JP6890109B2 - Information generator, information generation system and program - Google Patents

Description

Embodiments of the present invention relate to information generators, information generation systems and programs.

A fault simulator technique for generating a pseudo fault in a communication system between devices is disclosed (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2011-123783

However, according to the prior art as described in Patent Document 1, it is necessary to change the command execution procedure and the execution content for causing a pseudo failure according to, for example, the configuration of the target device, and the operation man-hours are required. There was a problem that it was very large.

An object of the present invention is to provide an information generation device, an information generation system and a program capable of reducing operation man-hours.

In one embodiment of the present invention, information indicating the type of network failure to occur in the network to be learned and network parameters related to the network to be learned are input data, and a task for operating the network is defined. The output data is a workflow that combines one or more of the tasks selected from the definition information, and the presence or absence of a network failure that occurs as a result of giving an operation based on the workflow to the network is used as training data, and the type of the network failure and Acquisition of the trained model in which the relationship between the network parameters and the workflow is learned, the failure type information indicating the type of network failure to occur in the network to be inspected, and the network parameters related to the network to be inspected, respectively. The trained model outputs the unit, the supply unit that supplies the failure type information and the network parameters acquired by the acquisition unit to the trained model, and the input data supplied from the supply unit. An information generation device including a selection unit that selects one or more of the above workflows based on the occurrence probability of a network failure of the type indicated by the failure type information, and an output unit that outputs the workflow selected by the selection unit. Is.

In the information generation device of the embodiment of the present invention, the trained model outputs a plurality of the workflows to the selection unit, and the selection unit is among the plurality of workflows output by the trained model. , Select the workflow having a relatively high probability of occurrence of the type of network failure indicated by the failure type information.

In the information generator according to the embodiment of the present invention, a failure determination policy in which the relationship between the situation generated as a result of giving the network an operation based on the workflow output by the trained model and the presence / absence of a network failure is defined. Based on the above, the determination unit that determines whether or not a network failure has occurred, the workflow, and the determination result that the determination unit determines the situation that occurs as a result of giving an operation based on the workflow to the network are stored in association with each other. The trained model is further provided with a storage unit, and the trained model is trained by using the teacher data as the presence or absence of a network failure determined based on the failure determination policy.

In the information generation device of the embodiment of the present invention, the trained model is relearned by using the correspondence between the workflow and the determination result stored in the storage unit as the teacher data.

In one embodiment of the present invention, the above-mentioned information generator, a workflow control unit that gives an operation based on the workflow output by the output unit to the network, and a situation that occurs in the network are monitored and the results of monitoring are obtained. It is an information generation system including a network monitoring unit that outputs to the determination unit of the information generation device.

In one embodiment of the present invention, the computer has an acquisition step of acquiring failure type information indicating the type of network failure to occur in the network to be inspected and network parameters related to the network to be inspected, and a network to be learned. Information indicating the type of network failure to be generated and network parameters related to the network to be learned are used as input data, and one or more of the tasks selected from the task definition information in which the task to operate the network is defined are combined. The relationship between the type of network failure, the network parameters, and the workflow is learned by using the above workflow as output data and using the presence or absence of network failure that occurs as a result of giving an operation based on the workflow to the network as training data. The trained model outputs the trained model based on the supply step that supplies the failure type information and the network parameters acquired in the acquisition step and the input data supplied in the supply step. A program for executing a selection step of selecting one or more of the workflows based on the occurrence probability of a network failure of the type indicated by the failure type information, and an output step of outputting the workflow selected in the selection step. Is.

According to the present invention, it is possible to provide an information generation device, an information generation system and a program capable of reducing operation man-hours.

It is a figure which shows an example of the functional structure of the information generation system of this embodiment. It is a figure which shows an example of the network parameter of this embodiment. It is a figure which shows an example of the task definition information of this embodiment. It is a figure which shows an example of the workflow of this embodiment. It is a figure which shows an example of the functional structure of the information generation apparatus of this embodiment. It is a figure which shows an example of the operation in the learning stage of the information generation system of this embodiment. It is a figure which shows an example of the operation in the utilization stage of the information generation system of this embodiment. It is a figure which shows an example of the relearning operation of the information generation system of this embodiment.

[Embodiment]
Hereinafter, an outline of the information generation system 1 of the present embodiment will be described with reference to the drawings.
FIG. 1 is a diagram showing an example of the functional configuration of the information generation system 1 of the present embodiment.

[Overview of information generation system]
The information generation system 1 generates a workflow WF related to the network N based on various conditions, and executes an operation based on the generated workflow WF on the network N. In addition, the information generation system 1 monitors the status of the network N caused by the result of executing the operation based on the workflow WF.
The workflow WF referred to here includes a "command CMD", a "column of a plurality of command CMDs", a "task TSK defining a sequence of command CMDs", a "column of a plurality of task TSKs", and the like. Further, the workflow WF includes not only the above-mentioned instructions for logical operations such as command CMD and task TSK, but also instructions for causing a physical failure such as disconnection or short circuit of the wiring constituting the network N. It may be.

In an example of this embodiment, the information generation system 1 generates a workflow WF for causing a network failure D in the network N. Examples of types of network failure D include delay in response, lack of response (no response), synchronization failure between components, loss of data to be communicated, and resource exhaustion due to process contention. The types of network failure D include, for example, a database failure built on a node (for example, a server) connected to network N, a switching failure when a redundant system is configured, a server operating system, and the like. There is an application failure.
Generally, in the network N, the workflow WF for generating a certain kind of network failure D changes according to the situation of the network N. The information generation system 1 generates a workflow WF based on the type of network failure D to be generated and the status of the target network N to generate network failure D.

[Functional configuration of information generation system]
Hereinafter, the functional configuration of the information generation system 1 will be described. The information generation system 1 includes an information generation device 10, an input / output device 20, a task definition server 30, a workflow control unit 40, and a network connection device 50.

The input / output device 20 is composed of, for example, a personal computer or the like, and functions as a user interface device for a user who operates the information generation system 1. The input / output device 20 is connected to the information generation device 10 and supplies various information based on the user's operation to the information generation device 10. Further, the input / output device 20 presents (for example, screen display or printing) various information output by the information generation device 10 to the user.
In this example, the input / output device 20 supplies the failure type information DI and the network parameter NP to the information generation device 10. Here, the failure type information DI is information indicating the type of network failure D generated in the network N. The network parameter NP is information indicating the status of the target network N that causes the network failure D. An example of the network parameter NP will be described with reference to FIG.

FIG. 2 is a diagram showing an example of the network parameter NP of the present embodiment. In the network parameter NP in this example, an index (Index) for identifying the network parameter NP, a unit (Unit), and a value range (Value range) are associated with each other.

Returning to FIG. 1, the task definition information TI is stored in the task definition server 30. The task definition information TI is information that defines a task TSK that operates the task TSK network. An example of the task definition information TI will be described with reference to FIG.

FIG. 3 is a diagram showing an example of the task definition information TI of the present embodiment. In the task definition information TI in this example, the task number that identifies the task TSK and the content of the process to be executed (task TSK) are associated with each other.

Returning to FIG. 1, the information generation device 10 generates a workflow WF based on the failure type information DI and network parameter NP supplied from the input / output device 20 and the task definition information TI supplied from the task definition server 30. To do. The information generation device 10 outputs the generated workflow WF to the workflow control unit 40. Here, an example of the workflow WF generated by the information generation device 10 will be described with reference to FIG.

FIG. 4 is a diagram showing an example of the workflow WF of the present embodiment. In this example, the workflow WF1 indicates that each task TSK should be executed in the order of task number 2, task number 38, and task number 3. Workflow WF2 indicates that each task TSK should be executed in the order of task number 8 and task number 22.

Returning to FIG. 1, the workflow control unit 40 includes a so-called workflow engine, and by interpreting the workflow WF output from the information generation device 10, the command CMD based on the acquired workflow WF is sent to the network connection device 50. And output. That is, the workflow control unit 40 gives the network N an operation based on the workflow WF output by the information generation device 10.

The network connection device 50 executes an operation for the network N based on the command CMD output by the workflow control unit 40. The network connection device 50 includes a network monitoring unit 510. The network monitoring unit 510 monitors the situation occurring in the network N and outputs the monitored result (monitoring result RS) to the information generation device 10.
Next, the functional configuration of the information generation device 10 will be described with reference to FIG.

[Functional configuration of information generator]
FIG. 5 is a diagram showing an example of the functional configuration of the information generation device 10 of the present embodiment. The information generation device 10 includes an acquisition unit 110, a supply unit 120, a learned model 130, a selection unit 140, and an output unit 150.
In the following description, when the network N to be learned and the network N to be inspected are distinguished from each other, they are described as "network NL to be learned" and "network NE to be inspected", respectively.

The acquisition unit 110 acquires various information from the input / output device 20. In an example of this embodiment, the acquisition unit 110 acquires the failure type information DI and the network parameter NP, respectively. The failure type information DI is information indicating the type of network failure D generated in the network NE to be inspected. The network parameter NP is information about the network NE to be inspected.
The acquisition unit 110 outputs the acquired information to the supply unit 120.

The supply unit 120 supplies the failure type information DI and the network parameter NP acquired by the acquisition unit 110 to the trained model 130.

The trained model 130 is a machine learning model obtained by machine learning. In 130 of the present embodiment, information indicating the type of network failure D generated in the network NL to be learned and the network parameter NP related to the network NL to be learned are used as input data, the workflow WF is used as output data, and the workflow WF is used. This is a machine learning model in which the relationship between the type of network failure D, the network parameter NP, and the workflow WF is learned by using the presence or absence of the network failure D that occurs as a result of giving the based operation to the network N as the teacher data TD. As described above, the workflow WF referred to here is a combination of one or more task TSKs selected from the task definition information TI in which the task TSKs that operate the network are defined.

The selection unit 140 selects one or more workflow WFs output by the trained model 130 based on the input data supplied from the supply unit 120 based on the occurrence probability P of the type of network failure D indicated by the failure type information DI. To do.

Here, the trained model 130 may be configured to output a plurality of workflow WFs to the selection unit 140. When configured in this way, the selection unit 140 selects a workflow WF having a relatively high probability P of occurrence of a network failure D of the type indicated by the failure type information DI among a plurality of workflow WFs output by the trained model 130. select.

The output unit 150 outputs the workflow WF selected by the selection unit 140. The output unit 150 of this embodiment outputs the workflow WF to the workflow control unit 40.

[Example of functional configuration modification of information generator 10]
As shown in FIG. 5, the information generation device 10 may further include a determination unit 160 and a storage unit 170.
The determination unit 160 determines whether or not a network failure D has occurred based on the failure determination policy PCY. Here, the failure determination policy PCY is information that defines the relationship between the situation that occurs as a result of giving the network N an operation based on the workflow WF output by the learned model 130 and the presence / absence of the occurrence of the network failure D.
The storage unit 170 stores the workflow WF and the determination result determined by the determination unit 160 as a result of giving the network N an operation based on the workflow WF in association with each other.
In the information generation device 10 configured as described above, the trained model 130 is trained by using the teacher data TD to determine whether or not a network failure D, which is determined based on the failure determination policy PCY, has occurred. That is, the trained model 130 is relearned based on the determination result of the determination unit 160.

More specifically, the trained model 130 is relearned by using the correspondence relationship between the workflow WF stored in the storage unit 170 and the determination result as the teacher data TD.

[Operation of information generation system (learning stage)]
Next, an example of the operation of the information generation system 1 will be described with reference to FIGS. 6 to 8.
FIG. 6 is a diagram showing an example of an operation in the learning stage of the information generation system 1 of the present embodiment. In the following description, a machine learning model that has not completed learning (before or during learning) is described as a learning model ML, and a machine learning model that has completed learning is described as a trained model M. The trained model 130 described above is an example of the trained model M.

(Step S110) The information generation device 10 acquires the failure type information DI and the network parameter NP from the input / output device 20.
(Step S120) The information generation device 10 acquires the task definition information TI from the task definition server 30.
(Step S130) The information generation device 10 supplies each of the information acquired in steps S110 and S120 to the learning model ML. As a result, the training model ML generates a workflow WF.
(Step S140) The information generation device 10 supplies the workflow WF generated in step S130 to the workflow control unit 40. The workflow control unit 40 performs an operation based on the supplied workflow WF to the network N via the network connection device 50.

(Step S150) The network monitoring unit 510 of the network connection device 50 monitors the status of the network N in which the operation based on the workflow WF is executed in step S140. The information generation device 10 acquires the result of monitoring the status of the network N by the network monitoring unit 510 (monitoring result RS).
(Step S160) The information generator 10 evaluates the monitoring result RS acquired in step S150. Specifically, the information generation device 10 evaluates whether or not the network failure D of the type indicated by the failure type information DI acquired in step S110 has occurred in the network N. When the network failure D of the type indicated by the failure type information DI acquired in step S110 occurs in the network N, the information generation device 10 evaluates that the workflow WF is correctly generated. Further, the information generation device 10 evaluates that the workflow WF is erroneously generated when the network failure D of the type indicated by the failure type information DI acquired in step S110 does not occur in the network N. .. Here, when the network failure D of the type indicated by the failure type information DI does not occur in the network N, the network failure D has occurred in the network N even though the operation based on the workflow WF has been executed. This includes the case where there is no network failure D and the case where a network failure D of a type different from the type indicated by the failure type information DI occurs.

(Step S170) The information generation device 10 reflects the evaluation result ER in step S160 as teacher data in the training model ML.
(Step S180) The information generator 10 is based on whether or not the learning result of the training model ML is stable (that is, whether or not the learning is sufficiently performed until the output of the machine learning model can be utilized). To determine whether or not learning is continued. When the learning result of the learning model ML is stable (step S180; YES), the information generation device 10 ends the learning process. When the learning result of the learning model ML is not yet stable (step S180; NO), the information generator 10 returns the process to step S110 and continues the learning process.

[Operation of information generation system (utilization stage)]
FIG. 7 is a diagram showing an example of operation in the utilization stage of the information generation system 1 of the present embodiment. Here, the utilization stage means a stage in which the machine learning model has been learned and the output of the machine learning model can be used or utilized for the original purpose.
(Step S210) The information generation device 10 acquires the failure type information DI and the network parameter NP from the input / output device 20.
(Step S220) The information generation device 10 acquires the task definition information TI from the task definition server 30.

(Step S230) The information generation device 10 supplies the information acquired in steps S210 and S220 to the trained model M (that is, the trained model 130). As a result, the trained model 130 generates a workflow WF.
In this example, the trained model 130 will be described as automatically generating the workflow WF based on the learning result by appropriately combining the task TSKs, but the present invention is not limited to this. For example, a workflow WF in which task TSKs are manually combined may be prepared in advance (for example, stored in a task definition server 30 or another storage device (not shown)). In this case, the trained model 130 may select a workflow WF according to the learning result from a plurality of types of registered workflow WFs prepared (for example, stored) in advance.

(Step S240) The information generation device 10 supplies the workflow WF generated in step S230 to the workflow control unit 40. The workflow control unit 40 executes an operation based on the supplied workflow WF to the network N via the network connection device 50.
(Step S250) The information generation device 10 determines the end of processing. When the information generation device 10 ends the execution of the process (step S250; YES), the information generation device 10 ends the process in the utilization stage. When the information generation device 10 does not end the process (step S250; NO), the information generation device 10 returns the process to step S210 and continues the process in the utilization stage.

[Operation of information generation system (re-learning)]
FIG. 8 is a diagram showing an example of the re-learning operation of the information generation system 1 of the present embodiment. Here, re-learning means updating the learning result of the trained model M by training the trained model M again.
Here, since the processing of each step from step S310 to step S340 is the same as the processing from step S210 to step S240 described above, the description thereof will be omitted.

(Step S350) The network monitoring unit 510 of the network connection device 50 acquires the status of the network N in which the operation based on the workflow WF is executed in step S340. The information generation device 10 acquires information indicating the status of the network N acquired by the network monitoring unit 510.
(Step S360) The information generation device 10 determines the end of processing. When the information generation device 10 does not end the process (step S360; NO), the information generation device 10 returns the process to step S310 and continues the process in the utilization stage. When the information generation device 10 ends the execution of the process in the utilization stage (step S360; YES), the information generation device 10 advances the process to step S370.

(Step S370) The information generator 10 evaluates the information indicating the status of the network N acquired in step S350. Since the specific operation is the same as the operation in step S160 described above, the description thereof will be omitted.
(Step S380) The information generation device 10 ends the process by reflecting the evaluation result ER in step S370 as teacher data in the trained model 130.

[Summary of Embodiment]
As described above, in the information generation system 1 of the present embodiment, the information generation device 10 automatically generates the workflow WF based on the trained model 130. In this trained model 130, information indicating the type of network failure D generated in the network NL to be trained and the network parameter NP related to the network NL to be trained are used as input data, and a task TSK for operating the network is defined. The workflow WF that combines one or more task TSKs selected from the existing task definition information TI is used as the output data, and the presence or absence of the network failure D that occurs as a result of giving the operation based on the workflow WF to the network N is used as the teacher data TD. The relationship between the type of failure D, the network parameter NP, and the workflow WF is learned. Therefore, according to the information generation system 1, the workflow WF for generating a desired type of network failure D can be automatically generated without using human hands. Therefore, according to the information generation system 1 of the present embodiment, the man-hours (that is, the operation man-hours) for creating the workflow WF for generating the network failure D can be reduced.

Further, the information generation system 1 of the present embodiment selects a workflow WF having a relatively high occurrence probability P of the type of network failure D indicated by the failure type information DI from among the plurality of workflow WFs output by the trained model 130. The selection unit 140 is provided. Here, even if an attempt is made to generate a certain type of network failure D, it may not be possible to generate a desired type of network failure D depending on the selection result of the workflow WF. In the information generation system 1 of the present embodiment, since the selection unit 140 selects the workflow WF having a relatively high probability P of occurrence of the desired type of network failure D, the desired type of network failure D cannot be generated. The frequency of occurrence of situations can be reduced. That is, according to the information generation system 1 of the present embodiment, it is possible to accurately generate a desired type of network failure D while reducing the operation man-hours for generating the workflow WF.

Further, in the trained model 130 of the present embodiment, the correspondence relationship between the workflow WF output by the trained model 130 and the situation of the network N resulting from giving the operation based on the workflow WF to the network N is set as the teacher data TD. Relearned.
Further, in the information generation system 1 of the present embodiment, in order to realize this re-learning, the situation and network failure that occur as a result of the determination unit 160 giving the network N an operation based on the workflow WF output by the trained model 130. The presence / absence of network failure D is determined based on the failure determination policy PCY in which the relationship with the presence / absence of D has been defined. Further, the information generation system 1 includes a storage unit 170 that stores the determination result of the determination unit 160 as the above-mentioned correspondence.
The information generation system 1 of the present embodiment can improve the accuracy of the type of network failure D to be generated by retraining the trained model 130. That is, the information generation system 1 of the present embodiment can accurately generate a desired type of network failure D while reducing the operation man-hours for generating the workflow WF.

Although the embodiments of the present invention and modifications thereof have been described above, these embodiments and modifications thereof are presented as examples, and are not intended to limit the scope of the invention. These embodiments and modifications thereof can be implemented in various other embodiments, and various omissions, replacements, and changes can be made without departing from the gist of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and at the same time, are included in the scope of the invention described in the claims and the equivalent scope thereof.

Each of the above-mentioned devices has a computer inside. The process of each process of each device described above is stored in a computer-readable recording medium in the form of a program, and the process is performed by the computer reading and executing this program. Here, the computer-readable recording medium refers to a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Further, this computer program may be distributed to a computer via a communication line, and the computer receiving the distribution may execute the program.

Further, the above program may be for realizing a part of the above-mentioned functions.
Further, it may be a so-called difference file (difference program) that can realize the above-mentioned function in combination with a program already recorded in the computer system.

1 ... Information generation system, 10 ... Information generation device, 20 ... Input / output device, 30 ... Task definition server, 40 ... Workflow control unit, 50 ... Network connection device, 110 ... Acquisition unit, 120 ... Supply unit, 130 ... Learned Model, 140 ... selection unit, 150 ... output unit, 160 ... judgment unit, 170 ... storage unit, N ... network, WF ... workflow, NP ... network parameter

Claims (6)

One or more selected from task definition information in which tasks for operating the network are defined, using information indicating the type of network failure to occur in the network to be learned and network parameters related to the network to be learned as input data. The workflow that combines the above tasks is used as output data, and the presence or absence of a network failure that occurs as a result of giving an operation based on the workflow to the network is used as training data, and the type of the network failure, the network parameters, and the workflow A trained model in which the relationship of
An acquisition unit that acquires failure type information indicating the type of network failure that occurs in the network to be inspected and network parameters related to the network to be inspected, respectively.
A supply unit that supplies the failure type information acquired by the acquisition unit and the network parameters to the trained model.
A selection unit that selects one or more workflows output by the trained model based on the input data supplied from the supply unit based on the probability of occurrence of a network failure of the type indicated by the failure type information.
An output unit that outputs the workflow selected by the selection unit, and an output unit that outputs the workflow.
An information generator comprising. The trained model is
A plurality of the workflows are output to the selection unit, and the workflow is output to the selection unit.
The selection unit
The information generation device according to claim 1, wherein the workflow having a relatively high probability of occurrence of a network failure of the type indicated by the failure type information is selected from the plurality of workflows output by the trained model. The presence or absence of a network failure is determined based on the failure determination policy that defines the relationship between the situation that occurs as a result of giving the network an operation based on the workflow output by the trained model and the presence or absence of a network failure. Judgment unit and
A storage unit that stores the workflow in association with a determination result that the determination unit determines a situation that occurs as a result of giving an operation based on the workflow to the network.
With more
The trained model is
The information generation device according to claim 1 or 2, wherein the presence or absence of a network failure determined based on the failure determination policy is learned by using the teacher data as the teacher data. The trained model is
The information generation device according to claim 3, wherein the correspondence between the workflow and the determination result stored in the storage unit is used as the teacher data and relearned. The information generator according to claim 3 or 4.
A workflow control unit that gives an operation based on the workflow output by the output unit to the network, and
A network monitoring unit that monitors the situation occurring in the network and outputs the monitored result to the determination unit of the information generator.
Information generation system equipped with. On the computer
An acquisition step for acquiring failure type information indicating the type of network failure that occurs in the network to be inspected and network parameters related to the network to be inspected, respectively.
One or more selected from task definition information in which tasks for operating the network are defined, using information indicating the type of network failure that occurs in the network to be learned and network parameters related to the network to be learned as input data. As output data, the workflow in which the above tasks are combined is used as output data, and the presence or absence of a network failure resulting from giving an operation based on the workflow to the network is used as training data, and the type of the network failure, the network parameters, and the workflow A supply step that supplies the failure type information acquired in the acquisition step and the network parameters to the trained model in which the relationship of
A selection step of selecting one or more workflows output by the trained model based on the input data supplied in the supply step based on the probability of occurrence of a network failure of the type indicated by the failure type information.
An output step that outputs the workflow selected in the selection step, and
A program to execute.

Images