System for requesting troubleshooting and method for requesting troubleshooting

Abstract

The appropriate analysis team will be responsible for troubleshooting information processing system failures. [Solution] A fault response request system for performing fault countermeasures for information processing systems comprises: a first prompt input unit that receives fault information and requests a first generation AI to summarize the fault information; an update processing unit that receives the summarized fault information from the first generation AI and registers it in the stored fault DB; a similarity comparison unit that searches the stored fault DB based on the summarized fault information and finds multiple faults similar to the summarized fault information; a second prompt input unit that receives the summarized fault information and inference criteria and requests a second generation AI to infer an analysis team that fits the inference criteria using the summarized fault information from the multiple faults obtained; and a fault response request output unit that receives the analysis team inferred by the second generation AI and outputs a countermeasure request to the received analysis team.

Description

Technical Field

[0001] The present invention relates to a failure countermeasure request system and a failure countermeasure request method.

Background Art

[0002] In a system for taking countermeasures against failures occurring in an information processing system to be maintained, it is necessary to determine, in a short time, which analysis team will analyze the failure when the failure is received. To make such a determination, an engineer with high skills is required, and it is necessary to allocate engineers for each product with a different architecture used in the maintenance target system.

[0003] For this reason, an accumulated failure database (DB) is created from failure data of past failures, and an analysis team for analyzing the failure is selected by searching the accumulated failure DB.

[0004] However, the failure data is not necessarily described in a unified format, and it is difficult to register the failure data in the accumulated failure DB in a form that is easy to search.

[0005] Furthermore, when searching the accumulated failure DB, there is a problem that it is difficult to obtain an appropriate analysis team unless the search is performed using the terms in use.

[0006] Therefore, it is desired to register and update the daily-occurring failure data in the failure accumulation DB in a timely manner in a form that is easy to search, and when a new failure countermeasure request is received, to determine the analysis team in a short time by searching with a query that can obtain an appropriate countermeasure team from the failure accumulation DB.

[0007] Patent Document 1 discloses a technique in which a system evaluates that a team is formed by members selected based on a project history database from among a plurality of members belonging to an organization.

Prior Art Documents

Patent Documents

[0008] [Patent Document 1] Japanese Patent Publication No. 2007-42059 [Overview of the Initiative] [Problems that the invention aims to solve]

[0009] Patent Document 1 describes how to obtain an evaluation value for an organization when a team is formed, based on the contents of a memory that stores information about each member belonging to the organization, but it does not disclose how the information about each member belonging to the organization is created.

[0010] One of the problems that this invention aims to solve is to provide a fault countermeasure request system and a fault countermeasure request method that enable a suitable analysis team to quickly handle fault countermeasures based on fault information of an information processing system. [Means for solving the problem]

[0011] The above problem is solved by a fault response request system for performing fault countermeasures for information processing systems, comprising: a first prompt input unit that receives fault information and requests a first generation AI to summarize the fault information; an update processing unit that receives the summarized fault information from the first generation AI and registers it in the stored fault database; a similarity comparison unit that searches the stored fault database based on the summarized fault information and finds multiple faults similar to the summarized fault information; a second prompt input unit that receives the summarized fault information and inference criteria and requests a second generation AI to infer an analysis team that fits the inference criteria using the summarized fault information from the multiple faults obtained; and a fault response request output unit that receives the analysis team inferred by the second generation AI and outputs a countermeasure request to the received analysis team. [Effects of the Invention]

[0012] A suitable analysis team can handle troubleshooting for information processing system failures. [Brief explanation of the drawing]

[0013] [Figure 1] This is an example of the system configuration diagram in the embodiment. [Figure 2] This is an example of the hardware configuration diagram in the embodiment. [Figure 3] This is an example of the screen of the first prompt input part in the embodiment. [Figure 4] This is an example of the failure information in the embodiment. [Figure 5] This is an example of the summarized failure information in the embodiment. [Figure 6] This is an example of the screen of the second prompt input part in the embodiment. [Figure 7] This is an example of the screen of the inference criteria and inference results in the embodiment. [Figure 8] This is an example of the screen of the inference results including the inference criteria and reasons in the embodiment. [Figure 9] This is an example of the analysis request in the embodiment. [Figure 10] This is an example of the accumulated failure DB in the embodiment. [Figure 11] This is an example of the failure information DB in the embodiment. [Figure 12] This is an example of the flowchart showing the processing content of the registration processing part in the embodiment. [Figure 13] This is an example of the flowchart showing the processing content of the search processing part in the embodiment. [Figure 14] This is an example of the system configuration diagram in the second embodiment in the embodiment.

Mode for Carrying Out the Invention

[0014] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In each drawing for explaining the embodiments, the same components are denoted by the same names and reference numerals as much as possible, and the repeated explanations thereof are omitted.

[0015] The present invention is not limited to the embodiments described below, and includes various modifications and equivalent configurations within the scope of the appended claims. For example, the above-described embodiments have been described in detail for the purpose of explaining the present invention clearly, and the present invention is not necessarily limited to those having all the configurations described.

[0016] In addition, the processing units and processing modules described in the embodiments may be realized in hardware by designing part or all of them, for example, by means of an integrated circuit, or may be realized in software by a processor interpreting and executing a program for realizing each function.

[0017] The information described in the embodiments may be a table or a database (DB), or may be data stored in the main memory.

[0018] FIG. 1 is an example of a system configuration diagram in an embodiment. The trouble countermeasure request system includes a registration processing unit 2 that registers trouble information summarized in the accumulated trouble DB 14, and a search processing unit 3 that searches for troubles close to the trouble information using the information in the accumulated trouble DB 14 and estimates the analysis team 15.

[0019] The registration processing unit includes a first prompt input unit 4 that creates a prompt for causing the newly received trouble information to be summarized by the first generation AI 8, and a division processing unit 5 that divides the summary of the trouble information obtained from the first generation AI 8.

[0020] Furthermore, it includes a vectorization processing unit 6 that vectorizes the divided summary, and an update processing unit 7 that updates the accumulated trouble DB 14 using the vectorized summary.

[0021] The search processing unit 3 includes a similarity comparison unit 9 that searches for a specified number of registered troubles close to the information vectorized by the vectorization processing unit 6 for the newly received trouble information and the information vectorized by the accumulated trouble DB 14.

[0022] The similarity comparison unit 9 may also accept the specification of a comparison method, such as cosine similarity, hashing, or comparison based on Levenshtein distance.

[0023] It includes a second prompt input unit 10 that creates a prompt to cause the second generating AI 11 to make a prediction based on the searched failures, newly received failure information, and prediction criteria that indicate what criteria will be used to predict the analysis team.

[0024] Furthermore, it includes a countermeasure request output unit 12 that outputs an analysis request to the analysis team 15 predicted by the second generating AI 11.

[0025] The countermeasure request output unit 12 may display multiple analysis teams predicted by the second generation AI 11 to the operator's screen, accept the prediction of the analysis team 15 to which countermeasures should be requested, and output the countermeasure request to the accepted analysis team 15.

[0026] Once the analysis team 15 completes the countermeasures and inputs the results into the fault information update unit 16, the fault information update unit 16 registers the results in the fault information DB 13.

[0027] This system configuration allows for the simultaneous registration of a summary into the stored fault database (DB14), a search for similar faults using the vectorized summary, and the prediction of an appropriate analysis team when new fault information is received.

[0028] Furthermore, since the registered summaries are summarized according to predetermined criteria, they contain less redundant information, allowing the similarity comparison unit 9 to find highly similar problems when searching for similar problems.

[0029] In this embodiment, the first generation AI 8 and the second generation AI 11 are described as being located outside the fault response request system. This makes it possible to use the appropriate generation AI.

[0030] Furthermore, if you wish to use a dedicated generation AI, you may set up a first generation AI and a second generation AI within the fault response request system.

[0031] Figure 2 shows an example of a hardware configuration diagram in the embodiment. The fault response request system 1 consists of a computer equipped with a processor 20, memory 21, external storage device 22, input / output device 23, and network interface 24.

[0032] This embodiment describes an example implemented on a standalone computer, but a cloud system providing computing resources may also be used.

[0033] The processor 20 consists of processing units such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), the memory consists of memory elements such as RAM (Random Access Memory) and ROM (Read Only Memory), the external storage device 22 consists of storage devices such as an HDD (Hard Disk Drive) and an SSD (Solid State Drive), the input / output device 23 consists of a keyboard, display, touch panel, etc., and the network interface consists of hardware such as a NIC (Network Interface Card).

[0034] Software modules such as the first prompt input unit 4, the splitting processing unit 5, the vectorization processing unit 6, the update processing unit 7, the second prompt input unit 10, the countermeasure request output unit 12, and the fault information update unit 16 are stored in memory 21. Based on user instructions received from the input / output device 23, the processor 20 refers to information such as the accumulated fault DB 14 and fault information DB 13 stored in the external storage device 22 and executes the software.

[0035] Figure 3 shows an example of the screen of the first prompt input unit in the embodiment. The first prompt input unit 4 outputs a first prompt screen that includes a system prompt 31 and a user prompt 32.

[0036] The user enters the command 33, the extracted items 34, the deleted items 35, the replacement terms 36, the output format 37, and the text 38 described in the failure information into the user prompt.

[0037] Command 33 is a command to summarize the failure information, and a default command may be provided. Extraction items 34 is where instructions are entered to extract the "phenomenon," "points," and "conditions" of the failure. The extraction items are not limited to those exemplified here, but in order to maintain consistency in the accumulated failure DB 14, it is preferable to specify the same items for all failure information.

[0038] Here, the "phenomenon" section briefly describes what happened. It should include "who," "to what," "what was done as a result," and "what happened." The "critique" section describes what is considered a problem with the phenomenon and what is being suggested as the solution. The "conditions" section describes the model name, program version, and the conditions under which the phenomenon occurred.

[0039] By including "phenomenon," "points raised," and "conditions" in the summary, the accuracy of the appropriate analysis team's predictions can be improved from fault information described in various forms.

[0040] Deletion item 35 instructs the system to remove content from the failure information. Failure information may include large amounts of log data such as operation logs and system logs, as well as dump data. By instructing the system to delete this information, the quality of the summary can be improved.

[0041] Replacement term 36 is an instruction to replace terms used in the failure information with terms used in the summary, thereby eliminating inconsistencies in terminology. Furthermore, by unifying locally used terminology within individual organizations with standardized, general terminology, the search efficiency when the search processing unit 3 searches the accumulated failure DB 14 can be improved.

[0042] Output format 37 is an instruction that specifies the format for outputting the summary. In this example, it is set to the phenomenon, observation, and condition specified in extraction item 34. Since output format 37 also affects the search efficiency of the search processing unit 3, it is desirable to specify the same output format 37 for all failure information.

[0043] The user prompt 32 can be entered by the user, but content other than the text 38 for entering fault information may be left as default, or may be included in the system prompt.

[0044] Figure 4 shows an example of fault information in the embodiment. The fault information includes standardized information 40 such as status, priority, and person in charge ID, and a free-text description 41.

[0045] Standardized information 40 is entered according to predetermined criteria, but the content of the explanation 41 varies greatly depending on the type of disability and the person entering the information. Therefore, it is necessary to summarize it according to established criteria.

[0046] For example, the dump storage location in (1 / 6) is meaningless information for a summary used to select an analysis team. The phenomenon in (2 / 6) contains information necessary for the summary, but it uses the terms "scan" and "CA," which are inconvenient terms for searching in the search processing unit 3 as they are.

[0047] The points raised in (3 / 6) also contain information necessary for summarizing, but the term "CV" is included here as well, which is inconvenient for searching with the search processing unit 3. The same applies to "scan" included in the conditions for occurrence in (4 / 6).

[0048] Figure 5 shows an example of summarized failure information in an embodiment. The summarized failure information includes the subject, phenomenon, findings, and initial analysis team.

[0049] In the subject line, the machine name listed in the incident information (5 / 6) is used, and "scan" is replaced with "sweep". In the description of the phenomenon, "CA" is replaced with "chronoamperometry". In the description of the issue, "CV" is replaced with "cyclic voltammetry" and summarized.

[0050] This issue has not yet been assigned to an analysis team, therefore the initial analysis team is unknown. Note that the summarized issue information does not include any conditions, but it is not necessary to include all items specified in output format 37.

[0051] Figure 6 shows an example of the screen of the second prompt input unit in the embodiment. The second prompt input unit 10 outputs a second prompt screen that includes a system prompt 61 and a user prompt 62.

[0052] The user prompt 62 specifies six failures 63 that are similar to the input failure information, selected by the similarity comparison unit 9 from the accumulated failure DB 14. The number of failures can be narrowed down to failures similar to the input failure information by specifying a maximum number of failures to the similarity comparison unit 9.

[0053] By specifying summarized and vectorized fault information 64 and prediction criteria 65 for selecting an analysis team, the second generating AI 11 can be asked to predict an analysis team suitable for handling the input fault information 64.

[0054] For prediction criterion 65, the text may be automatically generated by accepting user selections from a predetermined list of items on a separate screen.

[0055] For example, it is possible to include the following instructions in user prompt 62 and select an analysis team that matches the user's instructions. If cost-effectiveness is selected, the analysis team that minimizes the value obtained by multiplying the number of members in the analysis team by the period from the date of occurrence to the date of resolution will be selected. If growth potential is selected, the analysis team with the lowest average years of experience across the entire team will be chosen. If you select "rapidity," the analysis team with the shortest timeframe from the time of occurrence to the time of resolution will be selected. The parts of the user prompt 62 other than the fault 63, fault information 64, and inference criteria 65 may be written as defaults or may be written in the system prompt.

[0056] Figure 7 shows an example of the estimation criteria and estimation results screen in an embodiment. In this example, the user can select three estimation criteria: economic efficiency, growth potential, and speed, but other criteria may also be provided.

[0057] As in this example, selecting economic efficiency changes the second prompt screen to an instruction that reads, "For the problem-solving team, please answer based on the years of experience of the team members."

[0058] By selecting personnel with many years of experience, analysis can be performed efficiently, which is expected to improve cost-effectiveness. By pre-defining the conversion method, it becomes possible to flexibly change the estimation criteria.

[0059] Based on the economic estimation criteria, the second generation AI 11 selects and outputs email addresses that can identify Team A, Team B, Team C, and their respective personnel, based on the six problems 63 entered on the second prompt screen.

[0060] At this time, by also providing information such as "We are answering based on years of experience," which indicates the conditions actually entered in the prompt, users can more easily judge the appropriateness of the selected team.

[0061] Figure 8 shows an example screen of the prediction results, including the prediction criteria and reasons, in an embodiment. As in this example, the prediction results of the second generating AI 11 can also provide the user with the information necessary for their final decision by outputting the predicted analysis team and the reasons for that analysis team's prediction. The reasons for the prediction include information such as the team's expertise and problems they have handled in the past.

[0062] Figure 9 shows an example of an analysis request in an embodiment. In this example, the analysis request is sent to the analysis team by email, but it may also be sent using a system such as Wormflow.

[0063] In this example, the analysis request is sent to Team A, but it is also possible to request analysis from multiple analysis teams simultaneously.

[0064] When requesting analysis from multiple teams, obtaining confirmation from each team that receives the request regarding their availability and then selecting the appropriate team will enable more efficient analysis.

[0065] Figure 10 shows an example of the accumulated fault database in the embodiment. The accumulated fault database 14 is composed of a vector database, and the vector distance of the fault information summary can be calculated based on the words contained in the fault information summary.

[0066] The similarity comparison unit 9 uses the accumulated failure DB 14 to find past failure information summaries that are close in vector distance to the newly received failure information summary.

[0067] In this example, the past failure information summaries for failures 1 through 10, enclosed by dotted lines, which are close in vector distance to the new failure information summary (represented by a black circle), are retrieved from the past failure information summaries (represented by white circles) registered in the accumulated failure DB14.

[0068] In Figure 10, for the sake of simplicity, the explanation uses a two-dimensional vector distance between two words, "Problem Name" and "Analysis Team." However, an actual accumulated fault database is a vector database of three dimensions or more.

[0069] The requested obstacles 1 through 10 are as follows: Issue 1: (GUI display error, Analysis Team A) Failure 2: (Storage device failure, Analysis Team B) Issue 3: (Insufficient storage capacity, Analysis Team A) Issue 4: (Network failure, Analysis Team C) Issue 5: (Network connection failure, Analysis Team C) Failure 6: (Backup failure, Analysis Team B) Issue 7: (Network latency, Analysis Team C) Issue 8: (Network bandwidth limitations, Analysis Team D) Issue 9: (Data corruption, Analysis Team D) Issue 10: (Disk failure, Analysis Team E) Figure 11 shows an example of a fault information database in the embodiment. The fault information database is a database for managing received faults, and it registers information such as the fault information ID, problem name, description of the issue, phenomenon, conditions for occurrence, date and time, and initial analysis team, all associated with each other.

[0070] Since the fault information database is not used by the search processing unit 3, terms such as "scan" entered by the person who reported the fault are registered as they are. Information such as operation logs and system logs are also registered.

[0071] Furthermore, it includes "progress information" to register the progress of the analysis. By establishing a fault information database and reflecting the analysis results of the analysis teams, administrators can not only grasp the status of countermeasures for individual faults, but also see which analysis team is analyzing which fault, enabling more efficient operation of the analysis teams.

[0072] Figure 12 is an example flowchart showing the processing content of the registration processing unit in the embodiment.

[0073] First, the first prompt input unit 4 receives fault information (S1), and the registration processing unit 2 requests the first generation AI 8 to summarize the fault information (S2). The registration processing unit 2 receives the summary of fault information from the first generation AI 8 (S3).

[0074] The registration processing unit 2 determines whether the received summary is below a predetermined capacity (S4). In this example, 2KB is used as the standard. If it exceeds 2KB, the splitting processing unit 5 splits it into 2KB or less (S5), and the process proceeds to step S6 to vectorize the failure information.

[0075] If the value in step S4 is 2KB or less, the vectorization processing unit 6 performs vectorization of the summarized fault information (S6).

[0076] Next, the update processing unit 7 registers the vectorized summary failure information to the stored failure DB 14 (S7).

[0077] Figure 13 is an example flowchart showing the processing steps of the search processing unit in the embodiment.

[0078] First, the registration processing unit 2 receives a summary of new vectorized fault information to be searched (S20). The similarity comparison unit 9 refers to the stored fault DB 14 and finds past faults that are close in vector distance to the received summary of fault information (S21). At this time, it may also receive a specification of the number of faults to be found and find faults up to the specified number.

[0079] The system receives the user's specified inference criteria (S22), and based on the inference criteria and past failures determined by the similarity comparison unit 9, the second prompt input unit 10 creates a prompt and requests the second generation AI 11 to make an inference for the analysis team (S23).

[0080] The search processing unit 3 receives the analysis team from the second generation AI 11 (S24), and the countermeasure request output unit 12 requests the analysis team to analyze the new failure (S25).

[0081] When the countermeasures are completed, the analysis team 15 requests the fault information update unit 16 to update the fault information database 13, and the fault information update unit 16 updates the fault information database 13. This makes it possible to manage the status of fault countermeasures.

[0082] Figure 14 is an example of a system configuration diagram in the second embodiment of the embodiment.

[0083] In the first embodiment, when new fault information was received, registration in the stored fault DB 14 and the request for analysis from the countermeasure request output unit 12 to the analysis team 15 were performed synchronously. In the second embodiment, it is possible to perform registration in the stored fault DB 14 and the request for analysis to the analysis team asynchronously.

[0084] In the second embodiment, the first prompt input unit 4, the division processing unit 5, and the vectorization processing unit 6 provided in the registration processing unit 2 are added to the search processing unit, and the registration processing and search processing can be processed in parallel by requesting the first generation AI 8 to summarize the failure information from the search processing unit 3.

[0085] This configuration allows for the registration of summarized fault information in the accumulated fault database for faults that have been addressed or have low priority, as registered in the fault information DB13, during periods when computing resources are available. This increases the flexibility of system operation.

[0086] In this example, the first prompt input unit 4, the division processing unit 5, and the vectorization processing unit 6 are duplicated in the registration processing unit 2 and the search processing unit 3, but they do not need to be physically duplicated; it is sufficient if they can be launched independently in the registration processing unit 2 and the search processing unit 3.

[0087] The processing of the fault response request system 1 described herein is intended to facilitate understanding of the present invention. Processing may be added or removed as needed based on the type, configuration, application, and characteristics of the information processing system targeted for fault response. Therefore, the present invention does not need to include all the processing described herein and is not limited to the embodiments. [Explanation of Symbols]

[0088] 1. Incident Response Request System 2. Registration Processing Unit 3. Search Processing Unit 4. First prompt input section 5-part processing unit 6. Vectorization Processing Unit 7. Update Processing Unit 8. The first generation AI 9 Similarity comparison section 10 Second prompt input section 11 Second Generation AI 12 Countermeasure Request Output Unit 13. Failure Information Database 14. Accumulated Failure Database 15 Analysis Team 16. Incident Information Update Department

Claims

1. In a fault response request system for handling faults in information processing systems, A first prompt input unit receives fault information and requests a first generation AI to summarize the fault information, An update processing unit receives the summarized failure information from the first generation AI and registers it in the stored failure DB, A similarity comparison unit searches the stored fault database based on the summarized fault information and finds multiple faults similar to the summarized fault information. A second prompt input unit receives the summarized failure information and prediction criteria, and requests a second generating AI to find an analysis team that fits the prediction criteria using the multiple failures obtained and the summarized failure information, A fault response request system equipped with a countermeasure request output unit that receives analysis teams requested by a second generation AI and outputs countermeasure requests to the received analysis teams.

2. In the fault response request system described in claim 1, The aforementioned first prompt input unit is a fault countermeasure request system that requests the first generating AI to create a summary that includes at least the phenomenon of the fault, the content of the issue, and the conditions under which it occurred.

3. In the fault response request system described in claim 1, The aforementioned first prompt input unit is a fault response request system that, when requesting a summary from the first generation AI, requests that it create a summary that deletes at least the log data and dump data of the fault information.

4. In the fault response request system described in claim 1, It includes a division processing unit that divides the summarized fault information received from the first generating AI into summaries of a predetermined length, The aforementioned update processing unit is a fault countermeasure request system that registers the divided summaries in the accumulated fault database.

5. In the fault response request system described in claim 1, The aforementioned similarity comparison unit accepts a specification of a comparison method, and the fault countermeasure request system uses the accepted comparison method to find faults similar to the fault information summarized from the accumulated fault DB.

6. In the fault response request system described in claim 1, The aforementioned estimation criteria are a fault response request system that includes at least one criterion from the following: economic efficiency, speed, and growth potential.

7. In the fault response request system described in claim 1, A fault countermeasure request system comprising a fault information update unit that updates a fault information database, which manages fault information based on the fault information received when it receives the results of countermeasures from the aforementioned analysis team.

8. In the fault response request system described in claim 1, The aforementioned countermeasure request output unit receives multiple analysis teams from the second generation AI, outputs the received multiple analysis teams, accepts the selection of an analysis team to request analysis from among the outputted multiple analysis teams, and outputs a countermeasure request to the selected analysis team.

9. In the fault response request system described in claim 1, A fault response request system comprising a registration processing unit including a first prompt input unit, and a search processing unit including both the first prompt input unit and a second prompt, wherein the registration processing unit and the search processing unit are capable of asynchronous processing.

10. In a method for requesting troubleshooting for information processing system failures, The first prompt input unit receives the failure information and requests the first generation AI to summarize the failure information. The update processing unit receives the summarized failure information from the first generating AI and registers it in the stored failure DB. The similarity comparison unit searches the stored fault database based on the summarized fault information and finds multiple faults similar to the summarized fault information. The second prompt input unit receives the summarized fault information and inference criteria, and requests the second generation AI to infer an analysis team that fits the inference criteria using the summarized fault information from the multiple faults obtained. A fault response request method in which a countermeasure request output unit receives an analysis team predicted by a second generation AI and outputs a countermeasure request to the received analysis team.

Images