JP6163982B2 - Business application delay detection device, method and program - Google Patents

Description

  The present invention relates to a technology for detecting a delay of a business application in a system in which many business applications operate.

  Transaction processing is online real-time processing of a business application, which makes it possible to immediately process given information and receive the result at a required place. In a large-scale computer system in which many business applications operate, if there is a problem in the processing of a certain business application, it is difficult to find all the problems in advance testing. For this reason, when a delay occurs in transaction processing in a specific business application having a problem during system operation, processing of other normally operating business applications is also affected. Therefore, a means for detecting a processing delay due to a problem of a business application is necessary.

  As a delay detection method, a method of determining the presence or absence of delay using a threshold value for each type of business application is common. Further, as a more accurate delay detection method, the method of Patent Document 1 is disclosed.

  In Patent Document 1, in order to detect a delay in a business application, the processing delay is calculated from the average time of processing of business applications executed within the latest predetermined time and the average time of processing executed before that time. Is detected. According to this method, even if the overall average time of business application processing has subsided within the threshold, it is possible to detect a delay when the latest average time is significantly longer than the previous average time. It becomes. In addition, since it is not necessary to set a threshold value for each type of business application, the operation cost of the system administrator can be reduced.

JP 2007-249663 A

  However, in the case of a system in which the number of transactions varies depending on the time of day, if the number of transactions increases within a certain period of time, resource contention such as CPU core contention occurs, and processing time is extended even for business applications that are operating normally May occur.

  In this case, in the delay detection method of Patent Document 1, since the latest average time is compared with the previous average time, a case where the processing time is extended even in a normally operating business application is determined as a delay. Resulting in. As a result, there is a problem that it is impossible to distinguish and detect delays caused by business application problems.

  The present invention has been made in view of the above problems, and its purpose is to solve the problem of business applications without being affected by the processing time that varies depending on the time even if the number of transactions varies depending on the time zone. It is to be able to detect a delay caused by.

  The delay detection device according to the present invention includes: an average execution time of each business application; and a weighted average time of the execution time according to a difference between a current time and a time when the business application was last executed. The execution rate of the execution time is calculated from the business execution means to be calculated, the average time and the latest weighted average time, and based on the expansion rate of the business application, the average of the expansion rate, and the standard deviation A business application delay detection apparatus comprising: business monitoring means for determining whether there is a delay in the business application.

  The delay detection method according to the present invention includes: an average time of execution times of individual business applications; and a nearest weighted average time of the execution times according to a difference between a current time and a time when the business application was last executed. Calculating an expansion rate of the execution time from the average time and the nearest weighted average time, and based on the expansion rate of the business application, the average of the expansion rate and the standard deviation of the business application This is a business application delay detection method for determining the presence or absence of a delay.

  The delay detection program according to the present invention includes: an average execution time of individual business applications; and a nearest weighted average time of the execution times according to a difference between a current time and a time when the business application was last executed. Calculating the expansion rate of the execution time from the processing to calculate, the average time and the most recently weighted average time, and based on the expansion rate of the business application, the average and standard deviation of the expansion rate A business application delay detection program that executes a process for determining whether there is an application delay.

  According to the present invention, even in a system in which the number of transactions varies depending on a time zone, it is possible to detect a delay due to a problem of a business application without being affected by a processing time that varies depending on time.

It is a figure which shows the structure of the transaction execution apparatus which performs the delay detection of embodiment of this invention. It is a flowchart which calculates the latest weighted average time and average time in the delay detection method of embodiment of this invention. It is a flowchart which determines the delay of a business application by calculating the expansion rate and standard deviation of processing time in the delay detection method of the embodiment of the present invention. It is an example of the work average time list in the delay detection method of the embodiment of the present invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. However, the preferred embodiments described below are technically preferable for carrying out the present invention, but the scope of the invention is not limited to the following.

  FIG. 1 shows a configuration of a transaction execution apparatus having a function for detecting a delay of a business application according to an embodiment of the present invention. The transaction execution apparatus 100 includes the business execution unit 110, the business monitoring unit 120, and the business average time list 130, so that the business application delay can be detected. The transaction execution device 100 can use a CPU (Central Processing Unit) or a server device having a memory, and by incorporating a delay detection program, the CPU computational resources and memory storage resources can be reduced. The business execution means 110, the business monitoring means 120, and the business average time list 130 can be configured.

  The business execution unit 110 includes a latest weighted average time calculation unit 111 and an overall average time calculation unit 112. The latest weighted average time calculating unit 111 calculates the latest weighted average time of the execution times of the individual business applications, and the overall average time calculating unit 112 calculates the average time of the execution times of the individual business applications.

  The work monitoring unit 120 includes a standard deviation calculation unit 121 and a delay determination unit 122. The standard deviation calculation unit 121 calculates the expansion rate of each business application process, and the average value and standard deviation of the business application expansion rate. The delay determination unit 122 determines whether or not there is a delay in processing of each business application based on the expansion rate obtained by the standard deviation calculation unit 121, the average value of the expansion rate, and the standard deviation.

  The business average time list 130 stores business types of business applications, the number of executions of business application processing, the latest weighted average time, the average time, and the last execution time.

  First, calculation of the average time and the latest weighted average time will be described with reference to the flowchart of FIG. FIG. 2 shows the business application processing delay detection method according to the embodiment of the present invention, in which the business execution means 110 receives the business application effective command, executes the business application, terminates it, and saves it in the business average execution time list 130. The flowchart which calculates average time and the latest weighted average time is shown.

  First, upon receiving a business application start command (S101), the business execution unit 110 executes the designated business application (S102). Next, the business application is terminated (S103), and then the business execution time is calculated from the business start time and the business end time (S104).

Next, the execution count and average time of the business application in the business average execution time list 130 are updated using the calculated business execution time (S105). The execution count is updated to the execution count obtained by adding 1 to the previous execution count. The average time is calculated and updated by the following formula (Formula 1).
A ′ (n) = {A ′ (n−1) × (n−1) + V} / n (Formula 1)
Here, n is the current business execution count, V is the current business execution time, and A ′ (n) is the average time until this time (nth).

Next, the latest weighted average time is calculated and updated using the calculated task execution time (S106). The latest weighted average time is calculated and updated by the following formula (Formula 2) or (Formula 3). In the following calculation formula, A (n) is the latest (n-th) weighted average time, P is a weighting factor, and t is the elapsed time from the last execution time of the job average time list 130 to the current job start time. .
If n <1 / (1-P),
A (n) = {A (n−1) × (n−1) × P ^t + V} / {(n−1) × P ^t +1} (Formula 2)
If n ≧ 1 / (1-P),
A (n) = A (n-1) * ^Pt + V * (1- ^Pt ) (Formula 3)
P can be arbitrarily set by the user. t is a value for raising the power of the elapsed time to the weighting coefficient. For example, if the final execution time is 3 [s] and the current job start time is 5 [s], P ^(5-3) is set. The longer the elapsed time, the larger the weighting factor for the most recent execution time. The calculation of the elapsed time can be arbitrarily set by the user in units of [s] (seconds) or [ms] (milliseconds).

  Finally, the current business end time is updated to the final execution time of the business application (S107), and the process ends.

  Subsequently, from the average time calculated in FIG. 2 and the most recently weighted average time, the growth rate of each business application and the average value and standard deviation of the growth rates of all business applications are calculated. A method for determining the delay of the business application using the standard deviation will be described with reference to the flowchart of FIG.

  FIG. 3 shows a method for detecting a delay in business application processing according to the embodiment of the present invention. The business monitoring means 120 uses the average time for each business and the most recently weighted average time stored in the business average execution time list 130 to The flowchart which calculates the expansion rate of the processing time for every business application, the average value and standard deviation of the expansion rate of the said business application, and determines the delay of each business application is shown.

First, the task monitoring unit 120 calculates and updates the processing time extension rate according to the following (Equation 4) using the latest weighted average time and the average time for each type of task stored in the task average time list 130. (S201).
Elongation rate (K) = nearest weighted average time / average time (Formula 4)
Next, it is determined whether or not the number of types of business whose previous business end time (final execution time) is within T seconds from the current business end time is m or more (S202).

  Here, T seconds is an effective time and can be arbitrarily set by the user. By removing the data before T seconds from the standard deviation calculation target, the standard deviation can be calculated from the expansion rate in the same time zone. Further, m is an effective number and can be arbitrarily set by the user. Since the standard deviation of the expansion rate is calculated in the case of m or more, the higher the number of m, the higher the standard deviation can be calculated.

  When S202 is No, the process is terminated without calculating the standard deviation.

When S202 is Yes, the standard deviation is calculated according to the following (formula 5) (S203). At that time, a business whose last execution time of the business application last time has exceeded T seconds from the current business end time is excluded from the standard deviation calculation target.
Standard deviation σ = {(Σ (K−L) ² ) / s} ^1/2 (Formula 5)
Here, K is the business growth rate, L is the average value of the business growth rates for which the standard deviation is calculated, and s is the number of business operations for which the standard deviation is calculated.

  Next, from the calculated standard deviation σ and the expansion rate K of each business, whether or not there is a delay in the business is determined depending on whether or not K is a value within the range of L ± Qσ (Q is an arbitrary real number). (S204). In other words, if K is within the range of L ± Qσ (Yes), it is determined that there is no delay in the business process (S205). If K is outside the range of L ± Qσ (No), it is determined that there is a delay (S206).

  Here, L ± Qσ is a range in which a certain proportion of the expansion rate of all the work falls within this range. For example, assuming that Q = 2.17, 97% of the growth rate of all businesses falls within the range of L ± 2.17σ. Then, it is determined that there is no delay for business applications whose expansion rate is within this range, and it is determined that there is a delay for business applications outside this range. The value of Q can be arbitrarily set by the user each time according to the execution status of the business application.

  Further, in the case of the delay determination, the expansion rate K takes a value larger than the average L of the expansion rate, so that it can be determined that there is a delay when K is larger than L + Qσ.

  The business application delay detection method according to the present embodiment described with reference to FIGS. 2 and 3 will be described below specifically using numerical examples.

  FIG. 4 shows the transition of the task average time list 130 when delay detection is performed according to the flowcharts of FIGS. 2 and 3. In FIG. 4, there are 10 types of business applications A to J, and the business A to J is between the state << 1 >> (time 64 [s]) and the state << 2 >> (time 65 [s]). It shows the transition of the average task time list when executed. The task average time list 41 in the state << 1 >> is updated to the task average time list 42 in the state << 2 >> by executing the tasks A to J shown in the task execution time list 40. At this time, it is assumed that, among the tasks A to J, the task H is not executed, and tasks other than the task H are each executed once.

  In state << 1 >>, the average rate of expansion is 1.25 and the standard deviation is 0.07, and the expansion rates of all business applications other than the business H are when Q is 2.17. 1.25 ± 0.07 × 2.17 = 1.9981 to 1.4019. Therefore, it is determined that there is no delay due to the problem of the business application.

  The business average time list 41 in the state << 1 >> is updated using each processing time in which each business other than H shown in the business execution time list 40 is executed once. This example will be described in detail below using business C and business I. Note that the weighting coefficient P is set to 0.8 and the elapsed time t is set in [s] units.

First, a case where the average time of work C and the latest weighted average time are calculated is shown.
The average time A ′ (n) is calculated as follows using (Equation 1).
A ′ (41) = (A ′ (40) × 40 + V) / 41
= (110 × 40 + 400) / 41
= 117
In this case, the latest weighted average time A (n) is calculated as follows using (Equation 3). Since the final execution time of the business C is 61 [s] and the business start time is 64 [s], the elapsed time is weighted using the elapsed time 3 [s].
A (41) = A (40) * ^Pt + V * (1- ^Pt )
= 150 × 0.8 ^(4-1) + 400 × (1-0.8 ^(4-1) )
= 272
The calculated average time 117 [ms] and the latest weighted average time 272 [ms] are stored in the business average time list 42.

Next, the case where the average time of the work I and the latest weighted average time are calculated is shown.
The average time A ′ (n) is calculated as follows using (Equation 1).
A ′ (31) = (A ′ (30) × 30 + V) / 31
= (100 × 30 + 160) / 31
= 102
In this case, the latest weighted average time A (n) is calculated as follows using (Equation 3). The elapsed time is 64 [s] for the final execution time of the job C and 64 [s] for the job start time, so that the elapsed time is 1 or less, but if it is 1 [s] or less, it is weighted using 1 immediately. To do.
A (31) = A (30) * ^Pt + V * (1- ^Pt )
= 130 * 0.8 + 160 * (1-0.8)
= 136
The calculated average time 102 [ms] and the latest weighted average time 136 [ms] are stored in the business average time list 42.

  Similar to the above-described examples of tasks C and I, the result of calculating the average time and the latest weighted average time of other tasks is the task average time list 42 of the state << 2 >>. Using this, the extension rate of the business type is calculated from the latest weighted average time and the average time, and the delay of the business application is determined using the standard deviation. The effective time T is 60 [s], and the effective number m is 5.

  First, the expansion rate for each business type is calculated by (Equation 4).

  Next, the number of business types whose final execution time is within 60 seconds (valid time) of the current time is counted, and the standard deviation is calculated for 9 business operations excluding the business operation H. The job H is excluded because the final execution time is 4 [s] and the current time is 65 [s] and there is an opening of 60 seconds or more.

  When the average expansion rate and standard deviation are calculated from the expansion rate for each business type, the average expansion rate = 1.38 and the standard deviation = 0.36. From the calculated average expansion rate and standard deviation, an expansion rate that is greater than 1.38 + 2.17 × 0.36 = 2.16 is determined to have a delay. In this case, it is determined that there is a delay because the business C has an expansion rate of 2.32.

  The business application delay detection program according to the present embodiment is a delay detection program that causes the transaction execution device that performs the delay detection of the business application shown in FIG. 1 to execute the delay detection method of the flowcharts of FIGS.

  According to the present embodiment, even in a system in which the number of transactions varies depending on the time zone, the extension rate and standard deviation of the execution time of the entire business in the same time zone are obtained, and the standard deviation exceeds a predetermined range. Therefore, it is possible to detect a delay caused by a problem in a business application without being affected by a processing time that varies depending on the time.

  The present invention is not limited to the above-described embodiment, and various modifications are possible within the scope of the invention described in the claims, and it is also included within the scope of the present invention. Not too long.

  Moreover, although a part or all of said embodiment may be described also as the following additional remarks, it is not restricted to the following.

Appendix (Appendix 1)
Business execution means for calculating an average time of execution times of individual business applications and a latest weighted average time of the execution time according to a difference between a current time and a time when the business application was last executed;
The expansion rate of the execution time is calculated from the average time and the most recent weighted average time, and the presence or absence of delay of the business application based on the expansion rate of the business application, the average of the expansion rate, and the standard deviation Business monitoring means for determining
Business application delay detection device equipped with.
(Appendix 2)
The business application delay according to claim 1, wherein the business execution means increases the weight of the latest execution time in the calculation of the nearest weighted average time as the difference between the current time and the last executed time is larger. Detection device.
(Appendix 3)
The business application delay detection device according to appendix 1 or 2, wherein the business monitoring means calculates the expansion rate by dividing the latest weighted average time by the average time.
(Appendix 4)
4. The business application delay detection device according to claim 1, wherein the business monitoring means identifies the business application having a delay.
(Appendix 5)
The business monitoring means, when the expansion rate of the business application is outside the range of L ± Qσ (Q is an arbitrary real number) in the relationship between the average L of the expansion rate and the standard deviation σ, 5. The business application delay detection device according to one of appendices 1 to 4, wherein the business application is determined to have a delay, or when the business application is within the range of L ± Qσ, the business application is determined to have no delay.
(Appendix 6)
Calculating the average time of the execution time of each business application and the latest weighted average time of the execution time according to the difference between the current time and the time when the business application was last executed;
The expansion rate of the execution time is calculated from the average time and the most recent weighted average time, and the presence or absence of delay of the business application based on the expansion rate of the business application, the average of the expansion rate, and the standard deviation Determine
Business application delay detection method.
(Appendix 7)
The business application delay detection method according to appendix 6, wherein the weight of the latest execution time in the calculation of the nearest weighted average time is increased as the difference between the current time and the last executed time is larger.
(Appendix 8)
The business application delay detection method according to appendix 6 or 7, wherein the expansion rate is calculated by dividing the latest weighted average time by the average time.
(Appendix 9)
9. The business application delay detection method according to claim 6, wherein the business application having a delay is specified from the expansion rate, the average of the expansion rates, and the standard deviation.
(Appendix 10)
If the expansion rate of the business application is outside the range of L ± Qσ (Q is an arbitrary real number) in the relationship between the average L of the expansion rate and the standard deviation σ, there is a delay in the business application 10. The business application delay detection method according to any one of appendices 6 to 9, wherein if there is no delay in the range of L ± Qσ, the business application is determined to have no delay.
(Appendix 11)
A process of calculating an average time of execution times of individual business applications and a latest weighted average time of the execution times according to a difference between a current time and a time when the business application was last executed;
The expansion rate of the execution time is calculated from the average time and the most recent weighted average time, and the presence or absence of delay of the business application based on the expansion rate of the business application, the average of the expansion rate, and the standard deviation The process of determining
Business application delay detection program that executes
(Appendix 12)
The business application delay detection program according to appendix 11, wherein as the difference between the current time and the last executed time is larger, the weight of the latest execution time in the calculation of the latest weighted average time is increased.
(Appendix 13)
The business application delay detection program according to appendix 11 or 12, wherein the expansion rate is calculated by dividing the latest weighted average time by the average time.
(Appendix 14)
14. The business application delay detection program according to one of appendices 11 to 13, wherein the business application having a delay is specified from the expansion rate, the average of the expansion rates, and the standard deviation.
(Appendix 15)
If the expansion rate of the business application is outside the range of L ± Qσ (Q is an arbitrary real number) in the relationship between the average L of the expansion rate and the standard deviation σ, there is a delay in the business application 15. The business application delay detection program according to one of appendices 11 to 14, wherein the business application causes the business application to determine that there is no delay when it is within the range of L ± Qσ.

DESCRIPTION OF SYMBOLS 100 Transaction execution apparatus 110 Business execution means 120 Business monitoring means 130 Business average time list 111 Latest weighted average time calculation means 112 Overall average time calculation means 121 Standard deviation calculation means 122 Delay determination means 40 Business execution time list 41, 42 Business average time list

Claims (10)

Business execution means for calculating an average time of execution times of individual business applications and a latest weighted average time of the execution time according to a difference between a current time and a time when the business application was last executed;
The expansion rate of the execution time is calculated from the average time and the most recent weighted average time, and the presence or absence of delay of the business application based on the expansion rate of the business application, the average of the expansion rate, and the standard deviation Business monitoring means for determining
Business application delay detection device equipped with. The business application according to claim 1, wherein the business execution unit increases the weight of the latest execution time in the calculation of the nearest weighted average time as the difference between the current time and the last executed time is larger. Delay detection device. The business application delay detection apparatus according to claim 1, wherein the business monitoring unit calculates the expansion rate by dividing the latest weighted average time by the average time. The business application delay detection apparatus according to claim 1, wherein the business monitoring unit identifies the business application having a delay. The business monitoring means, when the expansion rate of the business application is outside the range of L ± Qσ (Q is an arbitrary real number) in the relationship between the average L of the expansion rate and the standard deviation σ, 5. The business application delay detection device according to claim 1, wherein the business application is determined to have a delay, or if the business application is within the range of L ± Qσ, the business application is determined to have no delay. . Calculating the average time of the execution time of each business application and the latest weighted average time of the execution time according to the difference between the current time and the time when the business application was last executed;
The expansion rate of the execution time is calculated from the average time and the most recent weighted average time, and the presence or absence of delay of the business application based on the expansion rate of the business application, the average of the expansion rate, and the standard deviation Determine
Business application delay detection method. The business application delay detection method according to claim 6, wherein the weight of the latest execution time in the calculation of the nearest weighted average time is increased as the difference between the current time and the last executed time is larger. If the expansion rate of the business application is outside the range of L ± Qσ (Q is an arbitrary real number) in the relationship between the average L of the expansion rate and the standard deviation σ, there is a delay in the business application The business application delay detection method according to claim 6, wherein if there is no delay, the business application is determined to have no delay. A process of calculating an average time of execution times of individual business applications and a latest weighted average time of the execution times according to a difference between a current time and a time when the business application was last executed;
The expansion rate of the execution time is calculated from the average time and the most recent weighted average time, and the presence or absence of delay of the business application based on the expansion rate of the business application, the average of the expansion rate, and the standard deviation The process of determining
Business application delay detection program that executes If the expansion rate of the business application is outside the range of L ± Qσ (Q is an arbitrary real number) in the relationship between the average L of the expansion rate and the standard deviation σ, there is a delay in the business application 10. The business application delay detection program according to claim 9, wherein the business application is caused to determine that there is no delay when it is within the range of L ± Qσ.

Images