Metrics processing system and metrics processing method

Abstract

The objective is to make metrics visible and enable the detection of subtle behavioral changes. [Solution] The monitoring device 30 stores mixed metrics, which are aggregated from multiple processes, and pure metrics, which are aggregated from only the target process. The execution device 20 waits until the start timing of the second window ahead, according to the window information, which includes the interval and timing of the window that is the aggregation interval for the mixed metrics, and repeats the target process a specified number of times to acquire processing information. The pure metrics calculation device 10 acquires the processing information and mixed metrics, calculates the average value of the metrics if there are metrics of the same classification in the mixed metrics, and uses the calculation result as the pure metrics. If there are no metrics of the same classification, it calculates the average value of the mixed metrics and uses the calculation result as the pure metrics, and outputs the pure metrics to the monitoring device 30.

Description

Technical Field

[0001] The present invention relates to, for example, a metrics processing system and a metrics processing method.

Background Art

[0002] Rollbacks in comprehensive tests due to non - fulfillment of non - functional requirements are highly damaging. Recent systems based on the cloud have limitations in empirical rules due to the interaction between components and dynamic behaviors such as scale and auto - recovery.

[0003] In particular, behaviors with limited timing, such as during fault recovery, are difficult to assume, and potential problems can be encapsulated in the system. For smooth development, it is desirable to quickly detect potential problems in addition to the problems manifested as test results and error logs in non - functional tests at the development stage and provide feedback to development.

[0004] There is a disclosed technique for identifying data not mapped to the steady state itself as an anomaly by detecting and removing trends through pattern detection from time - series data of metrics with multiple mixed trends (Patent Document 1).

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] In the above - mentioned conventional example, in a monitoring function such as a cloud service, anomalies in short - term fluctuations such as automatic fault recovery may be missed in detection. This is because the metrics in the monitoring function are defined, for example, as a 1 - minute average, and short - term changes are smoothed during the calculation of the average value.

[0007] Furthermore, in the aforementioned Patent Document 1, when pre- and post-test processing or processing of multiple tests is performed consecutively, fluctuations in metrics resulting from different factors are included within the same aggregation window, leading to the problem of mixed values ​​during smoothing.

[0008] This invention has been made in view of the above background, and aims to provide a metrics processing system and a metrics processing method capable of detecting minute behavioral changes. [Means for solving the problem]

[0009] To solve the above-mentioned problems and achieve the above objectives, one embodiment of the present invention is a metrics processing system comprising: an execution device that waits until the start timing of a specified window according to window information including the interval and timing of a window which is the aggregation interval of mixed metrics, in which multiple processes are aggregated, and repeats the target process a specified number of times to acquire processing information; and a pure metrics calculation device that acquires the processing information and the mixed metrics, recalculates the average value of the metrics of the target process based on that information, and outputs the calculation result as a pure metric, wherein the execution device and the pure metrics calculation device cooperate with each other to execute the processing.

[0010] Another embodiment of the present invention is a metrics processing system comprising: an execution device that waits until the start timing of a specified window, according to window information including the interval and timing of a window which is the aggregation interval of mixed metrics obtained by aggregating multiple processes together, and repeats the target process a specified number of times to acquire processing information; and a pure metrics calculation device that recalculates the average value of the metrics of the target process based on that information and outputs the calculation result as pure metrics.

[0011] Furthermore, another embodiment of the present invention is a metrics processing method for a metrics processing apparatus, characterized in that, according to window information including the interval and timing of a window which is the aggregation interval for mixed metrics aggregated from multiple processes, the apparatus waits until the start timing of a specified window, repeats the target process a specified number of times to acquire processing information, acquires the processing information and mixed metrics, recalculates the average value of the metrics of the target process based on that information, and obtains the calculation result as a pure metric. [Effects of the Invention]

[0012] According to the present invention, it becomes possible to detect subtle changes in the behavior of metrics. [Brief explanation of the drawing]

[0013] [Figure 1] This is an explanatory diagram illustrating the behavior and aggregation using the present-day implementation. [Figure 2] This is a block diagram showing an example configuration of a metrics processing system according to one embodiment of the present invention. [Figure 3] This is a configuration diagram showing an example of a computer hardware configuration according to one embodiment of the present invention. [Figure 4] This is an explanatory diagram showing an example of code according to one embodiment of the present invention. [Figure 5] This is an explanatory diagram showing the data structure of typical metrics according to one embodiment of the present invention. [Figure 6] This is an explanatory diagram showing processing information according to one embodiment of the present invention. [Figure 7] This is an explanatory diagram showing the data structure of the window storage unit according to one embodiment of the present invention. [Figure 8] This flowchart illustrates an example of the operation from window information to outputting processing information according to one embodiment of the present invention. [Figure 9] This flowchart illustrates an example of how to calculate net metrics using one embodiment of the present invention.

Best Mode for Carrying Out the Invention

[0014] Hereinafter, embodiments of the present invention will be described with reference to the drawings. Note that the following description and drawings are merely examples for explaining the present invention, and for the sake of clarity of the explanation, appropriate omissions and simplifications are made. In addition, the present invention can be implemented in various other forms. Also, unless otherwise particularly limited, each component may be in a single or plural number.

[0015] In the following description, the same or similar configurations may be denoted by the same reference numerals and redundant descriptions may be omitted. Also, in the following description, various types of information may be described using expressions such as "information" and "table", but the various types of information may be expressed in other data structures. Also, as expressions for identification information, there are expressions such as "identification information", "identifier", "name", "ID", and "number", and these can be replaced with each other. Also, in the following description, "database" is denoted as "DB" and "table" is denoted as "TBL", respectively.

[0016] First, an overview will be described using FIG. 1. FIG. 1 is an explanatory diagram for explaining the behavior and aggregation according to this embodiment. As shown in FIG. 1, in the actual behavior, a plurality of processes such as a separate process, a target process, and a separate process are mixed from the start to the end of the window. In such a case, as the value after aggregation of the metrics, the influences of the plurality of processes are mixed and averaged.

[0017] Therefore, the metrics processing system of this embodiment obtains a metrics value in which only the target process is aggregated, that is, a pure metric, from the metrics (mixed metrics) in which a plurality of processes are mixed and aggregated. In this embodiment, since it does not depend on the processes specific to testing, as an example of the target process, the processes of non-functional testing excluding pre-processing and post-processing are used.

[0018] Regarding the terms used below, a metric is time-series data of numerical values, such as the time change of CPU utilization. A window is the aggregation interval of metrics, and in the case of the present invention, aggregation refers to the calculation of an average value. In particular, in metrics, a mixed metric refers to a metric aggregated in a state where the influences of multiple processes are mixed, and a pure metric refers to a metric aggregated only for the target process.

[0019] Next, the configuration will be described using FIGS. 2 and 3. FIG. 2 is a block diagram showing a configuration example of a metric processing system according to the present embodiment, and FIG. 3 is a configuration diagram showing a hardware configuration example of a computer according to the present embodiment.

[0020] As shown in FIG. 2 for example, the metric processing system 1 is composed of a pure metric calculation device 10, an execution device 20, etc. connected via a network (wired, wireless, etc.) not shown.

[0021] The monitoring device 30 has a monitoring function provided as a cloud service, product, open source, etc. Examples of the monitoring device 30 include "AWS CloudWatch (registered trademark)", "DataDog (registered trademark)", "Prometheus", etc.

[0022] Note that the metric processing system 1 may be configured to include the monitoring device 30 inside, or may be a metric processing system or device in which the pure metric calculation device 10 and the execution device 20 are integrated into an integrated configuration.

[0023] The monitoring device 30 includes an input / output unit 31 that handles the input and output of data, etc., with the metrics processing system 1, and a metrics storage unit 32 that stores metrics. The input / output unit 31 communicates with the outside world using Web API (Application Programming Interface) or RPC (Remote Procedure Call). The metrics storage unit 32 stores metrics, and the schema is the same for both mixed metrics and pure metrics.

[0024] In the metrics processing system 1, the execution device 20 is a server having a program 21 that includes programs such as programs and application code, and the program 21 mainly executes processes such as the cooperation processing unit 22, the waiting processing unit 23, and the loop processing unit 24. The cooperation processing unit 22 is a process that exchanges information with the pure metrics calculation device 10, and may communicate directly via a Web API or exchange information via a file located in a mutually accessible location. The waiting processing unit 23 is a process that waits for processing until the next window, and the loop processing unit 24 is a process that repeats processing between windows.

[0025] The pure metrics calculation device 10 is a computer that calculates pure metrics using mixed metrics between the execution device 20 (which is a server) and the monitoring device 30, and includes a cooperation unit 11, a pure metrics calculation unit 12, a window response unit 13, a correlation calculation unit 14, a window storage unit 15, and the like.

[0026] The coordination unit 11 communicates with external devices such as the monitoring device 30 and the execution device 20. The pure metrics calculation unit 12 calculates pure metrics 43 based on the processing information 52 and the mixed metrics 41.

[0027] The window response unit 13 receives a request from the cooperation processing unit 22 via the cooperation unit 11, and returns the window information 51 to the cooperation processing unit 22 from the window storage unit 15, which is the largest window time that matches the conditions specified in the request. The correlation calculation unit 14 calculates the correlation between the target metric and the pure metric. The window storage unit 15 stores the window.

[0028] Regarding the data, the metrics processing system 1 is configured such that the execution device 20 outputs mixed metrics 40 to the input / output unit 31 of the monitoring device 30, the monitoring device 30 outputs mixed metrics 41 from the input / output unit 31 to the linkage unit 11 of the pure metrics calculation device 10, and the pure metrics calculation device 10 outputs pure metrics 42 from the linkage unit 11 to the input / output unit 31 of the monitoring device 30. Mixed metrics 40 and 41 are metrics aggregated with the combined effects of multiple processes and have the same data structure. Pure metrics 42, on the other hand, are metrics aggregated only from the target process.

[0029] Furthermore, the metrics processing system 1 is configured to output window information 51 from the linkage unit 11 of the pure metrics calculation device 10 to the linkage processing unit 22 of the execution device 20, and to output processing information 52 from the linkage processing unit 22 of the execution device 20 to the linkage unit 11 of the pure metrics calculation device 10. The window information 51 is data that conveys the specifications of the metrics window, and is a data structure that includes, for example, one sample of the most recent start time, the number of seconds in the window, and a message (error message, etc.). The processing information 52 is data that conveys the execution time of the target processing, and the format of this data is as shown in Figure 6, for example.

[0030] Next, the hardware configuration will be explained using Figure 3. The pure metrics calculation device 10, the execution device 20, and the monitoring device 30 are all computers, and their basic configurations are the same. Below, the computer configuration of the pure metrics calculation device 10 will be described as a representative example.

[0031] The pure metrics calculation device 10 is composed of a processor 201, a communication control device 202, an I / F 203, a main memory 204, an auxiliary memory 205, an internal bus 206, and the like, as shown in Figure 3, for example.

[0032] The processor 201 has a CPU that controls the entire device, a memory that stores processing programs (cooperation unit 11, pure metrics calculation unit 12, window response unit 13, correlation calculation unit 14) for the CPU to execute, and various data being stored during execution. The communication control device 202 is connected to the network 210 via the I / F 203 and is responsible for communication with various external devices (execution device 20, monitoring device 30, etc.).

[0033] I / F203 has the function of connecting the pure metrics calculation device 10 with an external device via the network 210. The main memory 204 functions as the main memory, and the auxiliary memory 205 stores and manages various data in formats such as DB and TBL, and also has the function of a window memory unit 15, and is an external memory device such as a hard disk.

[0034] The internal bus 206 is connected to each unit within the pure metrics calculation device 10 and is responsible for the communication of data, signals, etc., within the device. Although not shown in the figures, the pure metrics calculation device 10 may also be equipped with an operating device that includes a keyboard and a display device.

[0035] In the computer shown in Figure 3, in the case of the execution device 20, the program 21 is stored in the main memory 204. In the case of the monitoring device 30, the input / output unit 31 corresponds to the communication control device 202 and I / F 203, and the metrics storage unit 32 corresponds to the auxiliary storage device 205.

[0036] Next, the code will be explained using Figure 4. Figure 4 is an explanatory diagram showing an example of the code according to this embodiment. The specific processing will be explained in Figures 8 and 9. The program code 300 shown in Figure 4 uses Python, but other programming languages ​​may also be used.

[0037] In Figure 4, the general flow is as follows: window information acquisition 301, separate processing 302, control start processing 303 with repetition 304, return specification 305 and attributes 306, the target processing is the main processing 307, control end processing 308, and then the process continues to separate processing 309.

[0038] In the program code 300 shown in Figure 4, window information acquisition 301 is the process of obtaining window information from the pure metrics calculation device 10. `targets` is the window to be targeted. If multiple windows are specified, the window with the longest window time among the specified metrics is obtained. If left blank, the window with the longest window time among all metrics is obtained.

[0039] `maxWindowSeconds` is the maximum window duration in seconds (e.g., 180 seconds), and it is possible to specify it because it will increase the waiting time. If the metrics specified in `targets` exceed `maxWindowSeconds`, you may add an error message to the window information.

[0040] The control start process 303 / control end process 308 may use a programming language-specific syntax for specifying the start and end, such as the `with` clause, or they may be defined as functions that take the target process as an argument.

[0041] If `repeat` is False, the system waits for one window of processing to complete at the start of control processing 303. After the target processing is performed, the data is sent at the end of control processing 308, and the system waits until the next window. The behavior is the same even if the target processing does not fit into one window (this is handled on the aggregation side).

[0042] Next, various data structures will be explained using Figures 5 to 7. Figure 5 is an explanatory diagram showing the data structure of a typical metric according to this embodiment, Figure 6 is an explanatory diagram showing the data structure of processing information according to this embodiment, and Figure 7 is an explanatory diagram showing the data structure of the window storage unit according to this embodiment.

[0043] Regarding metrics, in principle, mixed metrics 40, 41 and pure metrics 42 have similar data structures and are stored in the metrics storage unit 32 of the monitoring device 30. As shown in Figure 5, the data structure of the metrics associates Time, Name, Attributes, and Value. Here, Attributes contain conditions that identify the metric, and include data such as measurement location (application name), URL, and processing success / failure status.

[0044] In the first example in Figure 5, Time is set to "2024-10-31T12:00:00Z", Name to "http.server.duration", Attributes to "{"app":"frontend","http.status":200}", and Value to "123".

[0045] Furthermore, the processing information 52 has, for example, the data structure shown in Figure 6, and is the data sent in the control termination process 308 (see Figure 4).

[0046] The data structure of the processing information 52, as shown in Figure 6, is a mapping of ID (identification information indicating classification), StartTime (control start time), EndTime (control end time), Correlation (correlation), and Attributes (attributes). Correlation and Attributes may be left blank. Also, for example, ID may be written as a UUID, and StartTime and EndTime may be written in the ISO 8601 format, but the method of writing is not limited to these.

[0047] In the first example in Figure 6, the ID is set to "48ba49ba-1377-4aff-b26b-55b0b5104a1c", the StartTime to "2024-10-30T 17:35:00Z", the EndTime to "2024-10-30T 17:35:35Z", the Correlation to "http.server.duration", and the Attributes to "{method: GET, status_code: 200}".

[0048] Furthermore, the data obtained through the repetition function, i.e., in range 52A of Figure 6, will have the same ID, "a37f0361-710f-48f3-93d2-54c0450b56fa". In this range 601A, the Correlation and Attributes fields are shown as examples of blank fields.

[0049] Furthermore, the data structure of the window storage unit 15 is such that, for example, as shown in Figure 7, Service, Metrics, StartTime (control start time), EndTime (control end time), and RegisteredTime (registration time) are associated.

[0050] In the first example in Figure 7, the Service is set to "Cloud A", Metrics to "http.server.duration", StartTime to "2024-10-30T 17:35:00Z", EndTime to "2024-10-30T 17:36:00Z", and RegisteredTime to "2024-10-30T 17:36:00Z".

[0051] The window memory unit 15 primarily records the aggregation windows for metrics. Regular expressions can be used for Service and Metrics. For example, jvm.* can be applied to both jvm.cpu.time and jvm.memory.used. Because specifications vary from cloud to cloud, the aggregation result from "StartTime" to "EndTime" is registered as a metric in "RegisteredTime," and the window is stored in that registration format.

[0052] The Service registers the providers of the monitoring devices 30, such as AWS CloudWatch, DataDog, and Prometheus. The Service is pre-configured in the window memory unit 15 as a startup option for the pure metrics calculation device 10. The data in the window memory unit 15 may also be pre-defined and registered.

[0053] Next, we will explain the operation using Figures 8 and 9. Figure 8 is a flowchart illustrating an example of the operation from window information to outputting processing information according to this embodiment, and Figure 9 is a flowchart illustrating an example of the operation for calculating pure metrics according to this embodiment.

[0054] The operation example in Figure 8 is a process based on the program 21 of the execution device 20, and aims to output processing information 52 (see Figure 6). The standby processing unit 23 performs steps P802 and P807 below, and the repeating processing unit 24 performs step P808. The cooperation processing unit 22 performs all other processing.

[0055] The following explanation will be based on the program code 300 in Figure 4. In step P801, window information is acquired (corresponding to window information acquisition 301). Here, the window interval and timing are acquired. The window information 51 may be acquired directly from the pure metrics calculation device 10, or the window information 51 stored in a specific location by the pure metrics calculation device 10 may be read.

[0056] Next, in step P802, a process is executed to wait until the start time of the next two windows. Here, a sample of metrics during the wait is obtained. Furthermore, in step P803, the control start time (StartTime) for the processing information is recorded (control start process 303), and in the following step P804, the target processing based on the processing information is executed.

[0057] Then, in step P805, the control end time (EndTime) for the processing information is recorded (control end processing 308), and in step P806, the processing is repeated only within the window time (repetition 304). For example, by specifying a value such as repeat=3, the process may be repeated more than the specified number of times even if it exceeds the window size.

[0058] In this way, when the iterative processing ends based on the judgment in step P806, the processing waits until the next window (step P807), and in the following step P808, Correlation and Attributes are obtained from the arguments of the control start processing. If Correlation and Attributes are not obtained, the fields are left blank. The processing information 52 may be sent directly to the pure metrics calculation device 10, or it may be obtained indirectly by outputting the data to a specific location and having the pure metrics calculation device 10 obtain it.

[0059] The operation example shown in Figure 9 is a process based on the program of the pure metrics calculation device 10, with the aim of outputting pure metrics 42. Steps P903 to P905 are performed by the pure metrics calculation unit 12, and step P906 is performed by the correlation calculation unit 14. All other processing is performed by the linkage unit 11.

[0060] First, in step P901, processing information 52 is obtained from the execution device 20, and then in the following step P902, mixed metrics 41 are obtained from the monitoring device 30. These mixed metrics 41 are shared with the mixed metrics 40 obtained by the monitoring device 30 from the execution device 20. Then, the following processing is performed for each ID of the processing information 52.

[0061] All metrics for both the standby and target processing phases are acquired from the monitoring service. For the standby phase, metrics are shown from the time before the StartTime (a window time period prior to StartTime in processing information 52) up to StartTime. For the target processing phase, metrics are shown from StartTime to EndTime. If Attributes are specified in processing information 52, those Attributes are used as filter conditions, and only obtainable metrics are acquired. Subsequent processing is performed for each mixed metric.

[0062] Next, in step P903, if a metric that satisfies the following conditions is obtained, the metric name and value are added as Key-Values ​​to the Attributes of the 42 pure metrics. If there are maximum and minimum metrics, they are used as reference values.

[0063] If the value during processing is greater than the value during waiting, the metric that refers to the maximum value of the target metric will, for example, use db.client.connection.max for db.client.connection.count as its reference value.

[0064] Furthermore, if the value during processing is smaller than the value during waiting, the metric that points to the minimum value of the target metric becomes the reference value.

[0065] Then, in step P904, it is determined whether to repeat based on whether there are multiple records with the same ID. If there are multiple identical records (YES route in step P904), the repeated calculation is performed in step P905. If there are no multiple identical records, the average recalculation is performed in step P908.

[0066] In step P905, the average of the metrics that fall within the time range of the repeating process becomes the pure metric. Since the target process is running during that time, there is no need to allocate the average to account for waiting time. Also, the Attributes property "repeat enable" is added.

[0067] In the following step P906, if there is a pure metric with the same name that has not been processed repeatedly in the past, its value is compared with the value in step P905. If they differ by a certain percentage (for example, 30% or more smaller), it is considered an anomaly in the repetition, and in step P907, repeat_anomaly:suspected is added to the attributes of the pure metric.

[0068] Furthermore, in step P908, since there is no repetition, the average value of the mixed metrics is recalculated under the assumption that the target process was performed only between StartTime and EndTime, and that the value was the same as the waiting metric value at all other times.

[0069] For example, if the mixed metric value is 100 (an example of a 1-minute average), the waiting metric value is 0 (an example of a 1-minute average), and the target process is executed for only 30 seconds, it is determined that the "target process with an average metric value of 200" and the "waiting process with an average value of 0" each lasted for 30 seconds, resulting in the generation of a mixed metric with an average value of 100, and the pure metric value is set to 200.

[0070] Given that the duration of the target process is t, the pure metric is x, the window interval is w, the waiting metric value is a, and the mixed metric is m, the pure metric is given by the following equation (1).

[0071]

number

[0072] Then, in step P909, the past values ​​of the same-named pure metric and the metric specified as correlations in the processing information 52 for that time period are obtained, the correlation coefficient is calculated, and in step P910, correlation=(correlation coefficient) is added to the Attributes of the pure metric, and the pure metric 42 is output to the monitoring device 30.

[0073] As explained above, this embodiment makes it possible to make metrics apparent and detect subtle behavioral changes. In particular, since the control function in the code controls the execution of the program based on the timing of metric aggregation, it is possible to determine behavioral changes for each test from the metrics. Furthermore, by controlling the start times of the target process and subsequent processes so that the target process fits independently within the time range of the metrics window, it is possible to suppress the mixing of processing content in the aggregated metrics.

[0074] Furthermore, by repeating the target process during the metrics aggregation period, it becomes easier to reflect the behavior of the target process in the aggregated metrics.

[0075] Furthermore, by calculating the correlation between the calculated net metric value and the specified metric value, it is possible to see how much a change in the net metric affects the specified metric.

[0076] Furthermore, each of the above-mentioned configurations, functional units, processing units, processing means, etc., may be implemented in hardware, in whole or in part, for example, by designing them as integrated circuits. Alternatively, each of the above-mentioned configurations, functions, etc., may be implemented in software by having the processor interpret and execute programs that realize each function. Information such as programs, tables, and files that realize each function can be stored in memory, hard disks, SSDs (Solid State Drives), or other recording devices, or in recording media such as IC cards, SD cards, or DVDs.

[0077] Furthermore, the arrangement of the various functional units, processing units, and databases described above is merely an example. The arrangement of the various functional units, processing units, and databases can be changed to the optimal arrangement from the standpoint of the performance, processing efficiency, and communication efficiency of the hardware and software of these devices.

[0078] Furthermore, the configuration of the database (schema, etc.) that stores the various types of data mentioned above can be flexibly modified from the perspective of efficient resource utilization, improved processing efficiency, improved access efficiency, and improved search efficiency. [Explanation of Symbols]

[0079] 1. Metrics Processing System 10. Pure Metrics Calculation Device 11. Liaison Department 12. Pure Metrics Calculation Unit 13 Window Response Unit 14. Correlation Calculation Unit 15 Window memory 20 Execution device 21 Programs 22. Cooperative Processing Unit 23 Standby Processing Unit 24 Repeating Processing Unit 30 Monitoring equipment 31 Input / output section 32 Metrics Memory Unit 40, 41 Mixed metrics 42 Pure Metrics 201 Processor 202 Communication control device 203 I / F 204 Main storage 205 Auxiliary storage device 206 Internal Bus 300 program codes

Claims

1. A metrics processing system, An execution device that, according to window information including the interval and timing of the window which is the aggregation interval for mixed metrics aggregated from multiple processes, waits until the start timing of the second window ahead, and repeats the target process a specified number of times to acquire processing information, A pure metrics calculation device that acquires the processing information and the mixed metrics, calculates the average value of the metrics if there are metrics of the same classification in the mixed metrics and uses the calculation result as a pure metric, calculates the average value of the mixed metrics if there are no metrics of the same classification and uses the calculation result as a pure metric, and acquires a pure metric based on a correlation process between the pure metric and a pre-prepared target pure metric, Prepare, A metrics processing system characterized in that the execution device and the pure metrics calculation device cooperate with each other to perform processing.

2. A metrics processing system according to claim 1, further comprising a notification function, wherein the pure metrics calculation device calculates the value of the pure metrics of only the target process from the average value of the mixed metrics using the average value of the mixed metrics and the execution period of the target process based on the time notified by the notification function.

3. A metrics processing system according to claim 1, characterized in that the pure metrics calculation device controls the start times of the target process and subsequent processes so that the target process falls within the time range of the metric window on its own.

4. A metrics processing system according to claim 3, characterized in that the pure metrics calculation device repeats the target processing during the metrics aggregation period.

5. A metrics processing system according to claim 4, wherein the pure metrics calculation device outputs an error indicating an abnormality when the metric value changes significantly from the value when there is no repetition as a result of the repeated execution.

6. A metrics processing system according to claim 4, wherein the pure metrics calculation device calculates the value over a period that combines multiple aggregation periods when it is not possible to perform many repetitions within the aggregation period.

7. A metrics processing system according to claim 3, wherein the pure metrics calculation device uses the values ​​of the metrics as reference values ​​when the maximum or minimum value of a metric has been measured within the aggregation period.

8. A metrics processing system according to claim 1, characterized in that the execution device and the pure metrics calculation device are interconnected via a network.

9. A metrics processing system according to claim 1, characterized in that it is a device that integrates the execution device and the pure metrics calculation device.

10. A metrics processing system, A monitoring device that stores mixed metrics, which are aggregated from multiple processes, and pure metrics, which are aggregated from only the target process. An execution device that, according to window information including the interval and timing of the window which is the aggregation interval of the aforementioned mixed metrics, waits until the start timing of the second window ahead, and repeats the target process a specified number of times to acquire processing information, A pure metrics calculation device that acquires the processing information and the mixed metrics, calculates the average value of the metrics if there are metrics of the same classification in the mixed metrics and uses the calculation result as a pure metric, calculates the average value of the mixed metrics if there are no metrics of the same classification and uses the calculation result as a pure metric, acquires the pure metrics based on a correlation process between the pure metrics and pre-prepared target pure metrics, and outputs the pure metrics to the monitoring device, A metrics processing system characterized by having the following features.

11. A metrics processing method for a device that performs metrics processing, A metrics processing method characterized by waiting until the start timing of the second window ahead, according to window information including the interval and timing of the window which is the aggregation interval for mixed metrics aggregated by multiple processes, repeating the target process a specified number of times to obtain processing information, obtaining the processing information and the mixed metrics, calculating the average value of the metrics if there are metrics of the same classification in the mixed metrics and taking the calculation result as a pure metric, calculating the average value of the mixed metrics if there are no metrics of the same classification and taking the calculation result as a pure metric, and obtaining a pure metric based on a correlation process between the pure metric and a pre-prepared target pure metric.

Images