JP5427665B2 - Spanning tree reconfiguration method and communication apparatus - Google Patents

Description

  The present invention relates to a technique for localizing a range for reconfiguring a spanning tree in a wide area Ethernet service.

The wide area Ethernet service interconnects LANs (Local Area Networks) of geographically distant subscribers with an Ethernet interface. The service can communicate between devices on the interconnected LAN even if a communication protocol other than IP (Internet Protocol) is used in the LAN. In addition, the service implements services with various topologies.
MEF (Metro Ethernet Forum) is an organization that prescribes wide area Ethernet services. Point-to-point connection E-LINE service, point-to-multipoint connection E-TREE service, and multipoint connection E-LAN service Three types of services are defined. It is defined as a specification that the service can select STP (Spanning Tree Protocol) defined by IEEE 802.1D by a telecommunications carrier's PE (Provider Edge) device, and can select transmission, termination, and destruction (for example, Non-Patent Document 1). ).

  In a subscriber's LAN interconnected by a wide area Ethernet service, when the LAN is interconnected by a service provided by another communication carrier, a loop path across the wide area Ethernet service is configured. When using the wide area Ethernet service, there is a problem that such a loop route is accidentally formed and it is necessary to avoid the failure of the subscriber network.

  On the other hand, in order to improve the reliability of interconnection by wide area Ethernet service, OAM (Operation Administration and Maintenance) for detecting a failure and APS (Automatic Protection Switching) for automatically switching a communication path when a failure occurs are described in ITU- T (International Telecommunication Union-Telecommunication sector), IEEE (Institut of Electrical and Electronic Engineers), etc. are standardizing work. In response to this, telecommunications carriers have applied OAM and APS to the communication path of the wide-area Ethernet service provided by their own companies to improve reliability. However, assuming large-scale disasters such as earthquakes and power outages, OAM and APS It cannot be said that it can respond sufficiently. This is because it is assumed that redundancy is performed by one communication carrier. Therefore, the subscriber uses the interconnection service provided by a plurality of communication carriers to make the communication path redundant and improve the reliability. Since such a configuration is the same as the above-described loop route, there is a problem that the subscriber must be able to intentionally perform the configuration in order to improve the reliability of the communication route.

  In order to solve the above problems and problems, it has been proposed to operate STP defined in IEEE 802.1D in a switch constituting a subscriber's LAN. By operating the STP, it is possible to automatically detect and eliminate a loop path that is accidentally or intentionally configured.

  In Patent Document 2, a communication device that provides a wide area Ethernet service provides a spanning tree root bridge function to a subscriber's LAN, thereby constructing one spanning tree in the interconnected subscriber's LAN. It is described that makes it possible to do.

  However, when STP is operated between LANs of subscribers interconnected by a wide area Ethernet service, a device configuration change event (device failure / addition, line failure, device setting) occurred within a certain subscriber's LAN. Change and topology change) are spread to other LANs of subscribers interconnected by wide-area Ethernet services that do not need to be affected. Spanning tree reconstruction and FDB (Filtering Database) are performed by switches in the LAN. Reset processing may be performed. In the case of STP, when the spanning tree reconstruction process is performed, data communication is interrupted for about 50 seconds. Further, when the FDB is reset, unnecessary traffic flooding may occur for MAC address learning, and the network load may increase.

  Patent Document 3 describes in which spanning tree instance when a VLAN ID is added / deleted by changing the method of using the BPDU field of Multiple Spanning Tree Protocol (MSTP) defined in IEEE 802.1Q. It does not affect MSTI (Multiple Spanning Tree Instance) that has been identified and VLAN ID is not added / deleted, and it is possible to provide a highly reliable wide area LAN service without restructuring the topology. Providing a multiple spanning tree protocol network. Specifically, the network device includes a network identification information processing unit that creates network identification information for each topology, a reception unit that receives and extracts network identification information from neighboring devices, the extracted network identification information, A comparison unit that detects a change by comparing with the network identification information of its own device generated by the network identification information processing unit, and when the change is detected by the comparison unit, the topology (spanning tree instance) in which the change is detected And a topology change detection processing unit having a topology information construction unit that performs reconstruction only for the document.

  Further, in the technique described in Patent Document 1, when it is determined that a topology change notification is received from a network device belonging to a second region different from the first region to which the network device belongs in MSTP, the topology change notification is sent to the first region. Although it is disclosed to determine whether or not to transmit to a network device belonging to one region, the root bridge ID (bridge MAC address and priority) is used to determine whether or not the topology change notification is transferred to another region. The value is a combination of degrees).

JP2007-97004 JP2007-158800 WO2005 / 057863

IEEE 802.1D-1998 Chapter 8

  However, the invention described in Patent Document 2 has a problem that it does not provide a function for localizing a configuration change event that has occurred in a subscriber's LAN. In other words, even if the influence range of the generated configuration change event converges in the LAN in which the event occurs and does not reach the entire network including the other LAN, the configuration change event is transferred to the other LAN. An increase in network load due to traffic flooding caused by unnecessary FDB reset due to a configuration change event occurs.

  In the technique described in Patent Document 1, since a communication path between network devices existing at the boundary of a region is fixed, it is determined whether or not a configuration change in another region affects its own region. It was sufficient to judge based on the size of the root bridge ID (a value obtained by combining the MAC address and priority of the bridge). However, since the communication path between the communication devices is not normally fixed in STP, reconfiguration is performed in the case of a configuration change such as a change of the shortest path to the root bridge, so whether or not localization is possible is determined. There is a need to. Even when the same root bridge is operating and a configuration change event such as a change of the shortest route to the root bridge occurs, it is necessary to determine whether or not local localization is possible.

  In addition, in the technology described in Patent Document 3, when a VLAN is added / deleted to one of multiple managed STP instances in the wide area LAN service, the purpose is to suppress the influence on other instances. Therefore, it is not determined whether or not the configuration change is localized within one STP instance. In addition, there was a problem with the connectivity with the MSTP installed in a communication device that has not been expanded.

  The present invention prevents a configuration change event occurring in an arbitrary LAN interconnected by a wide-area Ethernet service from being transmitted to another interconnected LAN and performing unnecessary processing such as spanning tree reconstruction. For the purpose.

  In addition, even when a spanning tree is configured with one LAN, by applying the apparatus of the present invention, it is determined whether or not it can be localized within a specific range, and shortening of data interruption within one LAN. Another object is to realize a reduction in the amount of packets.

  In addition, the E-TREE service, which is one of the wide area Ethernet services defined by MEF, aims to operate the STP correctly regardless of the location where the root bridge of the spanning tree is placed.

  The present invention monitors the STP packets exchanged between interconnected LANs by a PE device that provides a wide area Ethernet service, and grasps the root bridge of the spanning tree to be configured. The most important feature is to correctly recognize the range to be performed and to localize the influence of the event.

  The spanning tree restructuring method of the present invention has no effect by identifying whether the influence range of the generated configuration change event converges within the LAN in which the event has occurred or extends to the entire network including other LANs. Since the configuration change event is not forwarded to the determined LAN, it suppresses an increase in the network load due to data communication interruption due to unnecessary spanning tree reconstruction associated with the configuration change event and traffic flooding caused by FDB reset it can. In addition, even when a spanning tree is configured with one LAN, by applying the apparatus of the present invention, it is determined whether or not it can be localized within a specific range, and shortening of data interruption within one LAN. Alternatively, the amount of packets can be reduced. In addition, the STP can always operate correctly regardless of the location where the root bridge of the spanning tree is arranged in the E-TREE service.

The figure explaining the STP proxy response of PE apparatus to which the present invention is applied Diagram of redundant equipment and communication paths in wide area Ethernet service The figure explaining operation | movement of STP in the network comprised by the conventional PE apparatus. The figure explaining the flow of processing at the time of STP packet reception in PE apparatus The figure explaining the flow of processing at the time of Config BPDU reception The figure explaining the flow of processing at the time of TCN reception The figure which showed the contents of the information related to the root bridge held in the PE device The figure which shows the flow of an STP packet when a root bridge is arrange | positioned at the root segment of an E-TREE service. The figure which shows the flow of an STP packet when a root bridge is arrange | positioned to the leaf segment of E-TREE service. The figure which shows the change of the flow of a STP packet when a root bridge is arrange | positioned to the leaf segment of E-TREE service by applying this invention. The figure explaining the flow of a process of the Hello packet extended in order to use STP with an E-TREE service One configuration example in which a switch to which the present invention is applied is arranged in a corporate LAN One configuration example in which a switch to which the present invention is applied is arranged in a corporate LAN Example of internal structure of communication apparatus to which the present invention is applied The figure which showed the content of the information of the port which may transfer Config BPDU hold | maintained in PE apparatus The figure which showed the contents of the information of the port with which the device held in the PE device is equipped The figure explaining the flow of the master process of the hold timer The figure explaining the flow of the process which defers the update of the bridge information hold | maintained in PE apparatus by reception Config BPDU. Diagram explaining the flow of processing executed when the hold timer times out The figure explaining the flow of the processing performed at the time-out of a Hello timer

  Implemented a method to localize the effects of spanning tree restructuring to the minimum necessary when using STP to automatically detect and resolve loop paths in LANs interconnected using wide area Ethernet services .

  FIG. 1 is a network configuration diagram showing an embodiment of a spanning tree reconfiguration method of the present invention. A wide area Ethernet service is provided by a network 500 including 10 to 13 PE devices. Reference numerals 100, 110, 120, and 130 denote subscriber LANs interconnected by the service. Reference numerals 1 to 4 denote switches constituting the subscriber's LAN, which operate the STP for detecting and eliminating the loop path.

  Here, when a network configuration change occurs in the subscriber's LAN 120, the PE device 12 of the network 500 adjacent to the subscriber's LAN 120 does not change the subscriber's LAN 120 before the network configuration change occurs. Based on the content of the STP packet exchanged between the LANs, it is determined whether the configuration change can be localized in the LAN 120. When it is determined that the PE device 12 can be localized, the STP packet indicating the configuration change transmitted from the switch 3 to the network 500 is discarded according to the content of the configuration change event. Alternatively, the PE device 12 responds to the switch 3 by configuring the STP packet that the switch 1, 2, or 4 originally responds to the STP packet indicating the above-described configuration change as a proxy. If it is determined that the PE device 12 cannot be localized, the STP packet 50 is transferred to the subscriber LANs 100, 110, and 130 as usual.

  Figure 2 shows that when large-scale disasters such as earthquakes and power outages are assumed, OAM and APS are not always sufficient, and the interconnection service provided by multiple carriers is used to make communication paths redundant. In order to improve reliability, an example will be described in which a redundant configuration is applied to a communication device and a communication path between two LANs. The network 500 is a network provided by the first communication carrier. The PE devices 10 and 11 are located at the edge of the network 500 and are connected to the subscriber's LANs 100 and 110 via the network. The switch 1 is connected to the network 500 of the first carrier and is located in the subscriber's LAN 100. The network 600 is a network provided by the second communication carrier. The PE apparatuses 20 and 21 are located at the edge of the network 600 and are connected to the subscriber's LANs 100 and 110 via the network. The switch 2 is connected to the network 500 of the second communication carrier and is located in the subscriber's LAN 100. When the LAN 100 subscriber device communicates with the LAN 110 subscriber device, the LAN 100 subscriber device communicates via the networks 500 and 600 of the first communication carrier and the second communication carrier to realize redundancy. ing.

  In this configuration, two LANs are connected by services of different telecommunications carriers, but it is difficult for the PE devices of each telecommunications carrier to cooperate to eliminate the loop path. Therefore, it is necessary to operate the spanning tree protocol in the LAN to eliminate the loop path. For example, in this assumed environment, the present invention can be applied to the switches 1 to 4 constituting the subscriber's LAN or the PE devices 10 to 13 of the communication carrier.

  FIG. 4 shows the flow of processing when the PE device 12 in FIG. 1 targeted by the spanning tree reconfiguration method of the present invention receives an STP packet. In the present invention, Configuration BPDU (Configuration BPDU) and Topology Change Notification BPDU (TCN) are handled as STP BPDU (Bridge Protocol Data Unit) for transmitting a configuration change event.

  When the PE apparatus 12 receives a packet from the switch 3 at the input processing unit 1010 via the Ethernet IFs 601-1 to 601-n illustrated in FIG. 14, the NIF management unit 1000 performs the following processing. It is determined whether the received packet is an STP control packet based on the information in the received packet. If the received packet is an STP control packet, the flow of FIG. 4 is executed. If the STP packet is a Config BPDU packet based on the received STP control packet information (S100), the PE device 12 performs Config BPDU processing (S120). The Config BPDU process will be described later with reference to FIG. When the Config BPDU process is performed, the flow of FIG. 4 ends. On the other hand, if the received STP control packet is not a Config BPDU packet in S100, it is determined whether or not the received STP control packet is a topology change notification BPDU packet (S110). If the received STP control packet is a TCN packet, TCN processing is performed (S130). The TCN process will be described later with reference to FIG. When it is determined in S110 that the STP control packet is not a TCN packet, or when the TCN process in S130 is completed, the flow in FIG. 4 is terminated.

  FIG. 5 shows the flow of processing when a Config BPDU is received by the PE device.

In the NIF management unit 1000 of the PE apparatus in which the spanning tree reconfiguration method of the present invention is implemented, the value of the root bridge ID indicated by the received Config BPDU and the PE apparatus up to now are determined based on the table shown in FIG. The values of the root bridge ID values of the currently constructed spanning tree that monitors, recognizes and holds the STP control packet are compared (S301).
If the root bridge ID indicated by the received Config BPDU matches the root bridge ID held by the PE, the value of the path cost to the root bridge indicated by the received Config BPDU and the path cost to the root bridge held by the PE The values are compared in magnitude (S302).

  The root bridge ID indicated by the received Config BPDU is smaller than the root bridge ID held by the PE, or the root bridge ID indicated by the received Config BPDU is equal to the root bridge ID held by the PE, and the received Config BPDU indicates When the path cost to the root bridge is smaller than the path cost to the root bridge held by the PE, the hold timer stop process (S310) is performed, the Hello timer stop (S320), and the PE is determined by the information indicated by the received Config BPDU. The contents of the information related to the stored root bridge, that is, the table of FIG. 7 is updated (S330).

  Since the Config BPDU holds values corresponding to the fields of the table of FIG. 7, this update copies values from the fields of the received Config BPDU to the table. With this update, it is possible to always keep the latest information in the table shown in FIG. 7 for information necessary for changing the configuration of the target network. Therefore, it is possible to appropriately cope with a configuration change that occurs in the network. In the case of S330, that is, the root bridge ID indicated by the received Config BPDU is smaller than the root bridge ID 704 held by the PE, or the root bridge ID indicated by the received Config BPDU is equal to the root bridge ID 704 held by the PE. If the path cost to the root bridge indicated by the received Config BPDU is smaller than the path cost 704 to the root bridge held by the PE, the network configuration change by the configuration information carried by the Config BPDU cannot be localized. In the output processing unit 1020, the PE device transfers the Config BPDU to all ports other than the port that has received the Config BPDU (S340). For example, when the path cost to the root bridge changes due to a change in the root bridge or a change in the network configuration, the configuration change cannot be localized.

  In the NIF management unit 1000, the root bridge ID indicated by the received Config BPDU is equal to the root bridge ID held by the PE (S301), and the path cost to the root bridge indicated by the received Config BPDU is the route held by the PE. When it is equal to the path cost to the bridge (S302), the PE device determines that the Config BPDU is an STP packet (Hello packet) periodically transmitted by the root bridge of the spanning tree, and stops the hold timer. (S311) is executed, the Hello timer is stopped (S321), and the output processing unit 1020 turns off the Topology Change flag in the packet and switches to all ports other than the port that received the Config BPDU. Send (S350) and start the Hello timer (S360).

  A default value of the node management unit is set as the timeout time set in the Hello timer. In addition, the timeout time set in the Hello timer can be changed by the configuration of the network operated by the administrator. By this determination, it is possible to prevent an apparatus of another segment that is not affected by the configuration change from unnecessarily resetting the FDB with respect to the configuration change on the segment arranged by the root bridge. In addition, a Config BPDU that is not accompanied by a configuration change and is not a Hello packet can be discarded by the PE device.

  In the NIF management unit 1000, the root bridge ID indicated by the received Config BPDU is larger than the root bridge ID held by the PE (S 301), and the Config BPDU is received from the port that received the Config BPDU from the root bridge held by the PE device. Is received (S303), or the root bridge ID indicated by the received Config BPDU is equal to the root bridge ID held by the PE (S303), and the path cost to the root bridge indicated by the received Config BPDU is equal to the PE. If it is larger than the path cost to the root bridge held by (S302), a process of holding the bridge information update is implemented (S370). The process of deferring bridge information update will be described with reference to FIG. For example, when the spanning tree information learned so far is erased by restarting the subscriber unit, the own unit tries to reconfigure it as a root bridge. In response to this operation, information from the root bridge is directly transmitted to the switch arranged in the same segment as the root bridge, so that the PE device does not need to make a proxy response. However, when the switch recognized by the PE as the root bridge is not operating due to a failure or the like, it is also assumed that another switch becomes the root bridge. In order not to hinder the replacement operation of the root bridge, even when the bridge ID is higher or the path cost is higher than the information of the root bridge held by the PE, the information received for a certain period can be held and dealt with.

  Further, in the NIF management unit 1000, the root bridge ID indicated by the received Config BPDU is larger than the root bridge ID held by the PE (S301), and other than the port that received the Config BPDU from the root bridge held by the PE device. When the Config BPDU is received from the network (S303), the PE device configures the Config BPDU with reference to the root bridge information held in the table shown in FIG. 7, and transmits the Config BPDU to the port that has received the Config BPDU (S380). The PE device responds as a proxy to support the rapid convergence of learning operation by STP. That is, since the PE device does not transfer the Config BPDU to a port other than the port that has received the Config BPDU, the amount of communication can be reduced. Furthermore, since the subscriber apparatus that has transmitted the Config BPDU can receive the Config BPDU in which the information on the root bridge is updated not from the root bridge but from the PE apparatus, the information about the root bridge can be obtained quickly.

FIG. 17 shows the flow of the hold timer stop process.
The NIF management unit 1000 performs a hold timer stop process (S800), and initializes a save data area for hold that stores bridge information held by the PE (S810). In this initialization processing, the maximum value of the ID is set as the bridge ID as the bridge information, and the maximum value of the path cost is set as the path cost. Therefore, the switch having a large bridge ID in the Config BPDU information received after this processing But it can be treated as a root bridge candidate. The holding data area for holding bridge information has the data structure shown in FIGS. 7, 15, and 16, and one entry is prepared for each port of the PE device.

FIG. 18 shows a flow of bridge information update suspension processing.
The NIF management unit 1000 checks whether the hold timer is operating (S500). If the hold timer is stopped, the hold timer is started (S510). The timeout value set in the hold timer is set to a default value possessed by the node management unit. The timeout time set in the hold timer can be changed by the configuration of the network operated by the administrator. Next, the bridge ID of the storage data for holding is compared with the bridge ID of the received Config BPDU (S520). When the bridge ID of the stored data for holding is larger than the bridge ID of the received Config BPDU, or when the bridge ID is equal and the path cost of the stored data for holding is larger than the path cost of the received Config BPDU, the bridge information indicated by the received Config BPDU Is stored in the reserved storage data area (S540). In this case, the information on the root bridge held by the PE device is updated temporarily by the received Config BPDU, and the transfer of the received Config BPDU to a port other than the receiving port is temporarily held until the hold timer times out. As a result, when the switch that the PE device recognizes as the root bridge cannot operate normally as a root bridge due to a failure or the like, the other switch with a bridge ID higher than the previous root bridge can become the root bridge. It does not disturb the original operation of STP. When the bridge ID of the storage data for holding is smaller than the bridge ID of the received Config BPDU, or when the bridge ID is equal and the path cost of the storage data for holding is less than or equal to the path cost of the received Config BPDU, the received Config BPDU is discarded. (S550). This makes it possible to temporarily store the information of the bridge with the smallest bridge ID and the lowest path cost among the Config BPDUs other than the root bridge received by the PE device while the hold timer is operating. To. That is, only the most appropriate bridge information as information for updating the information of the root bridge held by the PE device after the hold timer times out is held in the PE device as hold storage data.

FIG. 19 shows the flow of timeout processing of the hold timer.
The bridge information stored in the storage data area for holding is overwritten and copied to the area where the root bridge information is stored, and stored as new root bridge information (S600). A Config BPDU is configured with reference to the root bridge information held by the PE device, and the root bridge information is transmitted to a port other than the received port (S610). This notifies the switch other than the segment where the root bridge is arranged that the spanning tree is reconfigured and the root bridge is changed. Finally, the reserved storage data area for storing the bridge information held by the PE is initialized (S810). In this initialization processing, the maximum value of the ID is set as the bridge ID as the bridge information, and the maximum value of the path cost is set as the path cost. Therefore, the switch having a large bridge ID in the Config BPDU information received after this processing But it can be treated as a root bridge candidate.

FIG. 20 shows the flow of timeout processing of the Hello timer.
A data area for storing root bridge information held by the PE is initialized (S700). In this initialization processing, the maximum value of the ID is set as the bridge ID as the bridge information, and the maximum value of the path cost is set as the path cost. Therefore, the switch having a large bridge ID in the Config BPDU information received after this processing But it can be treated as a root bridge.

  FIG. 6 shows the flow of processing when a TCN is received.

  In the NIF management unit 1000, the PE device compares the port that has received the TCN with the port that has received information on the root bridge of the currently established spanning tree (S200).

  When the PE device receives a TCN from a port other than the port that has received the information of the root bridge of the currently constructed spanning tree held by the own device, the output processing unit 1020 localizes the processing of the TCN. Instead of the switch that the PE device should originally respond to, a proxy response is made. As a result, the TCN is not transferred to other LANs. In the output processing unit 1020, the PE device refers to the information stored in the table of FIG. 7 about the Config BPDU with the Topology Change Notification Acknowledgment flag turned on and the Config BPDU periodically transmitted from the root bridge. Then, the TC flag is turned on and transmitted to the port that received the TCN (S210). When the TCN is received from the port that received the Config BPDU from the root bridge held by the PE device, it can be seen that the TCN is generated and transmitted in the segment where the root bridge is arranged. Since the TCN is originally transmitted to the root bridge, the TCN is delivered to the root bridge without going through the present apparatus. Therefore, the received TCN is discarded in the output processing unit 1020 (S220). As a result, the PE apparatus can execute the configuration change without transferring the TCN to a port other than the port that has received the TCN, so that the configuration change can be localized.

  FIG. 7 shows the contents of information related to the root bridge held in the NIF management unit by the PE device in which the spanning tree reconfiguration method of the present invention is implemented. The information is stored in a table format, and the node management unit instructs the NIF management unit in the PE device to create an entry corresponding to the port in the device. The meaning of each field of the entry corresponds to the information transferred by the Config BPDU. The information of the root bridge ID 703, path cost 704, bridge ID 705, port ID 701, message age 707, maximum age 708, hello time 709, and transfer delay 710 .

  The port ID 701 is an ID for identifying the port equipped in the PE device within the device. The group ID 702 is an ID set by the node management unit, and identifies a set of ports that may transfer a Config BPDU received at a certain port. The value of the group ID takes a positive number from 0 to N. When the PE device is initialized, the node management unit instructs all NIF management units installed in the device to create entries corresponding to all ports. The node management unit can hold the initial value of the entry corresponding to each port in the storage device. When creating each entry, the node processing unit initializes the initial value information of the entry corresponding to the port if it is stored, and initializes it with the default value possessed by the node management unit if there is no corresponding information. When the Config BPDU is received, the NIF management unit having the received port searches using the port ID that received the corresponding entry as a key. The NIF management unit performs the STP control packet reception process shown in FIG. 4 with reference to the received Config BPDU and the corresponding entry. By providing this table, the PE device of the present invention can identify whether the affected range of the configuration change event that has occurred converges within the LAN in which the event has occurred, or extends to the entire network including other LANs. The configuration change event can be controlled not to be transferred to the LAN that is determined to have no influence.

  FIG. 15 shows the contents of information for identifying a set of ports that may be transferred with a Config BPDU created by an instruction from the node manager and held by the NIF manager. Each entry in this table includes a group ID 1501, an E-TREE flag 1502, an E-TREE STP flag 1503, and one or more port information 1504. The group ID 1501 is set by the NIF management unit according to an instruction from the node management unit, and identifies the set in the PE apparatus. The value of the group ID 1501 set as an entry in this table is a positive number from 1 to N. A group ID value of 0 indicates that the function of the present invention is disabled at the port. The E-TREE flag 1502 is set by the NIF management unit according to an instruction from the node management unit, and indicates whether or not the group provides the E-TREE service. If the value of the flag is 0, it indicates that a service other than E-TREE is provided, and if the value is 1, it indicates that an E-TREE service is provided. The E-TREE STP flag 1503 is set by the NIF management unit according to an instruction from the node management unit, and indicates whether the STP support function in the E-TREE service provided by the present invention is valid / invalid in the group that provides the E-TREE service. Show. If this flag value is 0, it is invalidated, and if it is 1, it is validated. The port information stores information about all the ports constituting the set. The contents of the port information 1504 are shown in FIG.

FIG. 16 shows the contents of the information of the ports equipped in the PE device. This information includes a leaf flag 1601 and a port ID 1602. When providing the E-TREE service specified by the MEF, the leaf flag 1602 is set to 1 by the NIF management unit according to the instruction of the node management unit when the port is connected to the leaf segment of the E-TREE. When the port is connected to the root segment or used in a service other than the E-TREE service, the NIF management unit sets 0 according to the instruction of the node management unit. FIG. 3 is provided by a conventional PE apparatus. 1 is a network configuration diagram showing an embodiment in which STP is used in a LAN interconnected by a wide area Ethernet service. FIG. A wide area Ethernet service is provided by a network 500 including 10 to 13 PE devices. Reference numerals 100, 110, 120, and 130 denote subscriber LANs interconnected by the service. Reference numerals 1 to 4 denote switches constituting the subscriber's LAN. Here, the STP is operated to detect and cancel the loop path.

  Here, when a network configuration change occurs in the subscriber's LAN 120, the switch 3 transmits an STP packet 50 that conveys the configuration change to the subscriber's LANs 100, 110, and 130 interconnected by the network 500. The switches 1, 2, and 4 of each LAN return responses 61, 62, and 64 to the received STP packet 50 to the switch 3. As described above, in the wide-area Ethernet service configured by the conventional PE apparatus, the configuration change information of the spanning tree is transmitted to the LANs of all the interconnected subscribers regardless of the contents thereof.

  In FIG. 8, subscribers' LANs 100, 110, 120, and 130 are interconnected by an E-TREE service defined by MEF. In FIG. 8, LAN 100 corresponds to the root segment of the E-TREE service, and LANs 110, 120, and 130 correspond to leaf segments. Communication devices arranged in the root segment can communicate with all leaf segment devices. On the other hand, a communication device arranged in the leaf segment can communicate with a communication device arranged in the root segment, but cannot communicate with a communication device arranged in another leaf segment.

  A switch 1 in FIG. 8 shows a root bridge of a spanning tree constructed by interconnected subscriber LANs. In this case, since the root bridge is arranged in the root segment 100, the STP packet 60 transmitted from the root bridge is transmitted to all the leaf segments, and the STP operates normally.

  FIG. 9 shows a configuration in which the root bridge 1 of the spanning tree constructed by the interconnected subscriber LANs is arranged in the leaf segment 110. In this case, the STP packet 60 transmitted from the root switch 1 is transmitted only to the root segment 100 and not transmitted to the other leaf segments 120 and 130. Therefore, when the root bridge of the spanning tree is arranged in the leaf segment of the E-TREE service in this way, STP does not operate normally.

  FIG. 10 expands the Hello packet transfer function (S350) shown in FIG. 5 of the present invention, so that the PE packet is transferred to the other leaf segment by the PE apparatus 10, and the BPDU is transmitted by the E-TREE service. Indicates the range to be transmitted. With this extension, the STP can operate normally even if the root bridge is arranged in the leaf segment of the E-TREE service.

  FIG. 11 shows an extended processing flow of the Hello packet transfer function (S350) shown in FIG. 5 so that the STP can be used even in an environment using the E-TREE service.

In the NIF management unit 1000, it is determined whether the PE device that has received the Hello packet periodically transmitted by the root bridge provides the E-TREE service defined by the MEF (S400). This determination procedure searches the corresponding entry in the table shown in FIG. 7 using the reception port ID 701 of the received Hello packet as a key. Next, the corresponding entry is searched in the table shown in FIG. 15 using the group ID 702 in the found entry as a key. If the E-TREE flag 1502 of the found entry is 1, it is determined that the E-TREE service is provided, and if it is 0, it is determined that the E-TREE service is not provided.
If the PE device does not provide the E-TREE service, the output processing unit 1020 turns off the TC flag and transfers the packet in addition to the port that has received the Hello packet (S470).

  In the NIF management unit 1000, when the PE device provides the E-TREE service, it is determined whether the function of the present invention is enabled to use the STP in the E-TREE service (S410). This determination procedure searches the corresponding entry in the table shown in FIG. 7 using the reception port ID of the received Hello packet as a key. Next, the corresponding entry is searched in the table shown in FIG. 15 using the group ID 702 in the found entry as a key. If the E-TREE STP flag 1503 of the found entry is 1, it is determined that this function is enabled, and if it is 0, it is determined that this function is disabled.

  If the function is disabled in the PE device, the NIF management unit 1000 determines whether the segment that received the packet is a root segment or a leaf segment (S420). This determination procedure searches the corresponding entry in the table shown in FIG. 7 using the reception port ID 701 of the received Hello packet as a key. Next, the corresponding entry is searched in the table shown in FIG. 15 using the group ID 702 in the found entry as a key. Subsequently, the entry shown in FIG. 16 having the same ID as the receiving port in the found entry is searched. If the leaf flag 1601 of the entry is 0, it is determined that the route is 1, and if it is 1, the leaf is determined.

  In the NIF management unit 1000, when the packet is received from the leaf segment when the function is disabled in the PE device, the packet with the TC flag turned off is transferred to the root segment (S460).

  When the packet is received from the root segment when the function is disabled on the PE device, or when the packet is received from the route segment when the function is enabled on the PE device The unit 1020 transfers the packet with the TC flag turned off to the other route segment and all the leaf segments (S450).

  When the packet is received from the leaf segment when the function is enabled in the PE device, the packet with the TC flag turned off is transferred to all the root segments and other leaf segments (S440). As a result, even when the root bridge of the spanning tree is arranged in the leaf segment of the E-TREE service, the STP can be operated normally.

  FIG. 12 is a network configuration diagram showing an embodiment in which the spanning tree reconfiguration method of the present invention is applied to a corporate LAN. Reference numerals 100, 110, and 120 denote subscriber LANs deployed on different floors, which are interconnected by switches 1 to 6 installed in the subscriber LAN to which the present invention is applied. Reference numerals 1 to 6 denote switches constituting the subscriber's LAN. Here, the STP is operated to detect and cancel the loop path. In this configuration, the broadcast domain 20 is shared by the subscriber's LANs 100, 110, and 120, and the effect of spanning tree reconstruction is localized to each subscriber's LAN 100, 110, or 120 according to the content of the configuration change. Make it possible to do.

  FIG. 13 is a network configuration showing an embodiment in which switches 20, 21, and 22 to which the present invention is applied are arranged on each floor and subscriber LANs 100, 110, and 120 are interconnected in the embodiment shown in FIG. FIG. FIG. 14 shows an internal structure example of a communication apparatus to which the present invention is applied. The communication device includes a node management unit 2000 that manages the entire device, a plurality of network interfaces 610-1 to 610-n, and a switch unit 2100 for interconnecting the network interfaces. Each network interface processes Ethernet frames received via the Ethernet transmission / reception ports 601-1 to 601-n, interfaces 602-1 to 602-n to the switch units corresponding to the Ethernet transmission / reception ports, and the Ethernet transmission / reception units. An input processing unit 1010, an output processing unit 1020 that performs transmission processing of an Ethernet frame via the Ethernet transmission / reception unit, and a network interface management unit 1000 that controls the input processing unit and the output processing unit. The node management unit 2000 provides a management interface to the administrator of the communication device to enable various settings / operations on the device. The node management unit holds a storage device and can hold initial setting information necessary for initializing the device. When initializing the device, the node management unit initializes each function in the device with the initial setting information stored in the storage device or the default value of the node management unit. Settings and operations performed by the administrator on the node management unit of the device are appropriately reflected in an appropriate NIF management unit among the network interfaces 610-1 to 610-n installed in the device.

1-4: Switches 10, 11, 12, 13 installed in the subscriber LAN: PE devices 20, 21, 22 of the communication carrier: Switches 50, 60 installed in the subscriber LAN to which the present invention is applied: STP Packet 100, 110, 120, 130: Subscriber LAN
500, 600: Network 600-1,. . . , 600-n: Ethernet transmission / reception ports 602-1,. . . 602-n: Interfaces 610-1,... To switch units interconnecting network interfaces. . . , 610-n: Network interface 1000: Network interface management unit 1020: Network interface Ethernet frame output processing unit 1010: Network interface Ethernet frame input processing unit 2000: Node management for managing the entire communication device Part
2100: Switch unit for interconnecting network interfaces

Claims (11)

A communication device connected to a plurality of communication devices via a network and transferring packets using a communication path established with the plurality of communication devices,
An input processing unit for receiving control packets from the plurality of communication devices;
Based on the information of the communication path included in the control packet received by the input processing unit, discarding the other control packet including information about the change of the communication path received by the input processing unit, or to the other control packet An NIF management unit that determines whether a response packet to be transmitted is generated and transmitted by the own device;
An output processing unit that discards the other control packet based on the determination of the NIF management unit, or creates and transmits a response packet to be transmitted to the other control packet. A communication device. The communication device according to claim 1,
The communication path is a communication path formed by a spanning tree protocol,
The NIF management unit extracts information of a root bridge of a spanning tree from the control packet. The communication device according to claim 2,
The NIF management unit compares the information of the control packet with the information of the root bridge held in the communication device, and a spanning tree configuration change event transmitted by the control packet received by the input processing unit is: A communication device that determines whether the configuration change event can be localized. The communication device according to claim 1,
The output processing unit, based on the information of the root bridge in the spanning tree which the NIF management unit holds, the response packet to the configuration change event of the spanning tree constructed by the own device, the NIF management unit designated A communication device that transmits from a port of the communication device. The communication device according to claim 1,
A communication apparatus, wherein the spanning tree configuration change event is an event by a Config BPDU and a TCN, or a Config BPDU. The communication device according to claim 1,
When the wide area Ethernet service provided by the communication device is an E-TREE service and the NIF management unit receives the control packet from the leaf segment, the NIF management unit changes the configuration of the spanning tree to the root segment and other leaf segments. A communication apparatus characterized by selecting an event transfer. The communication device according to claim 2,
The communication apparatus according to claim 1, wherein the NIF management unit updates the information of the root bridge held by the communication apparatus according to the information of the control packet received by the input processing unit. The communication device according to claim 2,
The root bridge information includes a root bridge ID and a path cost. The communication device according to claim 8,
The NIF management unit has a root bridge ID included in the received control packet larger than the root bridge ID held by the communication device and the ports other than the port that has received the information of the root bridge ID held by the communication device. When a control packet is received, a new control packet is configured with reference to the root bridge information held by the communication device,
The output processing unit transmits the new control packet to a port that has received the control packet. The communication device according to claim 8,
The NIF management unit
Received when the root bridge ID held by the communication device is smaller than the bridge ID of the control packet, or when the root bridge ID is equal and the path cost held by the communication device is less than or equal to the path cost included in the control packet A communication apparatus for discarding a control packet. The communication device according to claim 8,
The spanning tree configuration change event is an event by Config BPDU and TCN, or Config BPDU,
A communication apparatus, wherein when a TCN is received from a port that has received a Config BPDU from a root bridge held by the communication apparatus, the received TCN is discarded.

Images