JP2013038767A - Communication path controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a communication path controller capable of preventing redundancy path from being established by processing a series of network events as an independent event, thus, enhancing the utilization efficiency of network resources.SOLUTION: A communication path controller includes: a control interface connected to a communication node in a communication network for receiving a monitoring report and transmitting a control signal; a memory part for storing the monitoring report received from the control interface as a network event record and a network state record, and storing a predict event classification and a processing method as a predict event processing table; and a control part for acquiring a path control policy by analyzing the network event record and the network state record on the basis of the predict event processing table, and performing the path control policy by transmitting a control signal to the communication node in the communication network through a control interface.

Description

  The present invention relates to a communication path control apparatus that controls an adjustable policy of an inter-domain route and can perform accurate route control for traffic in a communication network.

  Patent Document 1 discloses a control method for dynamically managing a network policy by an attached function of a network service related to a network system. In the control method, the network system monitors the trigger of the network system by acquiring information about the network system, and based on the monitored trigger, one or more static settings configured for the network service are performed. Modify the policy, the dynamic policy, or both. Triggers obtained by network system monitoring include timeouts, attached function changes, network structure changes, ingress detection events, firewall events, supervisor input, network service changes, network service change requests, and the like.

US Pat. No. 7,526,541

  However, Patent Document 1 considers only a simple network action of one independent event as a trigger, does not mention a complex action including a series of network events and an instantaneous network state, and Patent Document 1 Does not support route policy management for complex network activity because it does not include analysis of monitored network events and conditions.

  In order to solve the above-described problems, the present invention provides a communication path control apparatus capable of preventing the establishment of a redundant path by processing a series of network events as independent events and improving the utilization efficiency of network resources. The purpose is to provide.

  A typical example of the present invention is as follows. That is, a communication path control device that is installed corresponding to each domain in the communication network and controls communication paths in the domain, is connected to a communication node in the communication network, receives a monitoring report, and transmits a control signal A control interface that stores the monitoring report information received from the control interface as a network event record and a network status record, and stores a foreseeable event classification and processing method as a foreseeable event processing table, and the seen event processing table. To obtain a path control policy by analyzing the network event record and the network state record, and control to a communication node in the communication network via a control interface Characterized in that it comprises a control unit for executing the path control policy by sending No..

  According to the present invention, by analyzing a combination of a node failure event, a subsequent event (that is, a call event), and a subsequent call state information event, the redundant path caused when the above event is processed as an independent event. Establishment can be prevented and the utilization efficiency of network resources can be improved.

It is a network block diagram which performs the network policy of a centralized network management system. It is a figure which shows a multi domain network block diagram and a cross domain path | pass. It is a figure which shows the condition which received the new call request at the time of cross domain path failure. It is a figure which shows the condition of the intra-domain rerouting path | pass at the time of cross domain path failure. It is a figure which shows the condition of the rerouting path between domains at the time of cross domain path failure. FIG. 4 is a time series diagram for rejecting a new call request in a domain where a failure has occurred and rerouting a path between domains within the domain where the failure has occurred. It is a time series diagram which performs rerouting with respect to the path | pass between domains within the domain in which the failure occurred. It is a time series diagram which performs rerouting simultaneously with respect to the path | pass between domains between domains. It is a flowchart which monitors and analyzes an event and corrects a network policy. It is a flowchart which analyzes the event failure in a domain. It is a flowchart which analyzes a non-domain path | pass failure event. It is a flowchart which calculates the maximum time T1 where domain (1) waits for an access request event. It is a flowchart which analyzes a node recovery event in a domain. It is a foreseeing event processing table in a network management system. It is a figure which shows the multi-domain network block diagram in Example 2, and a cross domain path | route. It is a figure which shows the path | route of the multipath rerouting at the time of the cross domain path | pass failure in Example 2. FIG. 10 is a flowchart illustrating an analysis process of an in-domain failure event according to the second embodiment. 10 is a flowchart illustrating a plurality of routes and weights that can be calculated in the second embodiment.

  Hereinafter, Example 1 is demonstrated based on FIGS.

<Example 1: Case of only two boundary nodes>
FIG. 1 is a network configuration diagram for executing a network policy of a centralized network management system. The communication network 101 includes communication nodes 102, 103, and 104. These communication nodes respectively communicate with communication modules 105, 106, and 107 that communicate with each other based on a control signal and other communication modules. And monitoring units 108, 109 and 11 for monitoring the node status, transmitting a monitoring report and receiving a control signal. The communication nodes 102, 103, and 104 are connected to each other via the communication modules 105, 106, and 107 to perform communication, and at the same time, are connected to external networks (networks in other domains) 111 and 112 to perform communication. The communication network 101 further includes a network management system 113, which is connected to the monitoring units 108, 109 and 110 and receives monitoring reports and receives control reports 114 and receives control signals from the control interface 114. Network event record 115 and network state record 116 for storing the monitored report information, network event record 115 and network state record based on foresee event processing table 117 for storing foresight event classification and processing method, and foresee event processing table 117 Through an event analysis unit 118 that obtains a policy by analyzing information in 116 and a control interface 114 It sends a signal to the communication node 102, 103 and 104 has a policy execution unit 119 to execute the policy event analysis unit 118 has acquired. The network management system shown in FIG. 1 is centralized, but may be distributed, and their selection does not affect the implementation of the present invention. The network event record 115, the network status record 116, and the foreseeing event processing table 117 may be stored in one memory or in different memories in the network management system. When the distributed type is adopted, the above records and tables may be stored in different memories in different network management systems. In addition, the event analysis unit 118 and the policy execution unit 119 may be two functions included in one control unit (for example, CPU), and may be separately realized by two independent chips or integrated circuits. Good.

  2A to 2D show technical solutions for an intra-domain rerouting path and an inter-domain rerouting path that can be performed in a communication network when a cross-domain path fails in a multi-domain network.

  FIG. 2A includes a communication network domain (1), a communication network domain (2), and a communication network domain (3). Each domain has an independent network management system for managing communication nodes in the domain. Domain (1) includes Node A, Node B, Node C, Node D, Node E, Node F and Node G, Domain (2) includes Node H, Node I and Node J, and Domain (3) is Node K, node L, and node M are included. Among them, the domain (1) and the domain (2) are connected via boundary nodes A, D and I, and the domain (2) and the domain (3) are connected via boundary nodes G and K. In each domain, only border nodes can communicate with border nodes in other domains, and other nodes in domains such as B, C, E, F, H, J, L, and M are domain Called an intermediate node. The network has a cross-domain path, i.e. path 1, which starts from the source node H of the path and passes through the nodes J, I, D, E, F, G, K and the destination node of the path M is reached. Communication traffic is reliably and efficiently transmitted from the source node H to the target node M via the path 1. The establishment and deletion of path 1 are performed jointly by the above three domains (reference material RFC2205, Resource Reservation Protocol, 1997). For example, domain (1) is a path from D to G (hereinafter referred to as path DG). ), Domain (2) establishes a path from H to I, domain (3) establishes a path from K to M, and the border node concatenates these paths to form the complete path 1 Form.

  As shown in FIG. 2B, if node F in domain (1) fails due to a power failure or failure, both domain (1) and domain (2) know the failure event and perform appropriate processing. Do. At this time, two rerouting results may be obtained. First, as shown in FIG. 2B, the domain (2) transmits a new call request to the domain (1) via the boundary node I to change the route of the next hop. Or the boundary node I of domain (2) waits for path recovery of domain (1) without changing the next hop route, as shown in FIG. 1) is a case where a path from D to G (hereinafter referred to as path DG) needs to be recovered by intra-domain rerouting. The other is that, as shown in FIG. 2D, domain (2) sends a new call request to domain (1) via border node I and changes the route of border node I's next hop to node A. As a result, it is necessary for the domain (1) to newly establish a path from A to G (hereinafter referred to as a path A-G). This is a case of holding or deleting. Hereinafter, these two situations will be described based on respective time-series diagrams.

  3A and 3B show the situation for path rerouting between domains within the domain where the failure occurred, that is, the first situation.

  In FIG. 3A, the domain (2) sends a new call request to the domain (1) via the border node I to change the route of the next hop, but the domain (1) is rejected (see FIG. 2B). As a result, the domain (1) needs to recover the path DG by intra-domain rerouting (see FIG. 2C). After node F fails, the boundary node D detects that the path D-E has failed (step 301), and transmits a path failure message 302 to the network management system (1) in the domain (1). At the same time, the adjacent nodes E and G of the node F also detect the failure of the node F (step 303), and transmit a node failure message 304 to the network management system (1) of the domain (1). The network management system (1) recognizes the failure of the node F immediately after receiving the path failure message 302 and the node failure message 304 (step 305). Thereafter, the network management system (1) performs intra-node node failure event processing, wakes up the corresponding event analysis subprocess (see the analysis flowcharts of FIGS. 5 and 6A), and starts observation of the network state.

At the same time, the boundary node I of the domain (2) also detects the failure of the path DG (step 306), and sends an out-of-domain path failure message 307 to the network management system (2) of the domain (2). The management system (2) will recognize the out-of-domain path DG failure (step 308). At this time, the network management system (2) has various possible correspondence policies. For example, one of them transmits a transmission delay between the node I and the node A measured by the node I, that is, the delay _IA 309 to the network management system (2), and stops data transmission to the node D. (Step 310). The network management system (2) performs failure event processing based on the received event and the delay report message, wakes up the corresponding event analysis sub-process (see the analysis flowcharts in FIGS. 5 and 6B), and data transmission Is switched from the off-domain path DG to the out-of-domain path A-G (step 330) to obtain a processing result. As a result, path switching of the boundary node I is triggered and a call request 331 is transmitted to the other boundary node A of the domain (1) to establish a new off-domain path AG.

  After receiving the call request from the border node I of the domain (2), the border node A reports it to the network management system (1) of the domain (1) as a call event message 332. Therefore, in the observation period of the state of the boundary nodes A and D of the network management system (1) in the domain (1), the calling state information 313 transmitted to the network management system (1) of the boundary nodes A and D is included in the domain ( A call event message 332 for reporting a call event from node I in 2) is included.

  Based on the call event 332 included in the call state information 313 and other call events other than the call event 332, the intra-domain node failure event processing program of the network management system (1) 2) is performing path switching, and it is determined that the call event 332 is not an event triggered by newly arrived communication traffic but a protection operation in the neighboring domain due to the node failure 304. When considering the service quality parameters of the new path AG and the old path DG required for the call event 332, such as delay and packet loss rate, the network management system (1) A processing result that rejects and triggers intra-domain rerouting, that is, rerouting for path DG (step 312) is obtained. At this time, the network management system (1) needs to respond to the call request 331 from the boundary node I of the domain (2) by transmitting a call rejection command to the node A (step 333). Conversely, when the domain (1) processes the call event 332 and the node failure 304 as independent events, the path DG is rerouted and at the same time, the path AG requested for the call is newly established. In other words, two paths for transmitting the same communication traffic exist simultaneously, and network resources are wasted.

  Subsequently, the network management system (1) calculates a new route and substitutes for the old route, and the obtained calculation result is a D-E-B-C-G route, that is, node D to node E, node B, and The node G is reached via the node C (step 314). By sending a command for establishing a new path to node D (step 315) and sending it to nodes B, C, E, and G (step 317), a new path, ie, path 2 shown in FIG. Establish. After the path 2 is established, the boundary node D recovers the path DG using the new path (step 319), sends a path recovery message 320 to the network management system (1), and the network management system (1 ) Determines to delete the old path D-E-F-G (step 321), transmits a command to delete the old path to the node D (step 322), and transmits to the nodes E and G (step 324). ) To delete the old route.

  After the boundary node I of the domain (2) receives the call request, the network management system (2) recognizes the failure of path switching and determines to wait for recovery of the path outside the domain (step 335). After the path 2 is established, the boundary node I of the domain (2) detects the recovery of the path DG (step 326), and sends an out-of-domain path recovery message 327 to the network management system (2). The external path DG recovery is notified to the network management system (2) (step 328). Thereafter, boundary node I recovers data transmission to boundary node D (step 329), ie, path 2 completely replaces path 1 as rerouting.

  Another possibility of this situation is that in FIG. 3B, the border node I of domain (2) does not change the path of the next hop of border node I and waits for path recovery in domain (1). As a result, it is mentioned that the domain (1) recovers the path DG by intra-domain rerouting (see FIG. 2C). After node F fails, boundary node D detects the failure of path DG (step 301) and sends path failure message 302 to network management system (1) in domain (1). At the same time, the adjacent nodes E and G of the node F also detect the failure of the node F (step 303), and transmit a node failure message 304 to the network management system (1) of the domain (1). The network management system (1) recognizes the failure of the node F immediately after receiving the path failure message 302 and the node failure message 304 (step 305). Thereafter, the network management system (1) performs node failure event processing within the domain, wakes up the corresponding event analysis subprocess (see the analysis flowcharts in FIGS. 5 and 6A), and starts to observe the network state. To do.

At the same time, the boundary node I of the domain (2) also detects the failure of the path DG (step 306), and transmits an out-of-domain path failure message 307 to the network management system (2) of the domain (2). The external path DG failure is notified to the network management system (2) (step 308). At this time, the network management system (2) has various possible correspondence policies. For example, one of them transmits a transmission delay between the node I and the node A measured by the node I, that is, the delay _IA 309 to the network management system (2), and stops data transmission to the node D ( Step 310). The network management system (2) performs failure event processing based on the received event and the delay report message, wakes up the corresponding event analysis subprocess (see the analysis flowcharts in FIGS. 5 and 6B), and out of the domain A processing result waiting for path recovery is obtained (step 311).

  Therefore, during the period in which the network management system (1) in the domain (1) observes the states of the boundary nodes A and D, the calling state message 313 transmitted from the boundary nodes A and D to the network management system (1) includes the domain The call event from node I in (2) is not included. After the network management system (1) has observed for a predetermined time, the intra-domain node failure event processing program of the network management system (1) has determined that the domain (2) has not switched the path. It is not necessary to change the source node and the destination node for rerouting in (1). That is, the network management system (1) obtains a processing result that triggers intra-domain rerouting (analysis flowchart of FIG. 6A), that is, rerouting for the path DG (step 312).

  Subsequently, the network management system (1) calculates a new route and substitutes for the old route, and the obtained calculation result is a D-E-B-C-G route, that is, node D to node E, node B, and The node G is reached via the node C (step 314). By sending a command for establishing a new path to node D (step 315) and sending it to nodes B, C, E, and G (step 317), a new path, ie path 2 shown in FIG. Establish. After the path 2 is established, the boundary node D uses the new path to recover the path DG (step 319), sends a path recovery message 320 to the network management system (1), and the network management system (1 ) Determines to delete the old path D-E-F-G, and transmits a command for deleting the old path to the node D (step 322), and transmits to the nodes E and G (step 324); To delete the old route.

  After the path 2 is established, the boundary node I of the domain (2) also detects the recovery of the path DG (step 326), and sends an out-domain path recovery message 327 to the network management system (2). The external path DG recovery is notified to the network management system (2) (step 328). Thereafter, boundary node I recovers data transmission to boundary node D (step 329), ie, path 2 completely replaces path 1 as rerouting.

  The time-series diagram of FIG. 4 shows a situation where rerouting is simultaneously performed on an inter-domain path between domains, that is, a second situation. After node F fails, boundary node D detects the failure of path DG (step 401), and sends path failure message 402 to network management system (1) in domain (1). At the same time, the adjacent nodes E and G of the node F also detect the failure of the node F (step 403), and transmit a node failure message 404 to the network management system (1) of the domain (1). The network management system (1) recognizes the failure of the node F immediately after receiving the path failure message 402 and the node failure message 404 (step 405). Thereafter, the network management system (1) performs intra-domain node failure event processing, wakes up the corresponding event analysis subprocess (see the analysis flowcharts in FIGS. 5 and 6A), and observes the network status. Start.

At the same time, the boundary node I of the domain (2) also detects the failure of the path DG (step 406), and transmits an out-of-domain path failure message 407 to the network management system (2) of the domain (2). The network management system (2) is notified of the failure of the external path DG (step 408). At this time, the network management system (2) has various possible correspondence policies. For example, one of them transmits a transmission delay between the node I and the node A measured by the node I, that is, the delay _IA 409 to the network management system (2), and stops the data transmission to the node D ( Step 412). The network management system (2) performs event failure processing based on the received event and the delay report message, wakes up the corresponding event analysis sub-process (see the analysis flowcharts in FIGS. 5 and 6B), and data transmission Is switched from the off-domain path DG to the off-domain path A-G (step 410) to obtain a processing result. This triggers path switching of the boundary node I, and sends a call request 411 to the other boundary node A of the domain (1) to establish a new off-domain path AG (step 411).

  After the boundary node A receives the call request from the boundary node I of the domain (2), it reports it as a call event message 414 to the network management system (1) of the domain (1). Therefore, in the observation period of the state of the boundary nodes A and D of the network management system (1) in the domain (1), the calling state information 415 transmitted to the network management system (1) of the boundary nodes A and D includes the domain ( A call event message 414 for reporting a call event from node I in 2) is included.

  Based on the call event 414 included in the call state information 415 and other call events other than the call event 414, the intra-domain node failure event processing program of the network management system (1) 2) is performing path switching, and the call event 414 is determined not to be an event triggered by newly arrived communication traffic, but to be a protection operation in the neighboring domain due to the node failure 404. Therefore, in the domain (1), it is necessary for the source node or the destination node to newly establish a path A-G different from the old path DG to replace the path DG. That is, the processing result obtained by the network management system (1) triggers inter-domain rerouting (see the analysis flowchart of FIG. 6A). That is, in accordance with the path switching of the domain (2), routing is performed for the path AG, and the path DG is replaced with the path AG (step 413). Conversely, when the domain (1) processes the call event 414 and the node failure 404 as independent events at this time, the path DG required for the call is newly established and the path DG is simultaneously established. There is a need to reroute, i.e. two paths exist simultaneously to carry the same communication traffic, and network resources are wasted.

  Subsequently, the network management system (1) calculates a new route from A to G, and the obtained calculation result is an A-B-C-G route, that is, from node A to node B and node C. This is a route that reaches the node G via the route (step 416). A new path, that is, path 3 shown in FIG. 2D, is established by sending a command for establishing a new path to node A (step 417) and sending it to nodes B, C, and G (step 419). To do. After the path 3 is established, the boundary node A uses the newly established path A-G as a path from the node A to the node G (step 421), transmits the call permission 422, and the boundary of the domain (2) Responds to the call request 411 from the node I.

  After the boundary node I of the domain (2) receives the call permission 422, the communication traffic transmitted by the original path 1 is switched to be transmitted by the path 3 (step 423), that is, the data transmission to the node A is performed. Do. After the switching is successful, the node I transmits a switching success message 425 to the network management system (1) to notify the network management system (1) of the successful path switching (step 426).

  At the same time, the network management system (1) decides to delete the old route D-E-F-G (step 427), transmits an old path deletion command to the node D (step 428), and node E The old route D-E-F-G is deleted (step 432).

  Therefore, according to the present invention, by analyzing the node failure 304, 404 event and the subsequent event corresponding thereto, that is, the call event 332, 414 and the corresponding subsequent call state information 313, 415 in combination, (1) prevents the establishment of a redundant path caused when the above event is processed as an independent event, thereby improving the utilization efficiency of network resources.

  In the above process, the network management system (1) and the network management system (2) monitor and analyze events, and a flowchart of processing for correcting the network policy is as shown in FIG. During the operation period of the communication network, the control interface 114 of the network management system continuously receives network event and status report information from the communication node (step S501). Among them, network events can include, but are not limited to, intra-domain node failure, intra-domain node recovery, off-domain path failure, VPN (Virtual Private Network) establishment, VPN deletion, congestion, and decongestion. Not. The network state is generally a sampling value obtained by sampling state variables that are continuous over time in the network, and includes a call state, a node load, a transmission band, a packet loss rate, a delay, and an access traffic band. However, the present invention is not limited to these (see the foreseeing event processing table in FIG. 9). After receiving the network event and status report information, the control interface 114 writes them to the memory, that is, the network event record 115 and the network status record 116 in the network management system memory (step S502). At the same time, the event analysis unit 118 continuously reads the network event and network status sampling from the network event record 115 and the network status record 116 in the network management system memory (step S503). (See step S504, analysis flowcharts in FIGS. 6A and 6B). After that, it is determined whether or not the network policy correction such as the path rerouting method provided for one communication traffic is necessary as a result of the analysis in step S504 (step S505). If the determination result is “NO”, it indicates that all current policies should be maintained without modification, and the event analysis unit 118 returns to step S503 and continues to process the next event. Conversely, if the determination result in step S505 is “YES”, it indicates that at least one policy needs to be modified, and the event analysis unit 118 modifies the corresponding policy and notifies the policy execution unit 119 ( After step S506 (see the analysis flowchart of FIGS. 6A and 6B), the process returns to step S503 to continue processing the next event. At the same time, the policy execution unit 119 executes an existing policy, for example, a policy in which the path rerouting method to be provided for one communication traffic performs intra-domain rerouting, that is, a policy for recovering the path DG (step S507). When a policy modification command from the event analysis unit 118 is received, for example, when a path rerouting method provided for one communication traffic receives an inter-domain rerouting, that is, a policy substituting the path DG with the path A-G, A new policy is executed instead of the current policy (step S508).

  In the above process, when the event analysis unit 118 analyzes the network event and the network state (step S504), it is necessary to use different analysis processes and parameters for different network events. The specific analysis process is as shown in FIGS. 6A and 6B. The associated analysis process and parameter type may be foreseeing information stored in the foreseeing event processing table 117 (see the foreseeing event processing table in FIG. 9).

  FIG. 6A is a flowchart of processing in which the event analysis unit 118 of the network management system (1) analyzes an in-domain failure event. The event analysis unit 118 first searches the foresee event processing table 117 based on the intra-domain node failure event that is the current event (step S601), and finds a matching table item whose index is 11 in the foresee event processing table 117. Acquired and acquires the intra-domain node failure analysis process which is the corresponding analysis subprocess by the processing subprocess in the foreseeing event processing table 117, and wakes up the process (step S603). If the search event in step S601 is another event, other analysis sub-processes, for example, process sub-processes for other table items in FIG. 9, are woken up (step S602).

Thereafter, the event analysis unit 118 proceeds to the intra-domain node failure analysis process, and the subsequent observation time T1, the subsequent observation data source node A and the node D of the matching table item in the foreseeing event processing table 117, and the subsequent observation data type call. Based on the state information and the path service quality information, the call state information from the nodes A and D is observed and a timer is started (step S604). The event analysis unit 118 selectively reads only the call status information from the node A and the node D from the network event record 115 and the network status record 116 in the network management system memory, and the redundancy which does not belong to the failed path therein. It is checked whether a new call request from the sex boundary node (ie, node A) is included (step S605). If the determination result in step S605 is “NO”, it is checked whether the timer has exceeded the subsequent observation time T1 (step S606). When the timer times out, the domain (2) determines that the path has not been switched within the set time T1, and the obtained analysis result triggers intra-domain rerouting, that is, rerouting for the path DG ( Step S607). If the timer has not timed out, the process returns to step S604 and observation is continued. When the determination result in step S605 is “YES”, the request source, the target node, and the requested bandwidth of the existing call in the call request and the call state information of the node D are checked (step S608). It is determined whether the call request matches any existing call (step S609). If there is an existing call to match, it can be determined that domain (2) is switching paths. At this time, the service quality information between the source node and the destination node in the new path domain (1) requested by the call request from the network status record 116 in the network management system memory and the old path domain (1). It is necessary to read out and consider the quality of service information between the source node and the target node, such as delay or packet loss rate (step S610). When the delay _AG is not smaller than the delay _DG and the packet loss rate between the node A and the node G (that is, the packet loss rate _AG ) is not smaller than the packet loss rate _DG , intra-domain rerouting, that is, path _DG Is preferentially triggered (step S607). If the delay _AG is smaller than the delay _DG or if the packet loss rate _AG is smaller than the packet loss rate _DG , it triggers preferentially inter-domain rerouting, that is, replacing the path _{DG with the} path A-G. (Step S607). If there is no matching existing call, the call request is a call request triggered by new communication traffic, and it can be determined that it is not a path switching operation of domain (2) due to a node failure event. The call request need only be processed based on an existing policy that establishes a new path for the call request. In other words, if there is no matching existing call, there is no need to modify the network policy, so the program moves to step S606 to check whether the timer has timed out and continues observation based on the timeout determination result. Or end the observation.

  FIG. 6B is a flowchart of processing in which the event analysis unit 118 of the network management system (2) analyzes an out-of-domain path failure event. The event analysis unit 118 first searches the foreseeing event processing table 117 based on the current event, that is, the out-of-domain path failure event (step S601), acquires the corresponding matching table item, and in the foreseeing event processing table 117, An out-of-domain path failure analysis process, which is a corresponding analysis sub-process, is acquired by the processing sub-process, and the process is woken up (step S612). If the search event in step S601 is another event, the other analysis subprocess is woken up (step S602).

Then, the event analysis unit 118 proceeds to the out-of-domain path failure analysis process, and based on the succession observation time 0, the succession observation data source node I, and the succession observation data type transmission delay information in the matching table item in the foresee event processing table 117, The transmission delay information from the node I is observed (step S613). 9 shows only the status of the network management system (1) of the domain (1), and the status of the network management system (2) of the similar domain (2) is omitted. Therefore, the successor observation data source node here is the boundary node I of the domain (2), not the boundary node D of the domain (1). The event analysis unit 118 determines the quality of service of the redundant boundary node (ie, node A) that does not belong to the failure path from the node I to the domain (1) from the network event record 115 and the network state record 116 in the network management system memory. Information, for example, transmission delay information or packet loss rate is selectively read (step S613), and the transmission delay between node I and node A (ie, delay _IA ) is smaller than a predetermined value T2 or the packet loss rate between them. It is determined whether (that is, the packet loss rate _IA ) is smaller than a predetermined value P _th (step S614). Among them, the values of T2 and _Pth can be set _manually based on experience values. For example, T2 can be set to a maximum switching time of 50 milliseconds. When the determination result in step S614 is “NO”, if the communication traffic is switched from the path DG to the path AG, the communication quality may be deteriorated. Therefore, the obtained analysis result is waiting for the path DG recovery. Yes (step 615). When the determination result in step S614 is “YES”, when communication traffic is switched from the path DG to the path AG, there is no risk of communication quality deterioration, and at the same time, a large amount of rerouting traffic is concentrated on a node having traffic. Therefore, it is necessary to switch the communication traffic to the path A-G by replacing the path DG with the path A-G. is there. Therefore, the network management system (2) of the domain (2) controls to issue a call request from the border node I to the node A (step 616), and the call authorization from the node A (step 617) or Waiting for call rejection (step 618), until the call request is accepted (decision result of step 617 is “NO”) or call rejection (decision result of step 618 is “NO”) Continue retrying. That is, the process proceeds to step 616. After receiving the call rejection from the node A (the determination result of step S618 is “YES”), give up switching of the path and move to step S615 to wait for the recovery of the path DG, or from the node A After receiving the call permission (step S617 determination result is “YES”), the communication traffic of path 1 (ie, HJI-D-G-K-M) is transferred to path 3 (ie, HJ). -I-A-G-K-M) (step S619).

In the process in which the network management system (1) in FIG. 6A analyzes the in-domain failure event, the maximum possible delay in which the network management system (2) in FIG. 6B analyzes the out-of-domain path failure event, that is, the timer times out. It is necessary to wait for the subsequent observation time T1 for determining whether or not the domain (2) decides to switch the path. FIG. 7 is an exemplary flowchart of a process for calculating the maximum time T1 waiting for the domain (2) to issue an access request event in the domain (1). The event analysis unit 118 of the network management system (1) of the domain (1) first checks all external routing tables such as the routing table of the boundary node that executes BGP, and in the domain (1), the domain (2) All boundary nodes connected to, that is, node A and node D are acquired (step S701). Thereafter, the network topology information is searched to obtain the number of hops and the distance reaching the source node H of the path 1 from the node A and the node D, respectively, and the maximum hop number N _MAX and the maximum distance D _MAX are selected. (Step S702). Then, the successor observation time T1 is set so as to be the sum of the maximum queue delay, the transmission delay, and the event processing delay required for the network management system (2) of the domain (2), that is, Equation (1) (step). S703).

In Equation (1), N _max is the maximum number of transmission hops, T _avg is the queue delay per average hop for prediction, D _max is the maximum transmission distance, c is the speed of light, and T _pr is the event processing delay for prediction. The successor observation time T1 can be set by writing the time value calculated in the corresponding table item of the foreseeing event processing table 117 (see the foreseeing event processing table in FIG. 9).

  The processing process at the time of node failure has been described above, but it is necessary for the communication network to adopt a corresponding processing policy when the node recovers. FIG. 8 is a flowchart of processing in which the event analysis unit 118 of the network management system (1) analyzes the intra-domain node recovery event. The event analysis unit 118 first searches the foreseeing event processing table 117 based on the current event, that is, the intra-domain node recovery event (step S801), acquires the matching table item whose index is 12, and the foreseeing event processing table. The intra-domain node recovery analysis process, which is the corresponding analysis subprocess, is acquired by the processing subprocess in 117, and the process is woken up (step S803). If the search event in step S801 is another event, the other analysis subprocess is woken up (step S802).

Then, the event analysis unit 118 proceeds to the intra-node node recovery analysis process, the succession observation time T3 of the matching table item in the foreseeing event processing table 117, the succession observation data source node A, the node D, and the succession observation data type node load information. Based on the above, observation of node load information from node A and node D is started (step S804). The event analysis unit 118 selectively reads out only the node load information from the node A and the node D from the network event record 115 and the network state record 116 in the network management system memory, and within the time zone of the follow-up observation time T3. The average values, that is, the average load L _A of node _A and the average load L _D of node _D are calculated. Then, it is determined whether or not the difference between the average loads of the node A and the node D is larger than a predetermined value L _th (step S805). Of these, the value of T3 can be set manually based on experience values. For example, T3 includes 100 sampling points, and the sampling time is usually several nanoseconds. If the determination result in step S805 is “NO”, it indicates that the load assignments of the node A and the node D are relatively uniform, and it is not necessary to switch the communication traffic from the node A to the node D. Therefore, the obtained analysis result triggers intra-domain rerouting for node F recovery only (step 806). If the determination result in step S805 is “YES”, it indicates that the current load on node A is too large and that some communication traffic needs to be switched from node A to node D. Therefore, the obtained analysis result increases the path weight of the path DG and triggers the intra-domain rerouting (step 807), thereby causing more communication traffic to pass through the path DG, that is, the node D. To reduce the load on node A and improve the performance of the network.

  FIG. 9 shows a foreseeable event processing table 117 in the exemplary network management system, which includes an index for indexing each table item, wakes up for event types that can be identified and processed by the network management system, and different event types. The processing sub-process of the power analysis processing program, the successor observation time for which the information of the network event record 115 and the network status record 116 needs to be continuously read when processing different event types, and processing different event types Network event record 115 and network status record 116 information must be read continuously, the reporter's successor observation data source, and when processing different event types Successor observation data type of the data types that must be read to continue the information over click event record 115 and network state records 116 are defined.

  Hereinafter, Example 2 is demonstrated based on FIGS.

<Embodiment 2: Adopted simultaneously when a plurality of routes are available>
Example 2 network configuration diagram, time-series diagram, flowchart for monitoring and analyzing events and correcting network policy, flowchart for analyzing out-of-domain path failure event, flowchart for calculating waiting time T1, analyzing node recovery event in domain Since the flowchart to be processed and the foreseeing event processing table in the network management system are the same as those in the first embodiment, the description thereof is omitted here.

  10A and 10B illustrate a multipath rerouting method using an intradomain rerouting path and an interdomain rerouting path that a communication network may execute when a cross domain path in a multidomain network fails. .

  FIG. 10A is a configuration diagram of a multi-domain network and a cross-domain path based on the second embodiment. FIG. 10A includes a communication network domain (1), a communication network domain (2), and a communication network domain (3). Different domains have independent network management systems that manage the communication nodes in the domain. Domain (1) includes node A, node B, node C, node D, node E, node F, node G and node P, domain (2) includes node H, node I and node J, and domain ( 3) includes node K, node L and node M. Among them, the domain (1) and the domain (2) are connected by boundary nodes A, D, P, and I, and the domain (2) and the domain (3) are connected by boundary nodes G and K. In one domain, only border nodes can communicate with border nodes in other domains, and other nodes in the domain, such as B, C, E, F, H, J, L, and M are in the middle of the domain This is called a node. There is a path 1 which is a cross-domain path in the network, starting from the source node H of the path, and reaching the target node M of the path via the nodes J, I, D, E, F, G, K. . Communication traffic is reliably and efficiently transmitted from the source node H to the target node M through the path 1. The establishment and deletion of path 1 is performed jointly by the above three domains (see RFC 2205, Resource Reservation Protocol, 1997). For example, domain (1) is a path from D to G (hereinafter referred to as path DG). , Domain (2) establishes a path from H to I, domain (3) establishes a path from K to M, and a boundary node concatenates these paths to form a complete path 1.

  As shown in FIG. 10B, when the node F in the domain (1) fails due to a power failure or failure, both the domain (1) and the domain (2) recognize the failure event and perform processing corresponding thereto. To do. The difference from the first embodiment is that the domain (1) does not need to perform rerouting by selecting only one of a plurality of possible rerouting paths, and performs rerouting using a plurality of paths simultaneously. To balance traffic between multiple paths to better divert communication traffic when there is a node failure in the domain and prevent network congestion caused when rerouting large amounts of communication traffic It is to be. For example, domain (2) wants to send a new call request to domain (1) via border node I and change the next pop path of border node I to node A, but domain (1) A path from G (hereinafter referred to as path AG) and a path from P to G (hereinafter referred to as path PG) are newly established at the same time, and an old path DG (see the analysis flowchart of FIG. 11). Is deleted. In establishing a new path A-G and path PG, domain (1) can use the border nodes that receive the call request, eg, border nodes available via node A (ie, node A and node P). , Responding to a call request from the boundary node I of the domain (2) by transmitting a call permission including a plurality of paths (ie, paths AG and PG) and a weight of each path to the domain (2). Thus, after the boundary node I of the domain (2) receives the call permission, the original communication traffic transmitted via the path 1 is switched to the path 3 and the path 4 according to the proportionality of the path weight. assign. That is, data transmission is performed to the node A and the node P. Of course, if the domain (2) does not send a new call request to the domain (1), the domain (1) will pass from D to G only by intra-domain rerouting (hereinafter referred to as path DG). It is also possible to recover. Since the latter is similar to the case of the first embodiment, the description is omitted here.

  Hereinafter, a situation where rerouting is performed using a plurality of paths will be described. After the node failure event occurs, the event analysis unit 118 analyzes the network event and the network status. The specific analysis process is as shown in FIG. The associated analysis process and parameter type may be foreseeing information stored in the foreseeing event processing table 117 (see foreseeing event processing table in FIG. 9).

  The event analysis unit 118 first searches the foreseeing event processing table 117 based on the current event, that is, the intra-domain node failure event (step S1101), obtains the matching table item whose index is 11, and the foreseeing event. The intra-domain node failure analysis process, which is the corresponding analysis sub-process, is acquired by the processing sub-process in the processing table 117, and the process is woken up (step S1103). If the search event in step S1101 is another event, other analysis subprocesses such as the process subprocess for other table items in FIG. 9 are woken up (step S1102).

  Thereafter, the event analysis unit 118 proceeds to the intra-domain node failure analysis process, and the succession observation time T1, the succession observation data source node A, the node D, the node P, and the succession observation data of the matching table item in the foresee event processing table 117. Based on the type call state information and the path service quality information, the call state information from the nodes A, D, and P is observed and a timer is started (step S1104). The event analysis unit 118 selectively reads only the call state information from the node A, the node D and the node P from the network event record 115 and the network state record 116 in the network management system memory, and the failure path therein It is checked whether or not a new call request from a redundant boundary node (that is, node A or node P) that does not belong to is included (step S1105). If the determination result in step S1105 is “NO”, it is checked whether the timer has exceeded the subsequent observation time T1 (step S1106). If the timer has timed out, it is determined that the domain (2) has not switched the path all the time within the set time T1, and the obtained analysis result is an intra-domain rerouting, that is, a rerouting trigger for the path DG ( Step S1107). If the timer has not timed out, the process returns to step S1104 to continue observation. When the determination result in step S1105 is “YES”, the request source, the target node, and the requested bandwidth of the existing call in the call request and the call state information of the node D are checked (step S1108). It is determined whether or not the call request matches any existing call (step S1109). If there is an existing call that matches, it can be determined that the domain (2) is switching paths. It is necessary to calculate a plurality of possible paths at this time, and the quality of service information of these paths is read from the network status record 116 and the weight of each path is calculated (see step S1110, calculation flowchart of FIG. 12). Thereafter, inter-domain rerouting using a plurality of paths, that is, the substitution of the path DG with the path of the path AG and the path PG is triggered (step S1111). If there is no matching existing call, the call request is a call request triggered by new communication traffic and can be identified as not a path switching operation of domain (2) due to a node failure event. It is only necessary to process the call request based on an existing policy that establishes a new path for the call request. That is, if there is no matching existing call, there is no need to modify the network policy, so the program proceeds to step S1106 to check whether the timer has timed out and whether to continue observation based on the timeout determination result. Or stop observation.

  A flowchart for calculating possible paths and weights in the process is shown in FIG.

The event analysis unit 118 first checks all external routing tables, such as the routing table of the boundary node that executes BGP, and all boundary nodes or nodes connected to the domain (2) in the domain (1). A, a node D, and a node P are acquired, the boundary node D through which the original path passes is deleted, thereby acquiring all the redundant boundary nodes A and P (step S1201). Thereafter, the paths A-G and _PG from the boundary nodes A and P to the target node G are calculated (step S1202), the service quality information is searched, and the delay of each path, that is, the delay _AG and the delay _PG are calculated. (Step S1203). In the calculation process of step S1203, the delay of the path n can be expressed by the following formula (2).

In Equation (2), d _xy is a link delay between two adjacent nodes x and y in the path n. The weight of the path i can be further defined by the following mathematical formula (3) (step S1204).

In Equation (3), P _i is the weight of path i, d _i is the delay of path i, d _n is the delay of path n, and the value of n is all possible paths obtained in step S1202. In this embodiment, nε {3, 4}. After the above calculation is completed, an available boundary node (ie, node A and node P), a plurality of paths (ie, paths A-G and paths) via the boundary node that has received the call request such as node A. PG) and the weight of each path (ie, P3 and P4) are sent to the domain (2) to respond to the call request from the node I in the domain (2) ( Step S1205).

  The typical embodiments of the present invention will be exemplified below.

  A communication path control device according to the present invention is a communication path control device that is installed corresponding to each domain in a communication network and controls a communication path in the domain, and is connected to a communication node in the communication network and sends a monitoring report. A control interface that receives and transmits a control signal, a memory unit that stores monitoring report information received from the control interface as a network event record and a network state record, and stores a classification and processing method of foreseeing events as a foreseeing event processing table And obtaining a path control policy by analyzing the network event record and the network state record based on the foreseeing event processing table, and in the communication network via a control interface It sends a control signal to the signal node and a control unit for executing the path control policy.

  In the communication path control device, after an in-event node failure has occurred, the control unit determines that the neighboring domain is performing path switching for the old path affected by the node failure within a first predetermined time. In this case, the path control policy is determined based on the service quality information of the path in the domain in the desired switching path of the neighboring domain and the service quality information of the path in the domain in the old path.

  In the communication path control device, after an in-event node failure occurs, the control unit determines whether the new call request from the neighboring domain, the request source of the existing call request, the destination node, and the requested bandwidth are the same. If it is determined that they are the same, the neighboring domain has switched the path to the old path affected by the node failure, and the existing call request is one boundary in the domain. In the request received via the node, the new call request is a request received via another boundary node in the domain.

  In the communication path control device, when the delay and packet loss rate are included in the quality of service information, the delay of the part in the domain in the desired switching path of the neighboring domain is the part of the domain in the old path. When the packet loss rate of the part in the domain in the desired switching path in the neighboring domain is equal to or greater than the packet loss rate in the part of the domain in the old path, the control unit performs the intra-domain re-transmission. Trigger routing, the delay in the part of the domain in the desired switching path of the neighboring domain is smaller than the delay in the part of the domain in the old path, or in the domain in the desired switching path of the neighboring domain The packet loss rate of the part is more than the packet loss rate of the part in the domain in the old path When again, the control unit triggers the re-routing between domains.

  In the communication path control device, after an out-of-event node failure occurs, the control unit determines from the network event record and the network state record that the old path affected by the node failure in the neighboring domain from the boundary node of the domain. Quality of service information between redundant boundary nodes that do not belong to the redundant boundary node is read, and a call request for path switching to the redundant boundary node is sent to the redundant boundary node based on the quality of service information The control unit transmits a call request to the redundancy boundary node and receives a call permission from a neighboring domain, the communication traffic of the old path is changed to a path after path switching. Switch to.

  In the communication path control device, when the service quality information includes a delay and a packet loss rate, a transmission delay between a boundary node of the domain and the redundancy boundary node of a neighboring domain is greater than a first predetermined value. When the packet loss rate between the boundary node of the domain and the redundancy boundary node of the neighboring domain is smaller than a second predetermined value, the control unit transmits the redundancy boundary node to the redundancy boundary node. It decides to issue a call request for path switching to the node.

  In the communication path control apparatus, after the failure node in the domain is recovered, the control unit reads node load information from two boundary nodes from the network event record and the network state record, and the second of the two boundary nodes. When the average value within a predetermined time is calculated, and the difference between the two average values is larger than the third predetermined value, the path weight of the intra-domain path including the node with the small average value is increased, and the intra-domain Trigger rerouting.

  In the communication path control device, after the intra-domain node failure occurs, the control unit performs path switching for the old path in which the neighboring domain is affected by the failure of the node within the first predetermined time and If it is determined that there are two or more possible rerouting paths between a domain and a neighboring domain, read the quality of service information for these paths from the network status record, calculate the weight of each path, and then Trigger interdomain rerouting using paths.

  For example, when a node failure occurs in the domain (1) in the network and the path across the domain (1) and the domain (2) becomes unusable, the domain (2) may perform path switching. Send the call request to other available border nodes other than the border node through which the failed path passes. At this time, when the domain (1) processes the call event and the node failure event from the domain (2) as independent events, the path required by the call is newly established and the old path is rerouted at the same time. There is a need. That is, two paths exist simultaneously for transmitting the same communication traffic, and network resources are wasted. Conversely, if the domain (1) determines that the call event is not an event triggered by newly arrived communication traffic but a protection operation in the neighboring domain by a node failure event, the call request is made. It is possible to suppress waste of resources by establishing a new path and deleting the old path at the same time.

  In some other designs, domain (2) may inform domain (1) of the above information. However, it is necessary to monitor and analyze such event combinations when information exchange for operation management maintenance between the domain (1) and the domain (2) is not available.

Claims (12)

A communication path control device that is installed corresponding to each domain in a communication network and controls a communication path in the domain,
A control interface connected to a communication node in the communication network for receiving a monitoring report and transmitting a control signal;
A memory unit for storing the monitoring report information received from the control interface as a network event record and a network state record, and storing a foreseeable event classification and processing method as a foreseeable event processing table;
A path control policy is obtained by analyzing the network event record and the network state record based on the viewing event processing table, and a control signal is transmitted to a communication node in the communication network via a control interface. And a control unit that executes a control policy.   After the intra-domain node failure occurs, the control unit determines that the neighboring domain is performing path switching for the old path affected by the node failure within the first predetermined time. The path control policy is determined based on quality of service information of a path in a part of the domain in the path and quality of service information of a part of the path in the domain in the old path. Communication path control device. After an intra-domain node failure occurs, the control unit determines whether the new call request from the neighboring domain and the request source of the existing call request, the target node, and the requested bandwidth are the same. If it is determined that there is, the neighboring domain has switched the path for the old path affected by the node failure.
The existing call request is a request received via one boundary node in the domain,
3. The communication path control device according to claim 2, wherein the new call request is a request received via another boundary node in the domain. When the service quality information includes delay and packet loss rate,
The delay of the part in the domain in the desired switching path of the neighboring domain is equal to or greater than the delay in the part of the domain in the old path, and the packet loss rate of the part in the domain in the desired switching path of the neighboring domain Triggered by the packet loss rate of the part of the domain in the old path being the trigger, the control unit starts intra-domain rerouting,
The delay of the part of the domain in the desired switching path of the neighboring domain is smaller than the delay of the part of the domain in the old path, or the packet loss rate of the part of the domain in the desired switching path of the neighboring domain 3. The communication path control device according to claim 2, wherein the controller starts inter-domain rerouting by using a packet loss rate smaller than a packet loss rate in a portion of the domain in the old path as a trigger. After an out-of-domain node failure occurs, the control unit is a boundary node that does not belong to the old path affected by the node failure in the neighboring domain from the boundary node of the domain from the network event record and the network state record. Read quality of service information with respect to the redundant boundary node and, based on the quality of service information, determine whether or not to send a call request for path switching to the redundant boundary node to the redundant boundary node. ,
The control unit transmits a call request to the redundancy boundary node and switches communication traffic of the old path to a path after path switching when receiving a call permission from a neighboring domain. Item 4. The communication path control device according to Item 1. When the service quality information includes delay and packet loss rate,
The transmission delay between the boundary node of the domain and the redundancy boundary node of the neighboring domain is smaller than a first predetermined value, or the packet loss between the boundary node of the domain and the redundancy boundary node of the neighboring domain 6. The control unit according to claim 5, wherein when the rate is smaller than a second predetermined value, the control unit determines to send a call request for path switching to the redundancy boundary node to the redundancy boundary node. Communication path control device.   After the failure node in the domain is recovered, the control unit reads node load information from the two boundary nodes from the network event record and the network state record, and averages the two boundary nodes within a second predetermined time. When the difference between the two average values is larger than the third predetermined value, the route weight of the intra-domain path including the node having the small average value is increased, and the intra-domain rerouting is started. The communication path control device according to claim 2.   After a node failure in the domain occurs, the control unit performs path switching for the old path in which the neighboring domain is affected by the node failure within the first predetermined time, and the domain and the neighboring domain If it is determined that there are two or more reroutable paths in between, read the quality of service information of these paths from the network status record, calculate the weight of each path, and then between domains using multiple paths The communication path control apparatus according to claim 1, wherein rerouting is started.   The network events include intra-domain node failure, intra-domain node recovery, and off-domain path failure, and further include establishment of a virtual private network, deletion of a virtual private network, congestion and congestion deletion. The communication path control apparatus according to claim 1.   The network state is a sampling value obtained by sampling with respect to a state variable in which the network continues in time, and includes a call state, a node load, a transmission band, a packet loss rate, a delay, and an access traffic band. The communication path control apparatus according to claim 1, further comprising:   The communication path control apparatus according to claim 1, wherein the first predetermined time is a maximum time for waiting for a neighboring domain to transmit a call event.   The foreseeing event processing table includes a plurality of table items, and each table item includes an index, an event type, a processing sub-process, a subsequent observation time, a subsequent observation data source, and a subsequent observation data type. The communication path control device according to claim 1.