WO2004086697A1 - Node device having links and method for allocating user band to links - Google Patents

Abstract

A node device connected to adjoining other node devices through links and even transmitting distributed packets to the links while securing the packet order. The node device holds a table where user identification information is related to one link for transmitting the packet of the user of the user identification information. The node device extracts the user identification information contained in the received packet and finds the link corresponding to the extracted user identification information from the table, and transmits the received packet to the found link. Since one link is allocated to each user, the packet order of each user is secured.

Description

 TECHNICAL FIELD The present invention relates to a node device having a plurality of links and a method of allocating a user bandwidth to the plurality of links.

 The present invention relates to a node device provided in a communication network, connected to a node device adjacent on the downstream side via a plurality of links, and transmitting a received packet to the node device adjacent on the downstream side. The present invention relates to a node device that can guarantee the order of packets even when packets are distributed to a link and transmitted. The present invention also relates to a method of allocating a user band to a plurality of links that can guarantee the order of buckets. Background art

 With the explosive spread of the Internet in recent years, the emergence of media such as light that handles large-capacity and high-quality information, expectations have been placed on developing a large-scale communication infrastructure that can handle large-capacity data flexibly. I have. As a method of accommodating expanding traffic in a communication infrastructure, there is a method of providing a plurality of links between transmission devices. For example, there is a method of increasing the number of 1 Gbps band links between transmission devices from one to three. This method is called link aggregation in the Ethernet, and is specified in IEEE 802.3.

 When multiple links are provided between transmission devices, the ability to distribute (assign) packets (frames) from the transmission device to any of the multiple links is the key to realizing a bandwidth guarantee service and effectively using links (bands). It becomes important from a viewpoint.

 In the IEEE 802.3 link aggregation, when multiple links are provided, (1) the order of the buckets for each flow is guaranteed, and (2) the buckets are not duplicated. An algorithm that specifies how to distribute and transmit packets to multiple links is not specified.

One way to distribute packets to multiple links is to switch the link cyclically each time a bucket is sent. However, in this method, the packet order is not guaranteed because the packet length is variable. For example, if the first packet of the same user is sent on the first link and then the second bucket is sent on the second link, and the first bucket is longer than the second bucket, The second packet transmitted later may reach the transmission device of the transmission destination first, and the order may be reversed. To reliably prevent such order inversion, the source transmission device identifies the attributes of the packet, such as the MAC address, input port, IP address, and port number, and identifies packets with the same attribute value as the same. By allocating to the link, the order of packets for each flow is guaranteed and checked. In addition, in order to distribute packets evenly to a plurality of links, there is a method of randomly allocating packets to links based on a hash value obtained by substituting the attribute of the packet into a hash function. Buckets belonging to the same flow (or user) have the same hash value and are therefore distributed to the same link. This guarantees the order of packets for each port. However, the Nosh value obtained from the attribute value of a packet may be biased depending on the attribute value. As a result, a large number of buckets are allocated to a specific link and are not always uniformly distributed. For this reason, there is a problem that the extended link band cannot be used effectively. This is fine for the Best Effort service, but the guaranteed bandwidth with the bandwidth guarantee service is limited to the bandwidth of one link no matter how many links are added, considering the worst case. Will be done.

 As a conventional technology for realizing the bandwidth guarantee service, there is a technology related to a network management system.However, it does not presuppose a plurality of links.Therefore, distribution of a bucket to a plurality of links between transmission devices is considered. No (see Patent Document 1, for example).

 Also, as a conventional bandwidth control device, a plurality of physical links are trunked to one logical link, and certain physical links among the logical links are converged so as to match traffic under specific conditions. There is a sub-logical link that separates the traffic (for example, see Patent Document 2).

 Patent Document 1

Japanese Patent Application Laid-Open No. H11-1-177617 (page 2) Patent Document 2

 Japanese Patent Application Laid-Open No. 2002-232324 (page 2) Disclosure of the Invention

 An object of the present invention is to guarantee the order of packets in a node having a plurality of links.

 A node device according to the present invention is a node device provided in a communication network, connected to a node device adjacent on the downstream side via a plurality of links, and transmitting a received bucket to a node device adjacent to the downstream side. A user information extracting unit for extracting user identification information indicating the user who transmitted the bucket, which is included in the received bucket; user identification information of the user who transmitted the bucket; Data for retaining assignment data in which user identification information is associated with identification information of one link assigned to a user, and for obtaining identification information of a link corresponding to the user identification information extracted by the user information extraction unit. A holding unit, and transmitting the received packet to a link corresponding to the link identification information obtained by the data holding unit. And a packet transmitting unit.

 According to the present invention, a packet transmitted from a certain user is distributed to a predetermined one of a plurality of links and transmitted. As a result, the order of the buckets of the user is prevented from being reversed, and the order of the buckets can be guaranteed.

 Preferably, one link assigned to the user of the user identification information is a link that guarantees a required bandwidth of the user. As a result, a bandwidth guarantee service is realized.

Further, the node device according to the present invention is provided in a communication network, is connected to a node device adjacent on the upstream side via a plurality of links, and further receives a packet received from the node adjacent on the upstream side further downstream. A plurality of reception buffers for temporarily storing packets received from each of the plurality of links, monitoring the plurality of reception buffers, and monitoring the upstream side. A control bucket indicating link switching, which is transmitted when a node device adjacent to the node switches a link for transmitting a bucket, is received by at least one of the plurality of reception buffers. In this case, until the control bucket is received in all of the plurality of reception buffers, reading of the bucket subsequently received by the reception buffer in which the control bucket is received is stopped, and the plurality of reception buffers are stopped. And a read control unit for resuming the reading of packets from the plurality of receiving buffers and transmitting the packets when the control bucket is received by all of them.

 The user bandwidth allocation method according to the present invention is executed by a network management device that manages a communication network in which a plurality of links are provided between adjacent node devices, and allocates a user bandwidth to the plurality of links. Selecting a link having an available bandwidth equal to or more than the required bandwidth of the user from the plurality of links; and if one link is selected by the selecting step, the method further comprises the steps of: The requested bandwidth of the user is allocated to one link, and when two or more links are selected, of the two or more links, the link with the largest free bandwidth, the link with the smallest free bandwidth, and the link with the highest usage rate Selecting a high link or a link with the lowest utilization, and allocating the required bandwidth of the user to the selected link. That.

 According to the present invention, the required bandwidth of the user is allocated to a predetermined one of the plurality of links. As a result, a packet of a certain user is transmitted through one assigned link, and the order of the packets of the certain user can be guaranteed.

Further, the present invention is a user bandwidth allocation method executed by a network management apparatus which manages a communication network in which a plurality of links are provided between adjacent node apparatuses, and which allocates a user bandwidth to the plurality of links. In the case where the bandwidth of one or more users has already been allocated to the plurality of links, when allocating the required bandwidth of a newly added user, the allocation of the bandwidth of the already allocated user is performed. And selecting the largest requested bandwidth from the requested bandwidths of the users whose allocation has not been determined among the requested bandwidths of the released users and the newly added users. Selecting a link having an available bandwidth equal to or greater than the selected maximum required bandwidth among the plurality of links. And if one link is selected by the selecting step, Allocating the maximum requested bandwidth to the one link, and selecting two or more links in the selecting step, among the two or more links, the link having the largest available bandwidth, the link having the largest available bandwidth. Selecting a small link, a link with the highest usage rate, or a link with the lowest usage rate, and allocating the maximum required bandwidth to the selected link.

 According to the present invention as well, the required bandwidth of the user is allocated to one of the plurality of links, so that the order of the buckets can be guaranteed. BRIEF DESCRIPTION OF THE FIGURES

 FIG. 1 shows an example of a communication network to which the present invention is applied.

 FIG. 2 is a flowchart showing the flow of the entire process of NMS.

 FIG. 3 is a block diagram showing a configuration of each switch of the carrier network. FIG. 4 is a flowchart showing the flow of the user bandwidth allocation process executed by the NMS.

 FIG. 5 shows an example of the capacity of the additional user bandwidth when four 1 Gbps links are provided.

 FIG. 6 shows an example of accommodating an additional user band when there is one 1 Gbps link and three 100 Mbps links.

 FIG. 7 is a flowchart showing the flow of a user bandwidth allocation process executed by the NMS.

 FIG. 8 shows an example of band allocation according to the flowchart of FIG.

 FIG. 9 is a block diagram showing the configuration of the switch.

 FIG. 10 is a block diagram showing the configuration of the switch.

 FIG. 11 is a flowchart showing a flow of a link assignment process performed by the NMS when a link failure occurs.

 FIG. 12 shows an example of band allocation according to the flowchart of FIG.

 FIG. 13 is a flowchart showing a flow of a process of allocating a band according to the priority.

FIG. 14 is a block diagram showing the configuration of the switch. FIG. 15 is a block diagram showing the configuration of the switch.

 FIG. 16 is a block diagram showing the configuration of the switch.

 FIG. 17 is a flowchart showing a processing port of the read control unit of the receiving switch.

 FIG. 18 is a block diagram showing the configuration of the switch.

 FIG. 19 is a block diagram showing the configuration of the switch. BEST MODE FOR CARRYING OUT THE INVENTION

 FIG. 1 shows an example of a communication network to which the present invention is applied, and shows a configuration example of a wide area LAN using a virtual LAN (V LAN: Virtual LAN). At a plurality of bases A to E of a user (for example, a company or the like) such as a head office, a branch office, a factory, etc., a LAN (not shown) for the user is provided. And these sites

When the LAN is connected to carrier network 1, one V

LAN 2 is formed.

 Although FIG. 1 illustrates only one user's VLAN, the carrier network 1 is also connected to another user's base LAN, and the other user's VLAN is formed. You can.

 The carrier network 1 has switches 11 to 17 as an example of a node device, and a plurality of links are provided in a part between the switches. For example, two 1 Gbps links are provided between switches 1 1 and 1 2 and switches 13 and 16, and three 1 Gbps links are provided between switches 1 1 and 1 3. Therefore, two links of 100 Mbps are provided between the switches 13 and 17. In the present embodiment, the present investigation is applied to the distribution (allocation) of packets on a plurality of links between the switches.

The buckets input to the carrier network 1 from each base are added with the VLAN tag (VLAN-TAG) assigned to each user at the switches 14 to 17 which are the entrance nodes. The packet to which the VLAN tag has been added is switched in the carrier network 1 based on the VLAN tag and the MAC (Media Access Control) address included in the header of the bucket, and the packet is sent to the destination L. Transferred to AN. At the switches 14 to 17, which are the exit nodes connected to the transfer destination base, the VLAN tag is removed, and the removed packet is transmitted to the target base LAN.

 In the present embodiment, a packet of a certain user is distributed to one selected link among a plurality of links between switches, and is not distributed to two or more links. This guarantees the packet order of the user. In order to sort such packets, a VAN tag is used in the present embodiment.

 The setting for allocating the bucket of each user to which link is performed by ten network management systems (NMS: Network Management System) that manage the carrier network 1. In a large-scale carrier network, the NMS may be hierarchized, and an EMS (Element Management System) may be provided below the upper NMS.In this case, the EMS may set the distribution. . Hereinafter, the case where the NMS 10 performs the operation will be described as an example.

 First, the entire process from setting of the user bandwidth by the NMS 10 to setting of distribution to each switch will be described, and then a specific method of distribution will be described. After that, when changing from one distribution to another distribution, a description will be given of a modification method for ensuring the order of packets.

 FIG. 2 is a flowchart showing the flow of the entire process of the NMS 1◦.

 When a certain user (here, a user of V LAN 2 in FIG. 1) subscribes to a wide area LAN service using a VLAN, the NMS 10 sets a user band for each base for the user (S 1). . For example, in FIG. 1, 100Mbps is set for sites A to C and E, and 1Gbps is set for site D.

 Subsequently, the NMS 10 calculates the bandwidth of each link bundle used by the user based on the network configuration of the carrier network 1 (S2). For example, in FIG. 1, 300 Mbps of the 1 Gbps × 2 band between the switches 111 and 112 is calculated. In addition, among the numbers shown adjacent to the link between the switches in FIG. 1, the denominator indicates the entire band between the switches, and the numerator indicates the band calculated for the user.

Next, NMS 10 checks the free bandwidth of each link in each link bundle, and allocates The link is determined (S3). This method includes several methods, which will be described in detail later. Finally, the NMS 10 creates a distribution table for each switch based on the determined distribution link, and sets it for each switch (S4). The setting of the distribution table for each switch is performed via a control line connecting the NMS 10 and each switch, which is indicated by a broken arrow in FIG.

 FIG. 3 is a block diagram showing a configuration of each switch of the carrier network 1. As shown in FIG. The switch has a distribution table holding unit 31, a user information extraction unit 32, and an output link distribution unit 33.

 The distribution table holding unit 31 stores the distribution table 31a set by the processing of the above-described step S4 of the NM S10. In the present embodiment, the distribution table 31a includes a VLAN tag as an example of user information (user identification information), an assigned link number (link identifier) to which a packet of a user having the VLAN tag is distributed, and a power. Become. In Figure 3, the packet of the user with VLAN tag = 1 has link 2 and the bucket of the user with VLAN tag = 2 or 3 has link 1, the packet of the user with VLAN tag = 4 has link 3, Each is assigned.

 The packet input to the switch is input to the user information extraction unit 32. The user information extraction unit 32 extracts VLAN tags as user information included in the bucket, distributes the extracted VLAN tags to the distribution tape holder 31 and outputs the input bucket to the output link distribution unit 33. Give to. The distribution table holding unit 31 obtains the allocated link number corresponding to the VLAN tag given from the user information extracting unit 32 from the distribution table 31a, and supplies the obtained allocated link number to the output link distribution unit 33. .

The output link distribution unit 33 outputs the packet supplied from the user information extraction unit 32 to the link corresponding to the assigned link number supplied from the distribution table holding unit 31. As a result, the bucket of the user with VLAN tag = 1 is output to link 2, the bucket of the user with VLAN tag = 2 or 3 is output to link 1, and the packet of the user with VLAN tag = 4 is linked. Output to 3. As a result, the order of packets of each user is guaranteed. Also, make sure that the sorting table is appropriate As a result, the bandwidth of each link can be used effectively.

 Here, how to distribute each user's packet to a plurality of links \ In other words, how to create a distribution table \ That is, there are several methods in step S3 above, The following describes these methods.

 First, a method of allocating a user bandwidth to a plurality of links provided between switches will be described. FIG. 4 is a flowchart showing the flow of the user bandwidth allocation process executed by the NMS 10.

 In the bandwidth guarantee service, when a user contracts / subscribes, the guaranteed bandwidth of the user is determined. The NMS 10 selects a link having an available bandwidth equal to or larger than the guaranteed bandwidth from among a plurality of links between switches existing on the route determined in the carrier network 1 (S11). For example, in FIG. 1, if the user's guaranteed bandwidth of 300 Mb ps is allocated to three links between switches 1 1 1 1 3, the three links (link bundles) have an available bandwidth of 300 Mb ps or more. One link is selected. As a result, it is possible to guarantee the bandwidth for each user, and it is possible to guarantee the packet order of the user.

 Note that if there is no link having an available bandwidth equal to or greater than the guaranteed bandwidth, the guaranteed bandwidth of the user cannot be secured, so the NMS 10 changes the route of the user, and changes the route of the other user according to the flowchart of FIG. Execute the process.

 If there is a link having an available bandwidth equal to or more than the guaranteed bandwidth of the user (additional user), the NMS 10 determines whether or not there are a plurality of such links (S12), and (Y in S12), (1) the link with the highest available bandwidth, (2) the link with the lowest available bandwidth, (3) the link with the lowest load (utilization rate), (4) the load (used The link is selected based on one of the selection criteria of the link having the highest rate, and the user is assigned to the selected link (S13). The selection criterion (1) to (4) is determined in advance by an administrator or the like. If there is only one link whose available bandwidth is equal to or greater than the guaranteed bandwidth of the user (Ν in S12), the selected link, that is, the one link, is finally selected and assigned to the user. .

According to the selection criterion (1), the link with the largest available bandwidth is selected, and the additional user bandwidth is selected. Is assigned. This makes it possible to equalize the load between multiple links. According to selection criterion (2), the additional user bandwidth is allocated to the link with the least available bandwidth. As a result, when a link having a large free band is left, and another user using a larger band joins later, the left link can be allocated to the other user.

 According to the selection criterion (3), the load on each link can be equalized similarly to the selection criterion (1). Here, the load means a value obtained by dividing the used bandwidth of the link by the entire bandwidth (= used bandwidth + free bandwidth) of the link. If the full bandwidth of each link is equal, the same link will be selected regardless of the selection criteria (3) or (1), but the full bandwidth of each link may be different. However, in this case, the selection result of both may be different. The selection criterion (4) is the same as the selection criterion (2). 1 If the total bandwidth of each link is equal, the selection result of the selection criterion (4) matches the selection result of the selection criterion (2). If the total bandwidth of the link is different, the selection results of both may differ.

 For example, Fig. 5 shows an example of accommodating additional user bandwidth when four 1 Gb ps links are provided, and Fig. 6 shows one 1 Gb ps link and three 100 Mb ps links. 5 shows an example of accommodating an additional user band in the case of (1). In FIG. 5, assuming that the follow-up user bandwidth is 30 OMb ps, when using the above selection criterion (1) or (3), the user bandwidth is assigned to link 2 in either case, and When (2) or (4) is used, the user band is allocated to link 3 in either method. In FIG. 6, assuming that the additional user bandwidth is 5 OMb ps, when the above selection criterion (1) is used, the user bandwidth is allocated to link 1, while when the above selection criterion (3) is used, the corresponding user bandwidth is Bandwidth is allocated to link 3. Further, when the above selection criterion (2) is used, the user band is allocated to link 2, and similarly when the above selection criterion (4) is used, the user band is allocated to link 2.

A sorting table is created based on the link selected in step S14 of FIG. As a result, the bandwidth of the link bundle can be used effectively while guaranteeing the bandwidth for each user. In a carrier network using MPLS (Multi-protocol Label Switching), a link can be assigned for each label path instead of a VLAN tag.

 The setting of the link to which the packet of each user is to be allocated may be performed by the node itself (for example, the distribution table holding unit 31 or a tape generation unit (not shown) provided in the node). In this case, the bandwidth of the link bundle is determined by NMS or EMS, or by signaling between nodes, and the node performs only the distribution judgment.

 Next, a description will be given of a method of re-assigning a user band already allocated to the link bundle when the additional user band is allocated to the link bundle. FIG. 7 is a flowchart showing the processing flow of NMS 10 according to this method. When newly allocating the guaranteed band of the additional user to the link bundle, the NMS 10 first clears all the settings of the link bundle (S 21). Thereby, the band allocation of the user already set in the link bundle is also cleared. Subsequently, the NMS 10 selects a user with the maximum guaranteed bandwidth from among unset users (including both users who have already been assigned and additional users) (S22), The link is selected based on the additional user bandwidth allocation flow of FIG. 4 described above, and the bandwidth is set (S23). Such processing is performed for all users who use the link bundle (S24). When there are two or more links selected according to the selection criterion in step S13 in FIG. 4, one of the links is selected based on the criterion such as the link with the smallest number / the link with the largest number.

 As a result, the links to be allocated in order from the user with the largest used bandwidth are determined. Since the bandwidth of the already set user is also reset, even if the bandwidth of the additional user cannot be allocated to any of the links in the link bundle before resetting, it is allocated after resetting. May be possible. As a result, the link band can be effectively used.

FIG. 8 shows an example of band allocation according to the flowchart of FIG. As shown in the upper part of Fig. 8, in the link bundle consisting of two links 1 and 2 of 1 Gbps, link 1 has already been assigned the bandwidth of 500 Mbps for user 1, and link 2 has Is assumed that the bandwidth of 300 Mbps for user 2 has already been allocated. In this situation, let us consider a case where the additional bandwidth 800 Mbps of the new user 3 is allocated. In this case, the bandwidth of user 3 cannot be allocated to either link 1 or link 2. On the other hand, when the reconfiguration is performed according to the flowchart of FIG. 7, as shown in the lower part of FIG. 8, the bandwidth of 800 Mbps of user 3 is allocated to link 1 and the bandwidth of user 1 is allocated to link 2 Mbs and a bandwidth of 300 Mbs for user 2 are respectively allocated, and the bandwidths of all users can be allocated within two links.

 Note that, even with this method, for users who cannot perform link assignment due to lack of available free bandwidth, the link rate is changed, and link assignment is performed on the changed route.

 Next, the link allocation method (allocation method) when a failure occurs in one link of the link bundle will be described. FIG. 9 is a block diagram showing a configuration of a transmitting switch and a receiving switch for explaining a link distribution method when a link failure occurs.

 One way to guarantee bandwidth in the event of a failure is to provide a backup link that is not normally used. In Figure 9, of the three links, links 1 and 2 are the working links, and link 3 is the backup link used in the event of a failure. For example, when a failure occurs in the working link 2, the failure detection unit 41 provided in the receiving switch (downstream switch) 4 detects the failure. The failure detection unit 41 notifies the transmission side switch (upstream adjacent switch) 3 of the failure. This failure notification can be made via an uplink (not shown) from the receiving side switch 4 to the transmitting side switch 3, or can be made via the NMS 10 and the control line. The failure notification includes link identification information (for example, a link number) indicating the link in which the failure has occurred.

The transmission switch 3 is provided with a backup conversion table holding unit 34. The backup conversion table holding unit 34 holds a backup conversion table 34a used at the time of failure. As shown in Figure 9, this backup conversion table 34a contains the links that are normally used (when there is no failure). (Set by NMS 10), but changed to avoid the failed link when a failure occurs.

 The backup conversion table 34a is a table showing the links actually used in preparation for a failure, and shows the correspondence between the allocated links and the actually used links. In Fig. 9, packets that were distributed to link 1 before the failure occurred are distributed to the same link 1 after the failure occurs, and packets that were distributed to link 2 before the failure occured to link 3 after the failure occurred. It is shown that they can be sorted. If the failed link is link 1, the actually used link of assigned link 1 is set to link 3;

 Upon receiving the failure notification (including the failure link identification information) transmitted from the failure detection unit 41 of the receiving switch, the backup conversion table holding unit 34 changes the backup conversion table and changes the backup conversion table. The output link distribution unit 33 is instructed to change the bucket output link distribution based on the table. As a result, the output link distribution unit 33 distributes the packet to be distributed to the failed link 2 to the backup link 3. By doing so, it is possible to respond even when a failure occurs.

 The distribution table may be directly changed without using the backup conversion table. In this case, the table needs to be changed only for the user using the failed link. Next, a method of changing a link when a failure occurs when no backup link is provided is described. FIG. 10 is a block diagram of a transmitting switch and a receiving switch for explaining a link assignment method when a link failure occurs. FIG. 11 is a flowchart showing the flow of link assignment processing when a link failure occurs by the NMS 10.

 It is assumed that, of the three working links 1-3 between the transmitting switch 3 and the receiving switch 4, link 2 has failed. In this case, the failure detection unit 41 of the reception-side switch 4 transmits a failure notification including link identification information indicating a failure link to the transmission-side switch 3 and the NMS 10. When a failure notification is given to the sending switch 3 via the NMS 10, the notification is transmitted only to the NMS 10.

The NMS 10 receives the failure notification and, by receiving the notification of the failure, swings the switch 3 on the transmission side. Change the sorting table. Here, the assignment (distribution) of only the users who have been assigned to the failed link is reviewed, and the assignments of the users who have already been assigned to the links that have not failed are maintained as they are.

 First, the NMS 10 clears the setting of the failed link (S31). In the example of Fig. 10, the setting of link 2 is cleared. Subsequently, the NMS 10 selects a user with the maximum guaranteed bandwidth from among the users related to the cleared settings (ie, unset users) (S32). Next, the NMS 10 determines whether it is possible to allocate the bandwidth of the selected user among the non-failed links (links 1 and 2 in FIG. 10). Then, it is determined whether there is a link having an available bandwidth equal to or greater than the guaranteed bandwidth of the selected user (S33).

 If there are a plurality of such links (Y in S33), the NMS 10 selects and selects a link according to any of the selection criteria (1) to (4) in step S14 in FIG. The user's bandwidth is allocated to the selected link (S34). If there is one such link (Y in S33), NMS 10 allocates the user's bandwidth to the one link. In these cases, the bandwidth of the user can be guaranteed.

 On the other hand, if such a link does not exist (N in S33), the NMS 10 selects a link by the allocation process when the band is over and performs the setting (S35). Specifically, (a) a method of selecting links so that the load of each link becomes equal or close to equal, and (b) a method of distributing so that the excess bandwidth becomes equal or close to equal. There is a way. The links selected by both methods are the same if the total bandwidth of each link constituting the link bundle is equal. On the other hand, when the bandwidth of each link is different, (a) the method of equalizing the load or making the load close to the equal is preferable because the characteristics become equal. In this case, if all users assigned to the link (both newly assigned users after the occurrence of the failure and users assigned before the failure) have no particular priority, Does not guarantee the bandwidth of all users, and is a best-effort service.

By such an assignment change process, for example, the distribution table 31a in FIG. W 200

Is changed to the sorting tape hole 31b. The table 31b after this change is set in the distribution tape holder 31 of the transmission switch 3 by the NMS 10, and the output link distribution unit 33 of the transmission switch outputs the output link according to the table 31b. Select and output the bucket. As a result, even when a failure occurs, the order of packets is guaranteed, and packets can be sorted using the link bandwidth effectively.

 Also, a table before and after the change (only one link failure), such as table 31b ', can be prepared in advance, and the destination of the failure can be determined in advance and set. As a result, the link can be changed at high speed only when one link fails.

 FIG. 12 shows an example of band allocation according to the flowchart of FIG. Left 俱 The three links 1 to 3 of IJ show the assignment before the failure occurs, and the three links 1 to 3 on the right show the assignment after the failure of link 2. In this example, when a failure occurs in link 2, distribution is performed so that the usage load is as equal as possible, that is, the influence of the failure is equal for each user. First, in link 2, the bandwidth of 40 OMbps having the largest user bandwidth is allocated to link 3 having the largest available bandwidth (lowest load). The next largest bandwidth of 30 O M bps The bandwidth is allocated to link 1 with the largest available bandwidth (the least used load). Now that there is no more free bandwidth on both links 1 and 3, the last remaining 20 OM bps bandwidth is allocated to link 3 which has the least bandwidth overload (lowest usage load).

 Next, in the case where a backup link at the time of failure is not provided, a method of resetting the assignment to all users at the time of the occurrence of the failure, including the user who has already been assigned to the link where the failure has not occurred. explain.

When the priority is not set between the users, the bandwidth of each user can be allocated by the above-described flowchart shown in FIG. In this case, for the user who cannot secure the bandwidth, the link is selected and allocated by the processing of step S35 in FIG. 11 described above to provide the best-effort service or to change the route. To allocate bandwidth on the link on the changed route it can.

 When the priority is set between the users, the bandwidth can be allocated first to the high priority, the user, and then to the low priority user. FIG. 13 is a flowchart showing a flow of a process of allocating a band according to the priority. Here, it is assumed that the user has set one of two priorities, “high” and “low”.

 The NMS 10 clears the settings of all links including the failed link (S41). Next, the NMS 10 selects the user with the highest guaranteed bandwidth among the users whose priority is “high” and has not been set (S42), and executes the processing of FIG. The link is selected and set according to the user bandwidth allocation flow (S43). The NMS 10 allocates all the users of the "high" priority (S42 to S44), and, when the allocation is completed, subsequently, similarly allocates the bandwidth of the "low" user. (S45 to S47).

 If such allocations secure the bandwidth of all users with priority "high" and "low" on a link where no failure occurs, provide bandwidth-guaranteed service to all users can do.

 If the bandwidth can be secured for all the users with the priority “high”, but the bandwidth cannot be secured for all the users with the priority “low”, the user of the priority “low” can perform step S46. Instead of the processing, allocation can be performed by the allocation flow when the bandwidth is exceeded in step 35 of FIG. 11. In this case, the service for the user with the priority “high” is the guaranteed bandwidth service, and the service for the user with the priority “low” is the best-ef-auto service. In addition, for a user with a low priority, the route may be changed and the bandwidth may be allocated.

 If the bandwidth of both “high” and “low” users cannot be secured, allocation can be performed by the allocation flow when the bandwidth is over, and in this case, best effort It can be.

Next, the timing of the change when changing the link assignment will be described. FIG. 14 is a block diagram showing a configuration of a switch having a timer for performing assignment change control. The transmission-side switch has a timer 35, a selector 36, and transmission buffers 37a to 37c for links 1 to 3, in addition to the distribution table 31, the user information extraction unit 32, and the output link distribution unit 33. Can be

 If you change the link assignment, the packet order may be reversed before and after the change. In order to prevent this, a predetermined time corresponding to each of the links is calculated / set by the NMS 10 in the timer 35, and the link switching timing is determined. The predetermined time of each link set in the timer 35 is obtained by the following equation.

Setting time = (maximum packet length) ÷ (speed of link before change) + | (delay time of link before change) one (delay time of link after change) | + _α … h)

 Here, h (> 0) is a value to make the set time slightly larger than the theoretical value in consideration of link speed and delay time errors.

 For example, when changing the bucket distribution from link 1 to link 2, the maximum bucket length (number of bits) is Lma X, the link speed (bps) of link 1 is B1, the delay time of link 1 is 1, and the link time is 1. When the delay time of 2 is tapped at 2, the set time of link 1 is set to the timer 35 as Tl = LmaxmaBl + | T1−T2 | + α. The values of Lmax, B 1, T 1 and τ 2 can be obtained in advance. Upon receiving the packet, the user information extraction unit 32 distributes the VLAN tag included in the received packet and provides the VLAN tag to the table holding unit 31.

 The distribution table holding unit 31 stores a distribution table 31c in which a link before change and a link after change of each user (V LAN tag) are defined. The distribution table holding unit 31 provides both the pre-change link number and the post-change link number corresponding to the VLAN tag received from the user information extraction unit 32 to the selector 36, and also supplies the pre-change link number to the timer 35. For example, when the VLAN tag = 1, the link numbers 1 and 2 are given to the selector 36, and the link number 1 is given to the timer 35.

The selector 36 selects the pre-change link number and outputs it to the output link distribution unit 33 when the switching signal from the timer 35 is not input (initial state). As a result, in the initial state, the packet is transmitted to the link before the change. The timer 35 receives a link change instruction from NMS 10. After receiving the change instruction, the timer 35 starts measuring the set time corresponding to the pre-change link number immediately after receiving the pre-change link number from the distribution table holding unit 31. Then, when the timer 35 does not receive the same pre-change link number from the distribution table holding unit 31 until this time expires (that is, when the next packet allocated to the same link is not received). ), The switching signal of the link number before change is output to the selector 36 when the time expires. On the other hand, if the same pre-change link number is received before the expiration of this time (that is, if the next bucket allocated to the same link is received), the timer 35 resets the clock to 0 Repeat the timing from a new time. For example, when the link number 1 is given to the timer 35, the timer 35 starts counting the set time T1, and if the link number 1 is not received until the time T1 expires. Outputs the switch signal of link 1 to the selector 36 when the time T 1 expires, and resets the clock when the link number 1 is received, and repeats the clock from 0 again.

 Upon receiving the switching signal, the selector 36 selects the post-change link number for the link corresponding to the switching signal, and outputs the selected link number to the output link distribution unit 33. As a result, the packet is transmitted to the link having the changed link number. In this way, the timer 35 measures the time required for the transmission of the maximum packet length L max by taking into account the delay time difference between the links, and during that time, switches the link if no packet is received. As a result, the order of packets on the same link can be prevented from being reversed, and the order can be guaranteed.

 After outputting the switching signal for a certain link, the timer 35 does not time the time and output the switching signal, even if the same link number before change is received. This prevents link re-switching.

 The above-described switching by the timer is performed for each link. Next, a method of setting a timer time for each user (that is, a VLAN tag) and performing link switching control for each user will be described.

FIG. 15 is a block diagram showing a configuration of a switch when link switching control is performed for each user. The transmission-side switch is provided with a timer 35a and transmission buffers 37a to 37c for links 1 to 3 in addition to the distribution table 31, the user information extraction unit 32, and the output link distribution unit 33. .

 A predetermined time corresponding to each of the VLAN tags (users) is set in the timer 35 by the NMS 10, and the timing of link switching is determined. The predetermined time of each V LAN tag set in the timer 35 is the same as that of the above equation (1), except that the predetermined time is provided for each V LAN tag. '

 For example, when changing the bucket with VLAN tag = 1 from link 1 to link 2, the maximum packet length (number of bits) is Lmax, the link speed (bs) of link 1 is Bl, and the delay time of link 1 is 1. Assuming that the delay time of link 2 is τ 2, the timer 35 has a setting time T l = Lmax x B 1+ |

 Upon receiving the packet, the user information extraction unit 32 gives the V RAN tag included in the received packet to the distribution table holding unit 31 and the timer 35.

 The distribution table holding unit 31 stores a distribution table 31c in which a link before change and a link after change of each user (each VLAN tag) are defined. When no switching signal is received from the timer 35 (initial state), the sorting table holding unit 31 outputs the pre-change link number corresponding to the VLAN tag received from the user information extracting unit 32 to the output link sorting unit. Give to 33. For example, if the \ tag = 1, the pre-change link number 1 is given to the output link sorting unit 33. As a result, the output link distribution unit 33 outputs the received packet to the link 1 before the change via the buffer 37a.

A link switching instruction is input from the NMS 10 to the timer 35a. After receiving the switching instruction, the timer 35a starts measuring the set time corresponding to all VLAN tags. Then, at the time of receiving the packet, the timer 35a checks whether or not the timer corresponding to the VLAN tag received from the user information extracting unit 32 has expired. If the timer has expired, a table of the corresponding VLAN tag is determined. Perform a switch. On the other hand, if the time has not expired, the timer 35a resets the clock, and repeats the clock from 0 again. For example, timer 35a receives VLAN tag = 1. If the relevant timer value has expired, a switching signal with VLAN tag = 1 is output to the distribution table, while if the relevant timer value has not expired, the timer is reset. Repeat the timing from 0, a new time.

 In this method, the completion of switching may be delayed depending on the traffic because the switching depends on the received packet. Therefore, by checking the timer value sequentially for the user (VLAN tag) that has not been switched in parallel and switching the table whose timer value has expired, the completion of switching can be accelerated.

 Upon receiving the switching signal, the distribution table holding unit 31 changes the link of the VLAN tag corresponding to the switching signal from the link before the change to the link after the change, and thereafter, from the user information extracting unit 32 to the VLAN tag. Is received, the change link number corresponding to the VLAN tag is output to the output link distribution unit 33.

 In this way, the timer 35a measures the time required for the transmission of the maximum packet length L max by taking into account the delay time difference between the links, and during that time, switches the link if no packet is received. As a result, the order of packets of the same user (the same VLAN tag) can be prevented from being reversed, and the order can be guaranteed.

 After the timer 35a outputs the switching signal, the timer 35a does not output the clocking of the time and the output of the switching signal even if the same VLAN tag is received. This prevents link re-switching.

 Next, switching control by a control packet for notifying the receiving switch of a link change (hereinafter referred to as a “change notification packet”) will be described.

 FIG. 16 is a block diagram showing a configuration of a switch for performing switching control by a change notification packet. FIG. 17 is a flowchart showing a processing flow of the read control unit 42 of the receiving switch.

 The transmission side switch is provided with a control bucket input section 38. The NMS 10 gives a link switching instruction to the control bucket insertion unit 38 and the distribution table 31.

The sorting table holding unit 31 holds the same table as the sorting table 31c shown in FIG. 15 and outputs the link number changed from the link before the change to the link after the change according to the link switching instruction. Give to 3. As a result, Packets (data packets and user packets) are distributed to the link after switching.

 On the other hand, the control packet insertion unit 38 writes a change notification bucket (control bucket) into each of the buffers 37a to 37c in response to the link switching instruction from the NMS 10. The change notification packet is written so as to be inserted between a data packet (user bucket) written before switching and a data bucket (user bucket) written after switching. That is, upon receiving the switching instruction, a change notification packet is inserted after the last data packet that has already been distributed or is being distributed. The written change notification packet is transmitted to the receiving switch via links 1 to 3 in the same manner as the data packet.

 The receiving switches 43 a to 43 c for temporarily recording the packets respectively received via the links 1 to 3, the read control unit 42, and the buffer 4 A multiplexing unit 44 for multiplexing 3a to 43c and transmitting the multiplexed data to one output link is provided.

 Referring to FIG. 17, when the packets are received by the buffers 43a to 43c, the read control unit 42 determines whether the received packet is a change notification packet (control packet) (S 5 1). In the case of a control bucket (Y in S51), the read control unit 42 removes the control bucket from the buffer (S58), sets the switching flag provided in the read control unit 42 to ON, In addition to indicating that the link is being switched (S59), the control packet reception flag of the link that has received the change notification packet among the control bucket reception flags provided for each link in the read control unit 42 Is set to ON to indicate that the change notification packet has been received on the link (S60). After that, the read control unit 42 enters a reception waiting state for the next packet.

 On the other hand, if the received bucket is not a change notification bucket, that is, if it is a data bucket (N in S51), the read control unit 42 determines whether the link is currently being switched based on the switching flag. (S52).

If the switching flag is not ON (that is, the link is not being switched) (N in S52), the read control unit 42 reads the received data bucket from the buffer and sends the received data bucket through the multiplexing unit 44. And send it (S55). On the other hand, the switching flag In the case of す な わ ち N (that is, during link switching) (N in S52), the read control unit 42 determines whether the control packet reception flags of all the links (that is, all of the links 1 to 3) are ON. Is determined (S53). That is, it is determined whether or not the change notification bucket 1 has already been received for all the links.

 If the control bucket reception flags of all links are not ON (N in S53), the read control unit 42 determines whether the control packet reception flag of the link that received the data packet is 〇N (S56). When the control bucket reception flag is ON (Y in S56), since the change notification packet has already been received on the link, the read control unit 42 reads the received data bucket from the buffer. Is temporarily stopped (S57). On the other hand, when the control packet reception flag is OFF (N in S56), since the change notification bucket has not been received for that link, the read control unit 42 reads the received data packet from the buffer. The signal is transmitted via the multiplexing section 44 (S55).

 In step S53, if the control bucket reception flag of all links is ON (Y in S54), link switching for all links has been completed, and the received data bucket is immediately transmitted. However, the problem of the order reversal does not occur. Therefore, the read control unit 42 returns both the switching flag and the control packet reception flags of all the links to OFF (S554), and performs read / transmission processing of the received data packet (S555). As a result, the data packet whose transmission has been temporarily stopped in step S57 is also transmitted. Thereafter, the received data packet is read from the buffer and transmitted as usual.

As a result, when a change notification packet is received for a certain link, reading and transmission of data buckets of all links after the change notification bucket are suspended until a change notification bucket is received for other links. . Then, after the change notification bucket is received on all the links, the reading / transmission of the data bucket is restarted. Since the change notification packet is a control packet indicating the boundary between before and after the switching at the transmitting switch, by temporarily stopping transmission at the receiving switch after receiving the change notification packet, It is possible to guarantee the order of data packets before and after the allocation change of all links. Instead of the change notification packet, a flag or data indicating the change notification is written in the header of the data packet at the transmitting switch and transmitted to the receiving switch, and the read control unit 42 of the receiving switch sets the flag included in the header. Alternatively, the same processing as described above can be performed depending on data.

 In the above description, the case where the control bucket is applied to each link has been described, but the control packet may be applied to each user.

 Next, a description will be given of a method of realizing the order guarantee by stopping the transmission of the bucket for a predetermined time before and after the change of the link assignment in the transmitting switch.

 FIG. 18 is a block diagram showing the configuration of a switch that realizes order assurance by stopping transmission of packets at the transmitting switch. The transmission-side switch is provided with a read control unit 39.

 The switching instruction is input to the distribution table holding unit 31 from NMS 10. When a switching instruction is input from the NMS 10, the distribution table holding unit 31 reads a pause signal for temporarily stopping reading / transmission of packets from the buffer 37 a to 37 c and reads the pause signal. Give 3 to 9.

 When the read control unit 39 receives the pause signal from the distribution table holding unit 31, the buckets that have already been distributed by the output link distribution unit 33 or the buckets currently being distributed are stored in the buffers 37 a to 37 c. After the packet is sent from the buffer 37a to 37c, the packet transmission is stopped for a certain period of time. The stop time is a time at which the transmission of the bucket which has been allocated or is being distributed is completed, and is preferably a transmission completion time based on the maximum packet length (time obtained by the above equation (1)).

 After the reading is stopped by the temporary stop signal, the distribution table holding unit 31 switches from the link before the change to the post-change link in the distribution table 31c (see FIG. 15).

As described above, a predetermined time is provided between the data packet transmitted before the switching and the data packet transmitted after the switching, so that it is similar to the case where the reading of the receiving bucket is temporarily stopped by the receiving switch. Avoids bucket reversal Is done.

 The stop time by this method needs to be longer than the transmission time of the change notification packet (control packet), but the effect is small because the time difference is at most about one bucket. In addition, this method has significant advantages because it does not require special packets such as change notification packets and does not require special processing on the receiving side. In this case, the reading of the buffers 37a to 37c for each link is stopped.A common buffer is arranged in front of the force buffers 37a to 37c, and the reading of the common buffer is performed. May be stopped.

 Next, a method of realizing order guarantee by changing the bandwidth of the user corresponding to each link when the buffer of the link becomes empty will be described. FIG. 19 is a block diagram showing a configuration of a switch for changing a link by this method. The transmission-side switch is provided with a buffer management unit 40 that manages the buffers 37a to 37c of the links 1 to 3.

 The user information extraction unit 32 gives the V RAN tag of the received packet to the distribution table 31. The sorting table holding unit 31 holds the same sorting table 31 c as described above. The distribution table holding unit 31 gives the pre-change link number and the post-change link number corresponding to the VLAN tag from the user information extraction unit 32 to the selector 36, and the pre-change link number to the buffer management unit 40. give. The selector 36 is set so as to select the link number before the change and supply it to the output link distribution unit 33 when the switching signal from the buffer management unit 40 is not input (initial state). As a result, the output link distribution unit 33 transmits the reception packet to the link before the change via the buffer corresponding to the link before the change. ·

The switch instruction is input to the buffer management unit 40 from NMS 10. After receiving the switching instruction, the buffer management unit 40, upon receiving the pre-change link number from the distribution table holding unit 31, checks whether the buffer corresponding to the received pre-change link number is empty. For example, when the pre-change link number = 1, the buffer management unit 40 checks the status of the buffer 37a corresponding to the link 1. When the buffer is empty, the buffer management unit 40 gives a switching signal of the selector 36. I can. As a result, the selector 36 selects the post-change link and supplies it to the output link distribution unit 33. As a result, the reception buffer is transmitted to the link after the change. On the other hand, when the buffer is not empty, the buffer management unit 40 does not supply the selector 36 with the switching signal.

 The output link is switched after the buffer is empty, that is, after all the data constituting the previously transmitted packet has been transmitted, so that the output packet is switched between the previously transmitted packet and the packet transmitted after the link switching. However, the problem of the order reversal does not occur. This method has the advantage that the circuit can be easily realized because timer control is not required.

 In the embodiment described so far, the user who has subscribed to the wide area LAN service cancels the service contract or the service expiration date expires, and the user of the NMS 10 and each switch in the respective switches is switched. When the data is no longer needed, delete the data of the user from the data of NMS 10 and the distribution table of each switch, and remove the free bandwidth of the link managed by NMS 10 and each switch. It can be dealt with by increasing.

 The link may be a two-way link or a one-way link. Also, a case has been described where the node device of the carrier network 1 is a switch, but the node device may be a router. Industrial potential

 INDUSTRIAL APPLICABILITY The present invention is applicable to a node device (for example, a switch, a router, etc.) having a plurality of links, which is a node device constituting a communication network. Further, the present invention can be used for a network management system that manages a communication network.

Claims

The scope of the claims

1. A node device provided in a communication network, connected to a node device adjacent on the downstream side via a plurality of links, and transmitting a received bucket to the node device adjacent on the downstream side,

 A user information extraction unit for extracting user identification information included in the received bucket and indicating a user who transmitted the bucket;

 Holding assignment data in which user identification information of a packet transmission source user and identification information of one link assigned to the user of the user identification information among the plurality of links are stored; A data holding unit for obtaining identification information of a link corresponding to the user identification information extracted by the unit;

 A packet transmitting unit that transmits the received bucket to a link corresponding to the link identification information obtained by the data holding unit;

 A node device comprising:

2. In Claim 1,

 The node device, wherein one link assigned to the user of the user identification information is a link that guarantees a required bandwidth of the user.

3. In Claim 1,

 The node device, wherein the assignment data is generated by a network management device that manages the communication network, and is transmitted to and held by the data holding unit.

4. In Claim 1,

 The node device, wherein the allocation data is generated by the data holding unit or is generated by a data generation unit.

5. In Claim 1,

The allocation data indicates that the required bandwidth of the new user When the bandwidth is assigned to any of the links having the free bandwidth equal to or greater than the bandwidth required by the new user, the link is generated by allocating the bandwidth required by the new user to the link having the largest or the smallest bandwidth. Node device. In claim 1,

 The allocation data, when a required bandwidth of a new user is allocated to any one of the plurality of links, among the links having an available bandwidth equal to or more than the required bandwidth of the new user, is the lowest usage rate. A node device generated by allocating a required bandwidth of the new user to a link or the highest link. In claim 1,

 The allocation data may be such that, when a required bandwidth of a new user is allocated to any of the plurality of links, the required bandwidths of all users including the already allocated users and the new user are the plurality of links. No. 7 of Claim 7, which is generated by newly re-assigning the link to

 The node device, wherein the allocation data is generated by sequentially re-allocating, from among the requested bandwidths of all the users, a user having a larger requested bandwidth to a user having a smaller requested bandwidth. In claim 8,

 The allocation data includes a link having the largest available bandwidth or a link having the smallest available bandwidth among links having an available bandwidth equal to or more than the required bandwidth of the user selected in order from the user having the large required bandwidth to the user having the small required bandwidth. A node device generated by allocating a required bandwidth of the selected user. 0. In Claim 8,

The allocation data is transmitted from the user of the large requested band to the small requested band. A link is generated by allocating the required bandwidth of the selected user to the link with the lowest usage rate or the link with the highest usage rate among the links having the free bandwidth equal to or more than the required bandwidth of the user selected in order for the user. Node device. 1 1. In Claim 1,

 The data holding unit includes: user identification information; identification information of a first link assigned to a user of the user identification information at a normal time when a failure has not occurred in the plurality of links; and at least one of the plurality of links. In the event of a failure, allocation data in which the user identification information is associated with the identification information of the second link allocated to the user at the time of the failure is held, and in the normal state, the data is extracted by the user information extraction unit. Determining identification information of a first link corresponding to the user identification information; determining, at the time of the failure, identification information of a second link corresponding to the user identification information extracted by the user information extraction unit;

 The bucket transmitting unit transmits the received packet to a link corresponding to the identification information of the first link or the second link obtained by the data holding unit,

 Node device.

1 2. In claim 1,

 The data holding unit is configured to, when a failure occurs in at least one of the plurality of links, cause the user identification information of the assignment data to correspond to a link in which no failure has occurred. Rewriting all or part of the link identification information, and in the case of the failure, obtains link identification information corresponding to the user identification information extracted by the user information extraction unit based on the rewritten allocation data.

 No

3. In Claims 11 or 12,

At least one of the links is a backup used only in the event of a failure. The remaining links are the working links normally used, the first link is one of the working links, and the second link is one of the back-up links or the working link. Node device. 1 4. In claims 11 or 12,

 All of the plurality of links are working links that are normally used, the first link is one of working links, and the second link is a working link that has not failed at the time of failure. One of the links, a node device. 15. In claim 14,

 The second link of the allocation data is, in the event of a failure, one of the working links in which no failure has occurred and a link having an available bandwidth equal to or greater than the required bandwidth or a link having an available bandwidth equal to or greater than the required bandwidth does not exist. A node device that is one of the links selected so that the load on the working link that has not failed is evenly approached.

6. In claim 15,

 The second link of the allocation data is, in the event of a failure, one of the working links in which no failure has occurred and a link having an available bandwidth equal to or greater than the required bandwidth or a link having an available bandwidth equal to or greater than the required bandwidth does not exist. Is a node device that is a link selected so that the excess bandwidth of the working link that has not failed is approached evenly.

1 7. In claim 15,

In the second link of the allocation data, in the event of a failure, the requested bandwidth of a high-priority user is reallocated first to the working link in which a failure has not occurred, and the required bandwidth of a low-priority user is allocated later. Node determined by remediation

8. In Claim 1,

 The data holding unit stores the user identification information of the user who transmitted the packet, and the identification information of the first link and the identification information of the second link assigned to the user of the user identification information among the plurality of links. Holding the associated allocation data, obtaining identification information of first and second links corresponding to the user identification information, wherein the packet transmitting unit is a network management device that manages the communication network or the downstream device; The packet received before receiving the link switching command from the node device adjacent to the side is transmitted to the first link determined by the data holding unit, and when the switching command is received, the packet is received before the switching command is received. And determining whether the bucket to be allocated to the first link is received for a predetermined time. If not during the reception, the received packet is thereafter, transmitted to the second link determined by the data holding unit,

 Node device.

9. In claim 18,

 After receiving the switching command, if the packet of the user is not received on the first link for a predetermined time after receiving the switching command, the packet transmitting unit requests the data holding unit for a bucket received thereafter. The process of transmitting to the given second link is performed for each user,

 Node device. 0 · In Claim 1,

The data holding unit stores the user identification information of the user who transmitted the packet, and the identification information of the first link and the identification information of the second link assigned to the user of the user identification information among the plurality of links. Holding the associated allocation data, obtaining identification information of first and second links corresponding to the user identification information, wherein the bucket transmitting unit is adjacent to the network management device that manages the communication network or the downstream side Receives a link switching command from the node device The packet received before the transmission is transmitted to the first link obtained by the data holding unit, and upon receiving the switching command, a control packet indicating link switching is transmitted to the plurality of links. Transmitting a bucket to the second link determined by the data holding unit,

 Node device. 1. In Claim 1,

 The data holding unit stores the user identification information of the user who transmitted the packet, and the identification information of the first link and the identification information of the second link assigned to the user of the user identification information among the plurality of links. Holding the associated allocation data, obtaining identification information of first and second links corresponding to the user identification information, wherein the bucket transmitting unit is adjacent to the network management device that manages the communication network or the downstream side Regarding the bucket received before receiving the link switching command from the node device, the received bucket is transmitted to the first link obtained by the data holding unit, and when the switching command is received, the plurality of links are transmitted. The transmission of the bucket is stopped for a predetermined time for each, and thereafter, the data holding unit Starts to transmit the packets to the second link is because,

 Node device. 2. In Claim 1,

The data holding unit stores the user identification information of the user who transmitted the packet, and the identification information of the first link and the identification information of the second link assigned to the user of the user identification information among the plurality of links. Holding the assigned data associated with each other, obtaining identification information of first and second links corresponding to the user identification information, the data transmission unit has a transmission buffer for each of the plurality of links, and the communication For a bucket received before receiving a link switching command from a network management device that manages the network or a node device adjacent to the downstream side, the packet is received by the first link obtained by the data holding unit. After transmitting the received packet and receiving the switching command, it monitors whether or not each of the transmissions has become empty, and for the packets allocated to the first link corresponding to the empty transmission buffer, Switch to two links, send buckets,

 Node device. 3. The node device provided on the communication network, connected to the node device adjacent on the upstream side via a plurality of links, and receiving the bucket received from the node adjacent on the upstream side further on the downstream side. A node device for transmitting to

 A plurality of receive buffers for temporarily storing buckets received from each of the plurality of links;

 The plurality of reception buffers are monitored, and a control packet indicating link switching, which is transmitted when the node device adjacent to the upstream side switches a link for transmitting a packet, is at least one of the plurality of reception buffers. If the control packet is received in one of the plurality of reception buffers, the control unit stops reading the subsequently received bucket in the reception buffer in which the control packet has been received, until the control packet is received in all of the plurality of reception buffers. When a control packet is received by all of the plurality of reception buffers, a read control unit that restarts reading the packets from the plurality of reception buffers and transmits the packets,

 A node device comprising: 4. A user bandwidth allocation method executed by a network management apparatus that manages a communication network in which a plurality of links are provided between adjacent node devices, and allocating a bandwidth of a user to the plurality of links,

 Selecting a link having a free bandwidth greater than or equal to the bandwidth required by the user from the plurality of links;

If one link is selected in the selecting step, the required bandwidth of the user is allocated to the one link, and if two or more links are selected, the available bandwidth is selected from the two or more links. Link with the highest bandwidth, maximum bandwidth Selecting the smaller link, the link with the highest usage, or the link with the lowest usage, and allocating the required bandwidth of the user to the selected link. 25. In Claim 24,

 In the case where the bandwidth of one or more users has already been allocated to the plurality of links, when allocating the required bandwidth of a newly added user, the allocation of the bandwidth of the already allocated user is not changed. Allocating the required bandwidth of the newly added user by the selecting step and the allocating step.

 Method.

26. A user bandwidth allocation method executed by a network management apparatus that manages a communication network in which a plurality of links are provided between adjacent node devices, and allocating a bandwidth of a user to the plurality of links,

 When the bandwidth of one or more users has already been allocated to the plurality of links, when allocating the required bandwidth of a newly added user, releasing the bandwidth of the already allocated user; ,

 Selecting the largest requested bandwidth from among the required bandwidths of the users whose allocation has not been determined among the required bandwidths of the released users and the required bandwidths of the newly added users;

 Selecting a link having an available bandwidth equal to or greater than the selected maximum required bandwidth from the plurality of links;

When one link is selected by the selecting step, the maximum requested bandwidth is allocated to the one link. When two or more links are selected by the selecting step, the two or more links are selected. Of the links, the link with the largest free bandwidth, the link with the smallest free bandwidth, the link with the highest usage rate, or the link with the lowest usage rate is selected, and the maximum required bandwidth is assigned to the selected link. Steps to assign and ' A method comprising: