 of the LAN segment A is transferred to the port P24 by dynamic routing and transmitted to the opposite site D through the backup line 22b.

  Conversely, when an IP packet is received from the WAN side port P24, the backup router 20 transfers the packet to the LAN side port P22 having the highest routing priority that is released from the blocking, and outputs it to the LAN segment A through the hub 11. Thus, even when a failure occurs in the main router 10, the LAN segment A can send and receive IP packets to and from the opposite site D through the path Ra-d of the backup router 20 and the backup line 22b. The physical and logical asymmetric routing as described above does not occur.

<Fourth example>
As shown in FIG. 23, when normal communication is performed between the LAN segment A and the opposite site D through the main router 10 and the main line 12m, if a failure occurs on the LAN side of the router 10, the main router 10 continues to operate as the default gateway of the LAN segment A, and the backup router 20 cancels the STP blocking of the LAN side port P22 for the LAN segment A. In this way, the LAN segment A can continue communication with the opposite base D through the backup router 20, the transfer route S, the main router 10, and the route Ra-d of the main line 22m.

6). Network configuration example 2
6.1) Configuration In FIG. 24, the redundant configuration including the router 10 and the router 20 functions as a redundant CE router with respect to the LAN segment A, and the LAN segment A is directly connected to the redundant CE routers 10 and 20. . Further, the LAN segment A is connected to redundant firewalls FW1 and FW2. The redundant configuration of the routers 10 and 20 regarding the LAN segment A, the STP blocking control, and the routing operation are basically the same as the configuration and operation described in the second embodiment (FIGS. 13 to 17).

  In the redundant firewalls FW1 and FW2, the firewall FW1 is the main and the firewall FW2 is the backup, and the firewalls FW1 and FW2 are connected by an HA (High Availability) link and can notify each other of their operating states. When the interface on the main router 10 side goes down due to HA, the firewall FW1 operates to fail over from the firewall FW2.

  As shown in FIG. 24, the router 10 has three physical ports P11, P13, and P14, and the ports P11 and P13 are respectively connected to the LAN side interfaces # 1 and # 3 in the router 300 shown in FIG. The port P14 is connected to the WAN side interface 302. Similarly, the router 20 also has three physical ports P21, P23, and P24. The ports P21 and P23 are respectively connected to LAN side interfaces # 1 and # 3 in the router 300 shown in FIG. 15, and the port P24 is WAN. Is connected to the side interface 302.

  The redundant CE router for LAN segment A is configured as follows. In the router 10 which is the main router related to the LAN segment A, the port P11 is directly connected to the LAN segment A, the port P13 is connected to the port P23 of the router 20, and the port P14 is connected to the main line 12m provided by the WAN side. ing. In the router 20, which is a backup router for the LAN segment A, the port P21 is directly connected to the LAN segment A, the port P23 is connected to the port P13 of the router 10, and the port P24 is connected to the backup line 22b provided by the WAN side. ing.

  The port P13 of the router 10 and the port P23 of the router 20 are set by dynamic routing.

  As described above, the redundant CE router related to the LAN segment A performs substantially the same operation as that of the second embodiment (FIGS. 13 to 17) described above, and can perform physical asymmetric routing even when various failures occur. Occurrence can be avoided. Hereinafter, an access operation to the firewall FW1 or FW2 which is a communication device in the LAN segment A will be described with reference to the drawings.

6.2) Access operation <First example>
In FIG. 25, when there is no network failure, as in the routing operation shown in FIG. 20, access and communication from the opposite base C to the firewall FW1 are possible through the main router 10 and the main line 12m (route Ra). -c). Similarly, access and communication from the opposite site D to the firewall FW1 are possible through the main router 10 and the main line 12m. At that time, similarly to the routing operation shown in FIG. 17, physical and logical asymmetric routing does not occur between transmission traffic and reception traffic.

  Further, when a failure occurs on the main line 12m side of the counter site D, the interface on the main router 10 side is not down, so that no failover occurs from the firewall FW1 to the firewall FW2. In this case, similarly to the routing operation shown in FIG. 17, the access and communication from the opposite site D to the firewall FW1 continue through the route Ra-d of the main router 10, the transfer route S, the backup router 20, and the backup line 22b. And a physical asymmetric routing as described above does not occur.

<Second example>
As shown in FIG. 26, even when a failure occurs on the main line 12m side of the router 10, the access and communication from the opposite base D to the firewall FW1 are performed by the main router 10, the crossover route, as in the first example. S, through the backup router 20 and the backup line 22b.

  However, when the failure occurs at the WAN side IF of the main router 10, the communication path is as shown in FIG. 26. However, when the default gateway fails over to the router 20, the router 10 passes through L2, and from the router 20 It is transferred to the backup line by dynamic routing. Conversely, when an IP packet enters the WAN side port P24, the backup router is directly connected to the LAN segment A and the port P21 having the highest routing priority is in a blocking state, and is also logically directly connected to the LAN segment A. Since the MAC address of the destination IP is also registered via the transfer route S, the packet is transferred to the port P23 on the transfer route S side, and is output to the main router 10 through the transfer route S. This prevents asymmetric routing, both physically and logically.

<Third example>
As shown in FIG. 27, when a failure occurs in the main router 10 during normal operation through the main router 10, the backup router 20 takes over the role of the default gateway and performs STP blocking on the LAN side port P21 of its own station. Is released. Further, when the interface on the main router 10 side goes down, failover occurs from the firewall FW1 to the firewall FW2. Therefore, the access and communication from the opposite site D to the firewall FW2 can be continued through the backup router 20 and the backup line 22b as in the routing operation shown in FIG. Asymmetric routing does not occur.

<Fourth example>
As shown in FIG. 28, it is assumed that a failure occurs on the LAN side of the router 10 when normal communication is performed through the main router 10 and the main line 12m. In this case, since the interface on the main router 10 side goes down, failover occurs from the firewall FW1 to the firewall FW2. Communication between the LAN segment A and the opposite site D can be continued through the backup router 20, the transfer route S, the main router 10, and the main line 12a. Does not occur.

  The present invention can be applied to a redundant configuration of CE routers.

10 router (relay device)
11 Hub 12 WAN side line 12m Main line 12b Backup line 20 Router (relay device)
21 Hub 22 WAN side line 300 Router 301 LAN side interface unit 302 WAN side interface 303 Memory 303R Routing table 303A ARP table 304 Router control unit 305 STP control unit 306 Routing control unit 307 Redundant configuration control unit 308 Program memory

Claims (7)

A method for preventing occurrence of asymmetric routing in a redundant configuration of a plurality of relay devices connected directly to a local area network (LAN) and provided between the LAN and a wide area network (WAN),
A first relay device that operates as a main relay device and a second relay device that operates as a backup relay device set a route between them,
When the second relay device sets the LAN side port of its own station to a frame transmission / reception inhibited state, and the frame transmission / reception inhibited state is detected on the first relay device itself or on the LAN side of the first relay device. To release,
Asymmetric routing prevention method.   The setting and cancellation of the frame transmission / reception suppression state of the LAN side port of the second relay device is based on the MAC address learning suppression setting and cancellation, according to the STP (Spanning Tree Protocol) protocol, or the operation status of the first relay device The asymmetric routing prevention method according to claim 1, wherein the method is executed by blocking or opening a port triggered by detection of message output or detection by a network monitoring function.   The method of preventing asymmetric routing according to claim 1 or 2, wherein a transfer segment is set in the transfer route, and a dynamic routing cost value added to outgoing and return communication is uniformly adjusted. A redundant configuration of a plurality of relay devices connected directly to a local area network (LAN) and provided between the LAN and a wide area network (WAN),
A first relay device operating as a main relay device;
A second relay device that operates as a backup relay device and is connected to the first relay device via a crossover route;
The second relay device is
Monitoring means for monitoring failure occurrence on the LAN side of the first relay device itself or the first relay device;
Control means for setting the LAN side port of the second relay device to a frame transmission / reception inhibition state and releasing the frame transmission / reception inhibition state when the occurrence of the failure is detected;
Redundant configuration having   The control means sets the frame transmission / reception suppression state of the LAN side port by MAC address learning suppression setting and cancellation, by STP (Spanning Tree Protocol) protocol, or by port blocking and opening triggered by detection of Syslog output. The redundant configuration according to claim 4, wherein release is executed. A relay device directly connected to a local area network (LAN) and provided between the LAN and a wide area network (WAN),
Redundant configuration control means that operates as a main relay device or a backup relay device in a redundant configuration with an adjacent relay device;
When the relay device operates as a backup relay device, the adjacent relay device itself operating as the main relay device or monitoring means for monitoring the occurrence of a failure on the LAN side of the adjacent relay device;
When the relay device operates as a backup relay device, a control means for setting the LAN side port of the relay device to a frame transmission / reception suppression state and releasing the frame transmission / reception suppression state when the occurrence of the failure is detected;
A relay device.   The control means sets the frame transmission / reception suppression state of the LAN side port by MAC address learning suppression setting and cancellation, by STP (Spanning Tree Protocol) protocol, or by port blocking and opening triggered by detection of Syslog output. The relay apparatus according to claim 6, wherein cancellation is executed.

Images