JP2010268275A - Network management system, network management method, and program therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a network management technology that can be applied to a network with an uneven configuration and capacity and can evaluate the reliability of the network based on the variation in defective rate caused by sequential installation of multiple types of equipment and the influence of prolonged interruption of service on users when a trouble occurs. <P>SOLUTION: In a step 0, failure information, information on an object to be determined, a reference value for determining the execution, and a reference value for determining the reliability are input. In a step 1, the total number of failures of the network is calculated from the failure information. In a step 2, it is determined based on the total number of failures whether or not subsequent steps need be executed. In a step 3, the impact of failure is calculated based on the down time of each failure and the number of influenced users when each failure occurs. In a step 4, a probability distribution for the impact of failure is estimated as a degree of confidence from the total number of failures and the impact of failure. In a step 5, it is determined whether or not the reliability of the network is acceptable based on the estimation result for the degree of confidence. In a step 6, the determination result is displayed on a display unit. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a communication network technology realized by a network protocol typified by IP (Internet Protocol), and in particular, a component constituting a network (a device for transferring a packet such as an IP packet, a link for connecting devices). Based on the failure information of the network equipment such as transmission lines and transmission line devices to configure), the physical reliability of the network, the failure time (failure duration) associated with the failure, and the impact scale (number of affected users) ) To calculate the degree of failure impact, determine whether the network reliability is good or bad, and select failure cases and network equipment that have a large impact on reliability degradation. Network management that can automatically and accurately realize operation management of networks System and network management method, and to a program therefor.

  As the social infrastructure of communication networks advances, it is indispensable to reduce the impact of network failures and failures on user service usage. On the other hand, as the IP network technology progresses, the network is constructed on the premise of aggressively introducing commercial technology and general-purpose products, and therefore the development period of related devices and the network construction period tend to be shortened.

  In addition, there are increasing opportunities to introduce various types of network devices, such as newly developed devices with little operational track record and devices with various functions. Not only deterioration over time, but also component defects and non-compliance with setting conditions, etc. There is a risk that various causes of the device stop coexist and the cause of failure coexists.

  Conventionally, the reliability of a communication network depends on the equipment (for example, network components such as network devices and links) constituting the target network and the equipment configuration (facility connection configuration or redundant configuration). A model is created and calculated using the product or sum of failure probabilities (for example, failure rate and unavailability) of individual network equipment.

  Then, based on the failure probability required for the relevant communication network, the failure probability required for each network equipment is allocated, and based on the allocated failure probability and the reliability evaluation model. It has been evaluated by comparing the calculated failure probability.

  For the conventional calculation method of network reliability, see “To understand the NTT communication network” (NTT Communication Network Study Group, 1994), “Chapter 8 Stable Quality”, pp. 314-329 (Non-Patent Document 1). The reliability evaluation technique disclosed in this document is a technique for supporting accurate judgment in implementing network planning, design, and construction.

  In addition, in the conventional network design method considering reliability, the network designer calculates the failure probability of individual network equipment (components) constituting the network in the network section that is the target of the network design. Select the network equipment with a low failure probability, select a network equipment configuration with a low failure probability, or accommodate in the network equipment so that the calculated failure probabilities are added and the total value satisfies the predetermined target value. The number of subscribers to be set is being limited.

  For the network reliability design method, refer to “Reliability Handbook” (Reliability Society of Japan, 1997) “1.5.3 Network Improvement Plan Considering Reliability-Reliability of Communication Network Focusing on Trouble Level at Failure Sex prescription and design method- ", pp. 79-84 (Non-Patent Document 2).

  In addition, as will be described later, issues related to IP networks include "Safety and reliability standards for IP network (draft)", Ministry of Internal Affairs and Communications Information and Communication Council, Information and Communication Technology Subcommittee, IP Network Equipment Committee , Safety and Reliability Work Group, December 20, 2007 (Non-Patent Document 3).

“To understand the NTT communication network” (NTT Communication Network Study Group, 1994) “Chapter 8 Stable Quality”, pp. 314-329 "Reliability Handbook" (Japan Reliability Association, 1997) "1.5.3 Network Improvement Plan Considering Reliability-Communication Network Reliability Specification / Design Method Focusing on Troubles at Failure-", pp . 79-84 “Safety / Reliability Standards for Network IP (Draft)”, Ministry of Internal Affairs and Communications Information and Communication Council, Information and Communication Technology Subcommittee, IP Network Equipment Committee, Safety and Reliability Working Group, December 20, 2007 )

(Problem 1)
The conventional techniques (Non-patent Document 1) and (Non-Patent Document 2) described above are redundant from the network equipment closest to the end user for a plurality of end users who contract with a specific carrier (for example, a specific carrier). Network equipment deployment configuration (type of equipment deployed and number of stages deployed) in the communication section that goes through multiple network equipment that does not have a configuration are all the same, and the network specific device (specific function or model) It is assumed that the number of subscribers accommodated in devices belonging to the same level is the same. Under this assumption, it is possible to aggregate failure cases for each specific device of the network and calculate the influence of the failure on the user as the network failure probability by the sum or product of the average values of the failure probabilities.

  On the other hand, in a communication network represented by an IP network, the deployment configuration of network equipment is sequentially changed according to the change in the number of subscribers and the plan for service provision, so the number of network equipment deployment changes with time. . In addition, the number of subscribers that can be accommodated in a specific device varies greatly depending on the device.

  Therefore, in this case, since it is necessary to calculate failure probabilities for all the configurations every time there are various changes in the deployment configuration and the accommodation configuration of the network equipment, the calculation amount increases exponentially according to the situation. In addition, since the failure probability of the network cannot be calculated by a simple sum or product of failure probabilities calculated by summing up failure cases for each specific device of the network, in the conventional technology, in a communication network represented by an IP network, There is a problem that the failure probability of the network cannot be properly evaluated.

(Problem 2)
The conventional techniques (Non-Patent Document 1) and (Non-Patent Document 2) described above are based on the failure probability of the network using the average value of the failure rate and the service interruption time associated with the failure for the network equipment in the network section to be evaluated. Based on the reliability required for the network, set the reference value for each network facility in advance, and evaluate the failure probability of each network facility by comparing with the reference value. Is the method.

  On the other hand, in a communication network represented by an IP network, since a network is configured by combining a plurality of model devices provided as commercially available products, a plurality of types of network equipment (model devices) are also provided for specific functions. ), New model devices have been introduced successively after the start of network operation.

  In addition, since a plurality of model apparatuses having different operation periods are deployed, the statistical characteristics of the number of occurrences (frequency) and the rate of occurrence (occurrence frequency per hour) vary greatly.

  Similarly, the diversification of equipment types of network equipment has increased the types of equipment failures and shutdown events, making it difficult to quickly isolate the cause of failure and identify the location of the failure. May be prolonged.

  Regarding these issues, Non-Patent Document 3 mentioned above (Safety / Reliability Standards (draft) corresponding to IP of network), Ministry of Internal Affairs and Communications Information and Communication Council, Information and Communication Technology Subcommittee, IP Network Equipment Committee, Safety and Reliability Described in the Reliability Work Group, December 20, 2007).

  Therefore, when calculating the failure probability of the network using the average value of failure rate and service interruption time due to failure, failure cases with high impact and low occurrence frequency, or failure cases with low impact and high occurrence frequency Cannot properly evaluate the influence of the network on the network and the user who uses the network, and there is a risk that the influence will be underestimated or overestimated.

  At the same time, since the reference value cannot be set in advance for each network facility, the calculated failure probabilities cannot be compared and evaluated. Therefore, the conventional technique has a problem that the failure probability of the network cannot be sufficiently adequately evaluated.

  As described above, failure cases and facilities (equipment models, etc.) that have a large impact on reliability degradation are not clarified due to inability to accurately evaluate network reliability, or failure cases that are highly effective in improving reliability. And the equipment is not clear. As a result, there is a problem that the reliability of the network cannot be reliably implemented.

  And selection and determination of measures for constructing a network with high reliability or reliability desired by the network operation manager, equipment, equipment that requires replacement or repair, equipment that requires spare parts, recovery procedure It has been a big problem of the prior art that it is impossible to reliably select and determine equipment and failure cases that require modification, equipment that should be analyzed for failure causes, and failure cases.

  In order to solve the above problems, the object of the present invention is applicable to a network in which the deployment configuration and the accommodation configuration of network equipment in the network section to be evaluated are not uniform, and a plurality of model devices are sequentially deployed. Network management system and network management method capable of evaluating the reliability of the network in consideration of the influence on the user due to the variation in the failure rate of the network caused by it and the prolonged service interruption at the time of the failure, and for the same Is to provide a program.

In order to achieve the above object, the present invention adopts the following configuration.
(1) In order to solve the above problem (1), the present invention is suitable for calculating the failure probability in addition to information on the type and function of the failed equipment in the information on the failure history collected from the network. The network failure probability is calculated by using other information.

  Here, as information suitable for calculating the failure probability, the present invention uses information on the number of failures that have occurred in the network, the scale of influence due to failure in each failure case, and the service interruption time when the failure occurs.

  By using these pieces of information, it becomes possible to calculate the failure probability of the network even in a network where the arrangement configuration and the accommodation configuration of the network equipment in the network section to be evaluated are not uniform, and the reliability of the network is evaluated.

  In addition, in order to solve the above problem (2), the present invention estimates not only the average characteristic of the failure probability but also the probability distribution regarding the degree of influence of the failure, and evaluates the reliability of the network. Therefore, based on the failure history information collected from the network, the probability of failure impact (probability of failure) based on the total number of failures that occurred in the network and the failure impact of each failure case Estimate the distribution of.

  The degree of influence of a failure is calculated by the product of the scale of influence due to the failure (for example, the number of affected users when a failure occurs) and the service interruption time when the failure occurs. Here, it is possible to determine whether the reliability of the network is good or not by considering the spread as the probability distribution of the influence of the failure on the user by defining and estimating the probability distribution regarding the influence degree of the failure as the network reliability It becomes.

  Further, in the procedure for estimating the reliability of the network from the failure history information of the network, a cumulative probability distribution is calculated in order to easily estimate the failure probability with respect to the influence degree of the failure. Then, an upper cumulative probability distribution is calculated in order to consider the occurrence of cases where the degree of influence of the fault is extremely large. Furthermore, by using the logarithm conversion value for the probability corresponding to the failure influence degree, a parameter that characterizes the probability distribution is simply calculated to estimate the network reliability.

  And by using the estimation result of network reliability, it is possible to determine the network reliability even when the reference value for each network facility cannot be set or the failure probability of each network facility cannot be evaluated individually. .

Therefore, in order to realize the above means, the following functions are provided.
(A) Function for inputting network failure information, determination target information, execution determination reference value, reliability determination reference value (procedure 0)
(B) Function for calculating the total number of failures in the network from failure information (Procedure 1)
(C) A function for determining the necessity of executing the subsequent procedure based on the calculated total number of failures (procedure 2)
(D) Function for calculating the degree of influence of a failure based on the failure time of each failure case and the number of affected users at the time of the failure (Procedure 3)
(E) A function for estimating a probability distribution with respect to the failure influence degree as a network reliability from the calculated total number of failures and the failure influence degree (procedure 4)
(F) Function for determining whether the network reliability is good or not based on the estimation result of the network reliability (Procedure 5)
(G) Function for displaying determination result (procedure 6)

Hereinafter, a more specific configuration of the present invention will be described. (A) A network management system according to the present invention is a network management system that manages a network having a plurality of facilities as components, and includes a failure time for each facility. A first input means for inputting network failure information including the number of affected users, a second input means for inputting determination target information relating to equipment to be subjected to reliability determination, and input by the first input means Stored in the first storage means, the second storage means for storing the determination target information input by the second input means, and the first storage means Means for setting failure information as a determination target for equipment whose reliability is determined by the determination target information among the network failure information; and The failure time of the network failure information and the number of affected users, or the failure influence level of each equipment to be determined is calculated from the normalized failure time of the network failure information and the normalized number of affected users. Calculating means, calculating a failure probability with respect to the degree of influence of the failure calculated by the first calculating means, estimating network reliability based on the failure probability, and network reliability estimated by the estimating means And a display means for displaying the determination result determined by the determination means.

(B) In the network management system, a second calculation unit that calculates the total number of failures determined as a determination target from the network failure information, and the total number of failures calculated by the second calculation unit are given in advance. And a comparison unit that compares the execution determination reference value that is a criterion for determining whether or not the reliability determination execution is required, and as a result of comparison by the comparison unit, when the total number of failures is equal to or greater than the execution determination reference value The processing of the estimation means is executed, and the processing of the estimation means is not executed when the total number of failures is less than the execution determination reference value.

(C) In the network management system, the estimation unit uses an upper cumulative probability as a failure probability with respect to the influence degree of the failure.

(D) In the network management system, the estimating means may calculate an upper cumulative probability with respect to the failure influence degree based on a logarithmic value of the failure influence degree and the upper cumulative probability, or a reference It is calculated based on the logarithmic value of the influence degree of the failure and the upper cumulative probability.

(E) In the network management system, the estimation means uses a result obtained by performing linear regression on the logarithmic value of the normalized failure influence level and the upper cumulative probability, or the standardization parameter In the candidate value, use the logarithmic value of the failure impact and upper cumulative probability normalized by the standardization parameter with the largest linear regression correlation coefficient, or the probability for the regression line of linear regression and the failure impact It is characterized by using a difference from a point, or by setting a point at which a difference between a linear regression line and a probability point with respect to a failure influence degree is a certain value or more as a reference point.

  The program according to the present invention is a program for causing a computer to function as each means in the network management system.

  The network management method according to the present invention is a network management method for managing a network having a plurality of facilities as a component, and is a first input for inputting network failure information including a failure time and the number of affected users for each facility. A procedure, a second input procedure for inputting determination target information relating to equipment to be subjected to reliability determination, a first storage procedure for storing network failure information input by the first input procedure, A second storage procedure for storing the determination target information input by the second input means; and execution of reliability determination by the determination target information among the network failure information stored in the first storage procedure. A procedure for setting failure information for a target facility as a determination target, a failure time of the network failure information and the number of affected users, or A first calculation procedure for calculating the degree of influence of failure for each facility to be determined from the normalized failure time of the network failure information and the normalized number of affected users, and calculation by the first calculation procedure A failure probability with respect to the degree of influence of the determined failure is calculated, a network reliability is estimated based on the failure probability, and a network reliability estimated by the estimation procedure is determined based on a criterion value input in advance. A determination procedure; and a display procedure for displaying a determination result determined by the determination procedure.

  According to the present invention, by adopting the above configuration, the deployment configuration and the accommodation configuration of the evaluation network equipment, the variation in the failure rate due to the deployed device model, and the length of service interruption time when the failure occurs By taking into account the influence on the user, it becomes possible to realize a network management technique capable of accurately evaluating the reliability of the network.

  In addition, it is possible to perform highly reliable reliability determination by setting an execution determination reference value and executing network reliability estimation / determination only when the execution determination reference value is greater than or equal to the execution determination reference value. Therefore, useless processing that is not statistically reliable can be omitted.

  Then, by determining failure cases and network equipment that have a large impact on network reliability degradation, and selecting cases and devices that have a large impact on network reliability improvement measures, the operation management of the network is highly reliable. It becomes possible to manage a network having the reliability desired by a person.

  Here, in the present invention, since it has a function of estimating the reliability of the network from the failure information of the network equipment, based on the quantitative reliability evaluation, the device and the cause that have a large influence on the decrease in the reliability of the network It is possible to clarify.

  Also, by expressing the reliability of the network as the degree of failure impact and failure probability, it is possible to select measures that should be preferentially implemented and measures that should be taken preventively, or to avoid erroneous decisions. It is possible to reliably improve the reliability of the network.

  In addition, by determining the reliability of the network quantitatively, in network operation management, in order to quickly determine recovery measures in the event of a failure, and to determine failure countermeasures that reduce the impact on the user Indicators can be given.

It is a figure which shows an example of the processing flow (processing procedure) of the network management system which concerns on this invention. It is a figure which shows one Embodiment of the structure of the network management system which concerns on this invention, and the flow of its processing. It is a figure which shows an example of the flowchart which implements determination of reliability in an embodiment of this invention, and an example of the step of calculation. It is a figure which shows an example of the network failure information used in order to demonstrate the process in embodiment of this invention. It is a figure which shows an example of the failure information in the network failure database (DB) used in order to demonstrate the process in embodiment of this invention. It is a figure which shows the example of calculation of the influence degree of a failure used in order to demonstrate the process in embodiment of this invention (when the product of failure time and the number of affected users is defined as a failure impact degree). It is a figure which shows the example of the influence degree of the failure arranged in the ascending order used in order to demonstrate the process in embodiment of this invention (when the product of failure time and the number of affected users is defined as failure influence degree). It is a figure which shows the example of calculation of the probability point used in order to demonstrate the process in embodiment of this invention. It is a figure which shows the probability point and regression line on a log-log graph. It is a figure which shows the example of calculation of the parameter a and correlation coefficient used in order to demonstrate the process in embodiment of this invention. It is a figure which shows an example of the network used as the implementation object of this invention.

(Outline of the present invention)
First, a network configuration targeted by the present invention will be described.
FIG. 11 is a diagram illustrating a network configuration targeted by the present invention, and includes a control network 100, a core network 200, an access network 300, and an end user 400.

  It is assumed that the end user 400 uses the Internet access and various IP (Internet protocol) services. A failure in a configuration where the network element is not made redundant leads to service interruption, so it is highly likely that the service usage of the user will be affected. As a simple example, for example, the access network 300 is similar as shown in the figure. Or a component inside the facility (device A301, device B302 to device B303, device C304 to device C307) can be described as a model configured in a tree shape.

  Although a failure in a redundant configuration location such as the core network 200 is unlikely to lead to service interruption, service interruption due to switching or service interruption due to switching function failure may occur.

  Therefore, in order to evaluate the reliability of the network, it is necessary to consider all the failure cases that caused service interruption, but a huge failure history collected from a large-scale network that changes over time In the case of data, it is not always easy to classify and evaluate comparative points and factors.

  In view of the above, the present invention considers the above points, arranges all failure cases side by side, grasps the failure characteristics, and performs a quantitative evaluation. Failure impact level (= failure time x impact) based on date, failure recovery date, failure time (failure duration = failure recovery date-failure occurrence date), number of affected users (range affected by failure (people)), etc. The number of users), and the probability of the failure impact is estimated, and the reliability of the network is comprehensively evaluated by defining the probability of failure and the difficulty of the failure as the network reliability. It is what you do.

(Embodiment)
Hereinafter, embodiments of a network management system according to the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram showing an example of a processing flow (processing procedure) of a network management system according to the present invention.

  As shown in the figure, the processing procedure in the present invention includes a procedure for inputting network failure information and determination target information (procedure 0), a procedure for calculating the total number of failures (procedure 1), and whether or not to perform reliability determination. (Procedure 2), a procedure for calculating the degree of influence of a failure (Procedure 3), a procedure for estimating network reliability (Procedure 4), a procedure for determining network reliability (Procedure 5), and a determination result It consists of the procedure (procedure 6) which displays.

  FIG. 2 is a diagram showing an embodiment of the configuration of the network management system according to the present invention and the processing flow thereof, and also describes (procedure 0) to (procedure 6) in FIG.

  In the figure, 1 is network failure information input to the network management system in (procedure 0), 10 is a network management system according to the present invention, 11 is a receiver, and 12 is a network failure information DB for storing network failure information ( (Database), 13 is a total failure number calculation unit that executes procedure 1, 14 is an execution determination unit that executes (procedure 2), 15 is an influence calculation unit that executes (procedure 3), and 16 is (procedure 4). Network reliability estimation unit to be executed, 17 is a storage unit that stores the network reliability estimation value, 18 is a network reliability determination unit that executes (Procedure 5), 19 is a storage unit that stores determination target information, and 3 is a determination An input unit (input unit) for inputting target information, 4 is a display unit (display unit) for executing (Procedure 6), and the network determined in (Procedure 5). To display the over-click reliability determination result.

  The timing of execution of the processing of the network management system 10 in the present invention may be set in advance when a fixed period is set in advance, or may be executed by inputting an arbitrary period. In any case, the embodiment is the same. Therefore, in the following, the present invention is assumed to be repeatedly executed every fixed period, and a case where the execution timing is set in advance will be described as an example.

  Hereinafter, the processing in each procedure is associated with each step (step S10) to (step S50) in the flowchart of processing in the network management system according to the present invention shown in FIG. 3, and an example of the network failure information shown in FIG. 5, an example of the failure information in the network failure database (DB) shown in FIG. 5, an example of the calculation of the degree of influence of the failure shown in FIG. 6, an example of the degree of influence of the failure arranged in ascending order shown in FIG. 7, and the probability shown in FIG. This will be described in detail with reference to a calculation example of points, an explanatory diagram of probability points and regression lines on the log-log graph shown in FIG. 9, and a calculation example of the parameter a and the correlation coefficient shown in FIG.

(Procedure 0) <Input of network failure information and judgment target information>
This procedure is automatically executed in the present network management system 10 or is executed in advance by a network operation manager, and is described separately from the following procedures (procedure 1 to procedure 6).

  The network failure information 1 and the determination target information are input to the network management system 10 from the outside. The network failure information 1 is stored in a database (DB) (not shown) outside the corresponding system and is input to the receiving unit 11 of the network management system 10.

  As shown in FIG. 4, the network failure information 1 includes failure equipment, that is, a failure device (failed device model), failure occurrence date / time, failure recovery date / time, failure time (minutes), number of affected users (people), failure location / Contains information about the cause.

  Here, the failure time (minutes) is the time during which communication is interrupted due to a failure, and is defined by the difference (in minutes in this example) from the failure occurrence date and time to the failure recovery date and time, and is input by the network operation manager. . The number of affected users (persons) is defined by the number of failed devices accommodated.

  The network failure information 1 input to the receiving unit 11 is stored in the network failure DB 12.

  In addition, information related to equipment (generic name including equipment / device) that is a target of execution of reliability determination is input to the network management system 10 via the input unit 3 as determination target information. The determination target information input from the input unit 3 is temporarily stored in the storage unit 19.

  With reference to the determination target information stored in the storage unit 19, the failure information of the facility to be determined is set as the determination target information for the failure information input in the network failure information DB 12. Other failure information is excluded as information that is not subject to subsequent procedures.

  FIG. 5 is a diagram illustrating an example of network failure information stored in the network failure information DB 12. As shown in the figure, failure information in the network failure information DB 12 includes failure occurrence date / time, failure recovery date / time, failure time (minutes), number of affected users (people), failure for each failure facility, that is, failure device (device model). It consists of each item of location / cause. The failure time (minutes) is calculated by (failure recovery date / time) − (failure occurrence date / time) (unit: minutes).

(Procedure 1) <Processing by Total Failure Number Calculation Unit 13>
In the procedure 1, the total failure number calculation unit 13 calculates the number of cases (record number) in the network failure information DB 12 as the total failure number for the failure information set as the determination target (in step S10 of FIG. 3). Correspondence). Here, the total number of failures is N (total failure count calculation value).

(Procedure 2) <Processing of Execution Determination Unit 14 (Determining Necessity of Execution of Reliability Determination (Subsequent Procedure))>
In (Procedure 2), on the basis of the calculation result of the total number of faults, the execution determination unit 14 compares the total number of faults with a predetermined execution determination reference value by using a comparison unit. (Corresponding to step S20 in FIG. 3). For the predetermined execution determination reference value, the network operation manager may input an arbitrary value in advance from the input unit 3. From the standpoint of statistical advantage, about several tens of cases are necessary, and for example, a value of about 50 may be set as the execution determination reference value.

  The total number of failures calculated in (Procedure 1) and the execution determination reference value are compared by a comparison means. If (total number of failures) ≧ (execution determination reference value), the total number of failures from a statistical viewpoint Is determined to be large (step S20: Y in FIG. 3), and the process proceeds to the subsequent steps. If (total number of failures) <(execution determination reference value) (step S20: N in FIG. 3), it is determined that the total number of failures is small from a statistical point of view, and the subsequent (procedure 4) is executed. Without returning to step S10.

(Procedure 3) <Processing of Failure Influence Level Calculation Unit 15>
In the execution determination unit 14 (procedure 2), when it is determined that it is necessary to execute reliability determination with (total number of failures) ≧ (execution determination reference value) (step S20 in FIG. 3: Y), in this (procedure 3), Then, the influence degree calculation unit 15 calculates the influence degree of the failure (step S30 in FIG. 3).

There are the following two methods for calculating the degree of influence of a failure.
(Method 1) “When the product of the failure time and the number of affected users is defined as the failure impact level”
Numbers i are assigned to failure cases in order of the time and date of failure occurrence. The failure time (minutes) of each failure case is di, the number of users affected by the failure (people) is ri, and the failure influence degree (minutes × people) hi for the i-th failure case is the failure time (minutes) di and the affected users. It is calculated as the product di × ri of the number (person) ri (hi = di × ri). FIG. 6 is a diagram illustrating an example of a calculation result of the failure influence degree when the product of the failure time and the number of affected users is defined as the failure influence degree.

(Method 2) “When the product of the normalized failure time and the normalized number of affected users is defined as the failure impact level”
For the failure time and the number of affected users, use the values normalized by the respective statistical values (normalized failure time and normalized number of affected users), and define the product as the impact level of the failure in the same way as above. calculate.

  The failure influence degree hi ′ is defined and calculated as di′ri ′ using the normalized failure time di ′ and the normalized influence user number ri ′. At this time, as statistical values used for normalization, there are candidates such as respective average values, median values, maximum values, minimum values, quantile values (quartile values, etc.).

  Usually, in a network operation management system, failure time information is collected in units of minutes, and information on the number of affected users is collected in units of people. Therefore, the degree of influence of failure is calculated in units of [minutes / people].

  Since the subsequent procedures are common to both the method 1 and the method 2, they are described according to the description and symbols of the method 1.

(Procedure 4) <Processing of Network Reliability Estimation Unit 16 (Estimation of Network Reliability)>
In step 4, the network reliability estimation unit 16 estimates the network reliability based on the total number of failures calculated in step 2 and the degree of failure influence calculated in step 3 (step in FIG. 3). S40). The reliability of the network is a degree affected by equipment failure when using the network, and is defined as a failure probability with respect to the influence degree of the failure.

  As a method for estimating the distribution characteristic of failure probability related to the failure influence level, here, the upper cumulative distribution (in order to consider a region having a large influence degree and a small probability, if the failure influence degree h is set to an arbitrary h, On the other hand, estimation is performed on a probability distribution in which probabilities exceeding h are accumulated.

(4-0)
In order to estimate the network reliability, a parameter bj for standardizing the influence degree of the failure is introduced. A reasonable estimation result can be obtained by normalizing the degree of influence of the failure. Since it is necessary to select an appropriate value as the standardization parameter bj, it is necessary to set a candidate value and evaluate its validity.
The procedure for setting candidate values for the standardization parameter bj is shown below.

There are the following two methods (a) and (b) as setting methods.
(A) The maximum value of the candidate values of the standardization parameter bj is the average value of hk, the minimum value of the candidate values of the standardization parameter bj is 1, and the sequence from the minimum value to the maximum value is the candidate value of the standardization parameter bj Set as.
At this time, if the average value of hk does not become an integer, the integer value is set by rounding off.

As a method for setting the number sequence from the minimum value to the maximum value, for example, the following three methods (A) to (C) are conceivable.
(B) Change the numerical value by 1 from the minimum value to the maximum value.
(B) From the minimum value to the maximum value, 1/10 of the difference between the maximum value and the minimum value is changed as the range of change (in this case, the standardization parameter bj is 10).
(C) From the minimum value to the maximum value, 1/50 of the difference between the maximum value and the minimum value is changed as the range of change (in this case, the standardization parameter bj is 50).

(B) The maximum value of the candidate values of bj is the quantile value of hk, the minimum value of the candidate values of the standardization parameter bj is 1, and the sequence from the minimum value to the maximum value is set as the candidate value of the standardization parameter bj To do. At this time, if the quantile value of hk is not an integer, the integer value is set by rounding off.

  The third quartile value (75% value) or 90% value is used as a candidate for the quartile value. The method for setting the number sequence from the minimum value to the maximum value is the same as (a) above.

(4-1)
The initial value bj = b0 of the standardization parameter bj is set.

(4-2)
Next, the influence levels of failures are arranged in ascending order, number k is assigned in ascending order, and hk (k = 1... N) is set. FIG. 7 is a diagram showing a calculation example of the failure influence degree obtained in this way (failure influence degree (minute × person) hk = failure time (minute) dk × number of affected users (person) rk). .

(4-3)
From the above hk, N, bj, k, N (k = 1... N) probability points K (1 + hk / bj, (N−k + 1) / N) corresponding to the degree of influence of the failure are calculated. FIG. 8 is a diagram showing a specific calculation example of the probability point K (1 + hk / bj, (N−k + 1) / N) obtained in this way, and the influence degree (minute × person) hk of the failure. A specific example of normalized failure influence degree 1 + hk / bj and probability (N−k + 1) / N is shown (when N = 100 and bj = 100).

(4-4)
There is a high possibility that the degree of influence of a failure is widely distributed due to a longer failure time and a larger number of affected users. In addition, since there are many faults with a low degree of influence and few faults with a high degree of influence, the probability distribution characteristic regarding the degree of influence of the fault has a logarithmic distribution.

  Therefore, logarithmic conversion is performed on the probability of the failure and the corresponding probability. The probability point K (1 + hk / bj, (N−k + 1) / N) is plotted on a log-log graph as shown in FIG. 9, and a regression line is drawn. Here, the common logarithm is used as the logarithm setting.

(4-5)
An absolute value aj of the slope of the regression line and a correlation coefficient Rj for the probability point of the line are calculated. (Regression lines and correlation coefficients are well known and will not be described in detail (see FIG. 9)).

(4-6)
In the above procedure, j is changed to j = 1... J, and as shown in FIG. 10, the absolute value aj of the slope of the regression line corresponding to the standardization parameter bj and the correlation coefficient Rj for the probability point of the line. Is calculated.

(4-7)
As the square value of the correlation coefficient with respect to the regression line is closer to 1, the absolute value aj of the slope of the regression line and the standardization parameter bj are more appropriate values as estimated values. Therefore, j, which maximizes the square value Rj ² of the correlation coefficient, the absolute value aj of the slope of the regression line, and the standardization parameter bj are selected. In the case of the example of FIG. 10, j where the correlation coefficient Rj is maximum is 3, the standardization parameter bj at that time is 101, the absolute value aj of the slope of the regression line is 1.0, and the correlation coefficient Rj is 0. .99.

  The straight line determined by the absolute value aj of the slope of the regression line obtained in this way and the standardization parameter bj is the estimation result of the upper cumulative probability related to the influence degree of the failure, and is the estimation result of the network reliability.

  However, in the region where the degree of influence is large, the estimation result may deviate from (plotted probability points), and as the logarithmic graph proceeds in the horizontal axis direction, the probability point sequence may deviate from the straight line. .

  When the degree of divergence is large, the estimation result of the network reliability is not valid, and there is a possibility that the accuracy is lowered in the determination of the reliability. Therefore, correction of network reliability is shown in the following procedure. (The following procedures (4-8) to (4-10) are recommended procedures, but it is possible to provide the network management system according to the present invention even when not implemented.)

(4-8)
The absolute value aj of the slope of the regression line that maximizes the correlation coefficient Rj is set to ax, and using the value of ax, for k = 1... N, | log (1-ax · hk ′ ) −log {(N−k + 1) / N} | is calculated as N number sequences. Here, “| a |” is an absolute value of a. Further, hk ′ = 1 + hk / bj.

(4-9)
| log (1-ax · hk ′) − log {(N−k + 1) / N)} |
For k, update k sequentially from k = 1 and
| log (1-ax · hk) −log {(N−k + 1) / N)} |> c
Is selected first (k is the smallest k).

  Here, c is a fixed value set in advance. Since the difference between the common logarithmic values is evaluated, the value of c may be set to 1 or 0.5.

  When k satisfying the above magnitude relationship does not exist, k = N is set (meaning that the network reliability estimation result is not corrected).

(4-10)
Let the probability point K (1 + hk / bj, (N−k + 1) / N) corresponding to k in (4-9) be x and x (1 + hx / bj, px). Here, hk = hx, (N−k + 1) / N = px.
The values of hx and px calculated in this way are used as the corrected network reliability estimation results.

(Procedure 5) <Processing of Network Reliability Determination Unit 18 (Determination of Network Reliability)>
In procedure 5, based on the reliability estimation result transmitted from the network reliability estimation unit 16, the network reliability determination unit 18 performs network reliability determination (step S50 in FIG. 3).

  As a method for determining reliability, a method (determination method 1) for performing determination on a pair of probabilities and corresponding probabilities of failure, and an expected value calculated by the product of the probabilities of failure and probabilities There are two methods: a method for performing the determination (determination method 2).

(Judgment method 1)
When the determination is performed on the combination of the failure influence degree and the probability, there are the following four determination results (a) to (d).

(A)
(Hx) <(determination criterion value of influence degree) and (px) <(determination criterion value of probability)
(B)
(Hx) ≧ (determination criterion value of influence degree) and (px) ≧ (determination criterion value of probability)
(C)
(Hx) ≧ (determination criterion value of influence degree) and (px) <(determination criterion value of probability)
(D)
(Hx) <(determination criterion value of influence degree) and (px) ≧ (determination criterion value of probability)

(Judgment method 2)
When the reliability determination is performed based on the expected value: hx · (1−px) calculated by the product of the influence level and the probability of failure, there are the following two determination results (a) to (b).
(A) hx · (1-px) <(reliability determination reference value)
(B) hx · (1-px) ≧ (reliability determination reference value)

  Here, the network operation manager inputs an arbitrary value in advance as the determination reference value. As an example of setting the influence criterion value, for example, when measuring the failure time in minutes, a value of about 100,000 is set. In addition, as an example of setting the probability criterion value, a value of about 0.01 is set. Similarly, a value of about 1000 is set as an example of the reliability criterion value.

  Further, when the results (b), (c), and (d) are obtained in the determination method 1 or when the result (b) of the determination method 2 is obtained, the failure case corresponding to the probability point x is determined as the reliability. Judge as the failure case that has the greatest impact on the decline.

  Similarly, in a failure case corresponding to a probability point located on the left side with respect to the probability point x, that is, k satisfying hk <hx, it is determined as a failure case having a greater influence degree in descending order of k.

  Similarly, a probability point located on the right side of the probability point x, that is, a failure case corresponding to k satisfying hk> hx is determined as a unique case.

(Procedure 6) <Display of determination result on display unit 4>
In step 6, the determination result output by the network reliability determination unit 18 in the above (procedure 5) is displayed on the display unit 4.

  The evaluation results displayed on the display unit 4 include the items shown in the following (a) to (e), and any of these items or a combination of some items may be used.

(A) Display whether the set of network reliability is within the reference value.
For example, in the case of (a) of the determination method (1), “◯” is displayed, in the case of (b) of the determination method (1), “Δ” is displayed.

(B) Information on whether the expected value of the network reliability is within the reference value is displayed on the display unit 4.
For example, “◯” is displayed in the case of (a) of the determination method (2), and “X” is displayed in the case of (b) of the determination method (2).

(C) Information on the value of the probability point x (hx, px) and the corresponding failure example (information such as the name of the failure facility, the location name, the date and time of failure occurrence) is displayed on the display unit 4.

(D) Information on M (for example, 5) probability points (hk, pk) located on the left side of the probability point x (hx, px) and information on the corresponding failure case (failure equipment name, location name, failure occurrence) Information such as date and time) is displayed on the display unit 4.

(E) Displays the value of the probability point (hk, pk) located on the right side of the probability point x (hx, px) and information about the corresponding failure case (information such as the name of the failure equipment, location name, failure occurrence date and time) Part 4 is displayed.

  Note that the processes (each procedure in FIG. 1, each step in FIG. 3) and functions performed by each unit (each unit) of the network management system 10 shown in FIG. This is realized by executing a program corresponding to processing executed by each unit (each unit) using hardware resources such as a built-in CPU and memory. Further, the program can be distributed to the market via a recording medium such as FD, CD-ROM, DVD, or a network such as the Internet.

1: Network failure information 3: Input unit (input means)
4: Display unit for displaying the determination result (display means)
10: Network management system 11: Receiving unit (receiving means)
12: Network failure information DB (database)
13: Total failure number calculation unit (total failure number calculation means)
14: Execution determination unit (execution determination means)
15: Influence calculation unit (impact calculation means)
16: Network reliability estimation unit (network reliability estimation means)
17: Storage unit for storing estimated network reliability value 18: Network reliability determination unit (network reliability determination means)
19: Storage unit for storing determination target information 100: Control network 200: Core network 300: Access network 301: Apparatus A
302 to 303: Device B
304 to 307: Device C

Claims (10)

A network management system for managing a network having a plurality of facilities as a component,
First input means for inputting network failure information including the failure time and the number of affected users for each facility;
A second input means for inputting determination target information relating to equipment to be subjected to reliability determination;
First storage means for storing the network failure information input by the first input means;
Second storage means for storing the determination target information input by the second input means;
Means for setting, as a determination target, failure information for a facility that is subjected to reliability determination by the determination target information among the network failure information stored in the first storage unit;
The failure time and the number of affected users in the network failure information or the normalized failure time and the normalized number of affected users in the network failure information are used to calculate the failure impact level for each facility to be determined. 1 calculating means;
Estimating means for calculating a failure probability with respect to the degree of influence of the failure calculated by the first calculating means, and estimating network reliability based on the failure probability;
A determination unit that determines the network reliability estimated by the estimation unit based on a determination criterion value that is input in advance;
Display means for displaying the determination result determined by the determination means;
A network management system comprising: The network management system according to claim 1,
A second calculating means for calculating the total number of failures determined from the network failure information;
Comparing means for comparing the total number of failures calculated by the second calculating means with an execution determination reference value serving as a reference for necessity of reliability determination execution given in advance,
As a result of the comparison by the comparison means, when the total number of failures is equal to or greater than the execution determination reference value, the processing of the estimation means is executed, and when the total number of failures is less than the execution determination reference value, the estimation means A network management system characterized in that the above processing is not executed. The network management system according to claim 1 or 2,
The network management system according to claim 1, wherein the estimation means uses an upper cumulative probability as a failure probability with respect to the degree of influence of the failure. The network management system according to claim 3,
The estimation means calculates the upper cumulative probability for the failure influence level based on a logarithmic value of the failure influence degree and the upper cumulative probability, or a standardized failure influence level and upper cumulative probability pair. A network management system characterized by calculation based on numerical values. The network management system according to claim 4, wherein
The network management system according to claim 1, wherein the estimation means uses a result obtained by performing linear regression on a logarithmic value of the normalized failure influence level and upper cumulative probability. The network management system according to claim 5,
The estimation unit uses a logarithmic value of an influence degree of failure and an upper cumulative probability normalized by a standardization parameter having a largest correlation coefficient of linear regression among candidate values of the standardization parameter. system. The network management system according to claim 5,
The estimation unit uses a difference between a regression line of linear regression and a probability point with respect to the degree of influence of a failure. The network management system according to claim 7,
The network management system characterized in that the estimation means sets a point at which a difference between a linear regression line and a probability point with respect to a failure influence degree is a certain value or more as a reference point.   The program for functioning a computer as each means in the network management system in any one of Claim 1 to 8. A network management method for managing a network having a plurality of facilities as a component,
A first input procedure for inputting network failure information including a failure time for each facility and the number of affected users;
A second input procedure for inputting determination target information related to equipment to be subjected to reliability determination;
A first storage procedure for storing the network failure information input by the first input procedure;
A second storage procedure for storing the determination target information input by the second input means;
A procedure for setting, as a determination target, failure information for a facility to be subjected to reliability determination by the determination target information among the network failure information stored in the first storage procedure;
The failure time and the number of affected users in the network failure information or the normalized failure time and the normalized number of affected users in the network failure information are used to calculate the failure impact level for each facility to be determined. 1 calculation procedure;
An estimation procedure for calculating a failure probability with respect to an influence degree of the failure calculated by the first calculation procedure, and estimating a network reliability based on the failure probability;
A determination procedure for determining the network reliability estimated by the estimation procedure based on a determination criterion value input in advance,
A display procedure for displaying the determination result determined by the determination procedure;
A network management method comprising:

Images