JP4883139B2 - Monitoring system performance measuring program, monitoring system performance measuring method, and monitoring system performance measuring apparatus - Google Patents

Description

  The present invention relates to a technique for measuring the performance of a monitoring system.

  For the purpose of extracting statistical information of an information system, bottleneck analysis, etc., a monitoring system that processes messages generated in the information system sequentially online and monitors requests has been put into practical use. The monitoring system must be capable of processing messages sequentially online even when the information system to be monitored is operating at maximum load. In view of this, a technique has been proposed in which a message generated in an information system is temporarily buffered, and the buffered message is output to the monitoring system all at once, thereby loading the monitoring system and measuring performance.

JP-A-6-83578

  By the way, the monitoring system has a characteristic that when the number of requests in progress increases, the amount of resources consumed by a CPU (Central Processing Unit), memory, and the like increases. However, in the conventionally proposed technique, since the messages are sequentially stored in the buffer without considering the message generation time, the “frequency” representing the number of requests per unit time is increased, but “ There was a problem that “multiplicity” did not change. For this reason, in the conventionally proposed technique, a load with an inaccurate relationship between frequency and multiplicity is used, and thus it is difficult to accurately measure the performance of the monitoring system.

  Therefore, in view of the conventional problems, the proposed technique aims to provide a technique that can measure the performance of a monitoring system using a load in which the relationship between the frequency of requests and the multiplicity is correct.

  For this reason, the proposed technique shows a correlation between the average execution time of the request, the first frequency, and the first multiplicity based on the result of buffering the message flowing from the information system to the monitoring system for a predetermined time. Ask. Further, the proposed technique obtains the second frequency based on the transmission time and the number of requests for sending the buffered message to the monitoring system, and obtains the second multiplicity corresponding to the second frequency based on the mutual relationship. Ask. And this proposal technique repeats the following processes (1)-(3) until the difference of 1st multiplicity and 2nd multiplicity becomes below a predetermined value.

  (1) Buffering messages flowing from the information system to the monitoring system for a predetermined time by the number of times of buffering defined by the ratio of the second multiplicity to the first multiplicity, and interleaving each buffered message group Generated load.

(2) Substitute the second multiplicity for the first multiplicity.
(3) The second frequency is updated based on the transmission time and the number of requests sent to the monitoring system, and the second multiplicity corresponding to the second frequency is updated based on the mutual relationship.

  According to the proposed technique, the message flowing from the information system to the monitoring system is adjusted so that the relationship between the request frequency and the multiplicity is correct, so that the performance of the monitoring system can be accurately measured. Further, according to the proposed technique, the performance of the monitoring system is measured using an actual message, so that it is possible to cope with a change in request load caused by, for example, a change in information system hardware or software. it can.

It is explanatory drawing of the request performed with an information system. It is a relationship diagram between request frequency and multiplicity. It is a system configuration figure showing an example of a performance measuring device. It is a functional block diagram which shows an example of the various functions in a performance measuring device. It is a schematic diagram of the performance measurement procedure of a monitoring system. It is a flowchart of the performance measurement process of a control module. It is a flowchart of the buffering process of a buffering module. It is a flowchart of the present load calculation process of a request recognition module. It is explanatory drawing of the present load calculation procedure. It is a flowchart of the frequency measurement process of a buffering module. It is a flowchart of the multiplicity adjustment process of a buffering module. It is explanatory drawing of the interleaving method of a message group. It is a flowchart of a thinning process of a buffering module. It is a flowchart of the link process of a request recognition module. It is explanatory drawing of the load production | generation method of the multiplicity N. It is explanatory drawing of the load production | generation method of multiplicity 1.5. It is explanatory drawing of the convergence state of multiplicity.

Hereinafter, the proposed technique will be described in detail with reference to the accompanying drawings.
First, consider the relationship between request frequency and multiplicity.
In the information system, as shown in FIG. 1, it is assumed that requests 1 to N (N: natural number) are executed during time ΔT [seconds]. In this case, the request frequency F [request / second] is F = N / ΔT. Further, when the execution time of the request n is represented by “time (n)”, the multiplicity M of the requests is represented by the following expression.

  When the relational expression between the request frequency F and the multiplicity M is expressed as M = R × F using the conversion constant R, and the specific numerical values of the frequency F and the multiplicity M are substituted into this relational expression, It becomes like the following formula.

  Referring to this equation, it can be understood that the conversion constant R is the average execution time [seconds / request] of the requests 1 to N executed during the time ΔT. Therefore, if the average execution time of the requests executed during the predetermined time is obtained, the multiplicity corresponding to the request frequency F is obtained from the relationship of multiplicity M = average execution time R × frequency F as shown in FIG. It can be understood that M can be calculated.

Next, an embodiment of the proposed technique will be described.
FIG. 3 shows an example of a performance measuring apparatus that embodies the proposed technique.
A business system 10 as an example of an information system is connected to a monitoring system 30 via a performance measurement device 20 constructed using a computer. When executing a request from a user, the business system 10 issues a message group 40 including a plurality of messages that start with a request message and end with a response message. Each message of the message group 40 includes an attribute value related to the request so that the request, request message, and response message that are issued can be specified. The performance measurement device 20 takes in a message issued by the business system 10, generates a load for measuring the performance of the monitoring system 30, sends the load to the monitoring system 30 at a stretch, and measures the performance. The monitoring system 30 monitors a request issued by the business system 10 and performs, for example, statistical information extraction and bottleneck analysis of the business system 10.

  The performance measurement device 20 implements a control module 22, a buffering module 24, and a request recognition module 26 by executing a performance measurement program installed in a storage such as a hard disk from a computer-readable recording medium.

  The control module 22 controls the entire performance measuring apparatus 20, and specifically includes a control unit 22A for controlling the buffering module 24 and the request recognition module 26 as shown in FIG.

  The buffering module 24 performs buffering of messages flowing from the business system 10 to the monitoring system 30, measurement of request frequency, load generation of target multiplicity, load output, and the like. Specifically, as shown in FIG. 4, the buffering module 24 includes a buffer 24A, a buffering unit 24B, a load generation unit 24C, a load sending unit 24D, and a performance measurement unit 24E. The buffer 24A is composed of, for example, a memory or a file, and temporarily stores a message or the like. When the buffering unit 24B receives a message issued by the business system 10, the buffering unit 24B stores the message in the buffer 24A in the order of reception. The load generation unit 24C generates a target multiplicity load using the message stored in the buffer 24A. The load sending unit 24D sends the load generated by the load generating unit 24C to the monitoring system 30 at once. The performance measurement unit 24E measures the performance of the monitoring system 30 by monitoring the load transmitted by the load transmission unit 24D, that is, the request transmission rate.

  The buffering module 24 controlled by the control module 22 determines the second multiplicity corresponding to the second frequency and means, and the difference between the first multiplicity and the second multiplicity. It functions as a step and means for repeating the process until it becomes a predetermined value or less.

  The request recognition module 26 associates the message buffered by the buffering module 24 for each request, and calculates the average execution time R, frequency F0, and multiplicity M0 of the request as the current load. Specifically, as shown in FIG. 4, the request recognition module 26 includes a linking unit 26A and a current load calculation unit 26B. The associating unit 26A associates a message stored in the buffer 24A of the buffering module 24 for each request. The current load calculation unit 26B calculates the current load of the request linked by the linking unit 26A, that is, the average execution time R, the frequency F0, and the multiplicity M0. Here, for example, the method described in Japanese Patent Application No. 2008-207889 proposed by the present applicant can be applied to the message linking process by the linking unit 26A.

  The request recognition module 26 controlled by the control module 22 functions as a step and means for obtaining a correlation among the average execution time of requests, the first frequency, and the first multiplicity.

FIG. 5 shows an outline of the performance measurement procedure of the monitoring system 30.
In step 1 (abbreviated as “S1” in the figure, the same applies hereinafter), an operator who intends to measure the performance of the monitoring system 30 using the performance measuring device 20 buffers a message based on experience, for example. The time ΔT is determined. Further, the operator changes the message flow so that the message issued by the business system 10 flows into the monitoring system 30 via the performance measuring device 20. Then, the operator transmits a performance measurement command to the control module 22 of the performance measuring device 20 using the time ΔT as an argument.

  In step 2, the current load (average execution time R, frequency F0, and multiplicity) of requests executed in the business system 10 in cooperation with the control module 22, the buffering module 24, and the request recognition module 26 of the performance measuring apparatus 20. Check M0). That is, the modules cooperate to store a message flowing from the business system 10 to the monitoring system 30 in the buffer 24A, and link the message for each request to obtain the current load of the request.

  In step 3, the control module 22 and the buffering module 24 of the performance measuring apparatus 20 cooperate to measure the frequency Fi of requests when the message stored in the buffer 24A is sent to the monitoring system 30.

  In step 4, the control module 22, the buffering module 24, and the request recognition module 26 of the performance measuring device 20 cooperate to adjust the multiplicity of the load sent to the monitoring system 30. If the multiplicity has converged to the target multiplicity, the control module 22 ends the performance measurement. If the multiplicity has not converged to the target multiplicity, the control module 22 returns the process to step 3.

FIG. 6 shows a performance measurement process that is executed when the control module 22 receives a performance measurement command.
In step 11, the control unit 22 </ b> A transmits a current load confirmation request to the buffering module 24 in order to confirm the current load of the request being executed in the business system 10.

  In step 12, the control unit 22A determines whether or not the average execution time R, frequency F0, and multiplicity M0 of the request have been received as the current load. If the control unit 22A determines that the current load has been received, the process proceeds to step 13 (Yes). If it is determined that the current load has not been received, the reception determination process is repeated (No).

In step 13, the control unit 22A substitutes (sets) the multiplicity M0 for the current multiplicity Mi.
In step 14, the control unit 22 </ b> A transmits a frequency measurement request to the buffering module 24 in order to measure the frequency Fi of the request sent from the performance measuring device 20 to the monitoring system 30.

  In step 15, the control unit 22A determines whether or not the frequency Fi has been received. If the control unit 22A determines that the frequency Fi has been received, the process proceeds to step 16 (Yes), while if it determines that the frequency Fi has not been received, the reception determination process is repeated (No).

In step 16, the control unit 22A calculates a target multiplicity Mj corresponding to the frequency Fi using a relational expression of multiplicity M = average execution time R × frequency F (Mj = R × Fi).
In Step 17, the control unit 22A determines whether or not the absolute value (difference) of the difference between the current multiplicity Mi and the target multiplicity Mj is greater than a predetermined value ε. Here, the predetermined value ε is a threshold value for determining whether or not the current multiplicity Mi has converged to a predetermined range centered on the target multiplicity Mj. Decided in consideration. If the control unit 22A determines that the absolute value of the difference between the current multiplicity Mi and the target multiplicity Mj is greater than the predetermined value ε, the process proceeds to step 18 (Yes), while the current multiplicity Mi and the target multiplicity Mi If it is determined that the absolute value of the difference from the multiplicity Mj is equal to or less than the predetermined value ε, the process proceeds to step 21 (No).

In step 18, the control unit 22 </ b> A transmits a multiplicity adjustment request to the buffering module 24 in order to bring the load multiplicity close to the target multiplicity Mj.
In step 19, the control unit 22A determines whether or not a multiplicity adjustment completion notification has been received. If the control unit 22A determines that the multiplicity adjustment completion notification has been received, the process proceeds to step 20 (Yes), while if it determines that the multiplicity adjustment completion notification has not been received, the reception determination process is repeated. (No).

In Step 20, the control unit 22A substitutes (sets) the target multiplicity Mj for the current multiplicity Mi, and returns the processing to Step 14.
In step 21, the control unit 22 </ b> A outputs the frequency Fi as the maximum performance value of the monitoring system 30 because the current multiplicity Mi has converged to a predetermined range centered on the target multiplicity Mi.

FIG. 7 shows a buffering process executed when the buffering module 24 receives a current load confirmation request.
In step 31, the buffering unit 24B sequentially stores the messages in the buffer 24A while adding a time stamp to the messages flowing from the business system 10 to the monitoring system 30.

  In step 32, the buffering unit 24B determines whether or not a time ΔT has elapsed since the start of message buffering. If the buffering unit 24B determines that the time ΔT has elapsed since the start of buffering, the process proceeds to step 33 (Yes), whereas if it determines that the time ΔT has not elapsed, the process proceeds to step 31. Return to No (No).

  In step 33, the buffering unit 24B collectively transmits the buffered message (buffering message) to the request recognition module 26 in order to notify that the buffering is completed.

FIG. 8 shows a current load calculation process executed when the request recognition module 26 receives a buffering message.
In step 41, the associating unit 26A associates each message included in the buffering message for each request, and adds an identifier for identifying the request to each message.

  In step 42, the current load calculation unit 26B calculates the average execution time R, frequency F0, and multiplicity M0 of requests as the current load. That is, the current load calculation unit 26B obtains the number of requests N linked in step 41, and obtains the execution time of each request by referring to the request message of each request and the time stamp of the response message. Further, the current load calculation unit 26B obtains the average execution time R of requests by dividing the integrated value obtained by integrating the execution times of the requests by the number of requests N, and divides the number of requests N by the time ΔT. The frequency F0 is obtained. Then, the current load calculation unit 26B obtains a multiplicity M0 corresponding to the frequency F0 using a relational expression of multiplicity M = average execution time R × frequency F.

In step 43, the current load calculation unit 26B transmits the current load (average execution time R, frequency F0, and multiplicity M0) to the control module 22.
According to the buffering process and the current load calculation process, as shown in FIG. 9, a message flowing from the business system 10 to the monitoring system 30 is buffered with a time stamp (t11 to t24) added for a time ΔT. 24A is sequentially stored. When the message storage is completed, the message included in the buffering message is linked for each request. Then, the average execution time R, frequency F0, and multiplicity M0 are obtained as the current load of the request from the message linking result. Accordingly, in the subsequent processing, the multiplicity corresponding to the frequency can be calculated using the relational expression M multiplicity M = average execution time R × frequency F.

FIG. 10 shows a frequency measurement process that is executed when the buffering module 24 receives a frequency measurement request.
In step 51, the performance measuring unit 24E starts measuring the transmission time ΔT ′ required for sending the message stored in the buffer 24A to the monitoring system 30.

  In step 52, the performance measurement unit 24E determines whether or not an unprocessed message remains in the buffer 24A. If the performance measurement unit 24E determines that an unprocessed message remains, the process proceeds to step 53 (Yes), whereas if it determines that no unprocessed message remains, the process proceeds to step 56. Advance (No).

In step 53, the performance measurement unit 24E refers to the identifier added to the message and counts the number of requests N ′ transmitted to the monitoring system 30.
In step 54, the performance measurement unit 24E refers to the thinning flag added to the message by the process described later, and determines whether or not the message to be sent to the monitoring system 30 is out of the thinning target. If the performance measuring unit 24E determines that the message is not subject to thinning, the process proceeds to step 55 (Yes), while if the message is determined not to be subject to thinning, the process returns to step 52 (No ).

In step 55, the load sending unit 24 </ b> D sends a message to the monitoring system 30. Thereafter, the performance measurement unit 24E returns the process to step 52.
In step 56, the performance measurement unit 24E stops measuring the transmission time ΔT ′.

In step 57, the performance measurement unit 24E calculates the request frequency Fi using the formula: Frequency Fi = number of requests N ′ / transmission time ΔT ′.
In step 58, the load sending unit 24D sends the thinned message to the monitoring system 30.

In step 59, the performance measurement unit 24 </ b> E transmits the frequency Fi to the control module 22.
According to the frequency measurement process, the time ΔT ′ required to send a message to which the thinning flag is not added to the monitoring system 30 among the plurality of messages stored in the buffer 24A is measured. Further, the identifier of each message stored in the buffer 24A is sequentially referred to, and the number of requests N ′ sent to the monitoring system 30 is counted. Then, the request frequency Fi is calculated by dividing the number of requests N ′ by the time ΔT ′. On the other hand, the message with the thinning flag added is sent to the monitoring system 30 to be processed by the monitoring system 30 after the measurement of the frequency Fi is completed.

FIG. 11 shows multiplicity adjustment processing that is executed when the buffering module 24 receives a multiplicity adjustment request.
In step 61, the load generation unit 24C calculates the number N of times that the message is buffered in order to generate a load corresponding to the target multiplicity Mj. That is, the load generation unit 24C calculates a minimum integer that is equal to or greater than the ratio (Mj / Mi) of the target multiplicity Mj to the current multiplicity Mi, and sets this as the buffering count N.

In step 62, the buffering unit 24B sequentially stores the messages flowing from the business system 10 to the monitoring system 30 in the buffer 24A for the time ΔT.
In step 63, the load generation unit 24C determines whether or not message buffering has been executed N times. If the load generation unit 24C determines that buffering has been performed N times, the process proceeds to step 64 (Yes), whereas if it is determined that buffering has not been performed N times, the process proceeds to step 62. Return (No).

  In step 64, the load generation unit 24C interleaves the buffered N message groups. That is, as shown in FIG. 12, when the load generation unit 24C executes buffering twice, the multiplicity is obtained by sequentially extracting and arranging the messages from the two message groups 40 stored in the buffer 24A. Two loads are generated.

In step 65, the load generation unit 24 </ b> C transmits a message association request to the request recognition module 26.
In step 66, the load generation unit 24C determines whether or not a linking completion notification has been received. If the load generation unit 24C determines that the association completion notification has been received, the process proceeds to step 67 (Yes), while if it determines that the association completion notification has not been received, the reception determination process is repeated ( No).

  In step 67, the load generator 24C determines whether or not the ratio (Mj / Mi) of the target multiplicity Mj to the current multiplicity Mi is a real number. If the load generation unit 24C determines that the ratio is a real number, the process proceeds to step 68 (Yes), while if it determines that the ratio is not a real number, the process proceeds to step 69 (No).

In step 68, the load generation unit 24C calls a thinning process subroutine for determining a request to be thinned out from a load of multiplicity N.
In step 69, the load generation unit 24 </ b> C transmits a multiplicity adjustment completion notification to the control module 22.

FIG. 13 shows a thinning process executed by the buffering module 24.
In step 71, the load generation unit 24C selects one request in order from the load of multiplicity N that is associated with each request.

In step 72, the load generation unit 24C generates random numbers in the range up to N.
In step 73, the load generating unit 24C determines whether or not a division value (Mi / Mj) obtained by dividing the current multiplicity Mi by the target multiplicity Mj is larger than a random value. If the load generation unit 24C determines that the division value is greater than the random value, the process proceeds to step 74 (Yes), whereas if the load generation unit 24C determines that the division value is equal to or less than the random value, the process proceeds to step 75. (No).

In step 74, the load generation unit 24C adds a thinning flag indicating that it is a thinning target to each message of the request selected in step 71.
In step 75, the load generation unit 24C determines whether or not all requests included in the load have been processed. If the load generation unit 24C determines that all requests have been processed, the process returns to the main routine (Yes), whereas if it determines that all requests have not been processed, the process returns to step 71 (No). ).

FIG. 14 shows a linking process executed when the request recognition module 26 receives a message linking request.
In step 71, the associating unit 26A associates each message included in the load of multiplicity N for each request, and adds an identifier for identifying the request to each message.

In step 72, the linking unit 26 </ b> A transmits a linking completion notification to the buffering module 24.
According to such multiplicity adjustment processing, thinning-out processing, and association processing, an integer that is equal to or greater than the ratio (Mj / Mi) of the target multiplicity Mj to the current multiplicity Mi is obtained, and this is set as the buffering count N. A message flowing from the business system 10 to the monitoring system 30 is repeatedly buffered by the number of times of buffering N, and as shown in FIG. 15, N message groups 40 are interleaved to generate a load of multiplicity N. In addition, a message included in the load of multiplicity N is linked for each request, and if the ratio of the target multiplicity to the current multiplicity Mi is a real number, a message to be thinned out is determined using a random number. The

Here, the multiplicity adjustment method will be described using mathematical expressions.
Assuming that the average execution time R, the number N of requests, and the multiplicity M0 of the requests are obtained as a result of buffering the message for the time ΔT, the following relationship is established among them.

  When the buffered message is sent to the monitoring system 30 at once, the frequency of requests increases, so that the average execution time R ′ of requests processed by the monitoring system 30 is smaller than the average execution time R (R ′ <R). Therefore, if the request transmission time is ΔT ′ (ΔT ′ <ΔT), the following relationship is established. For example, when R ′ = R / 2, ΔT ′ = ΔT / 2, so M0 = R · N / ΔT = R ′ · N / ΔT ′.

  Therefore, assuming that the target multiplicity Mj is X times the multiplicity M0, the target multiplicity Mj can be expressed by the following equation.

  That is, by interleaving the current multiplicity message by buffering it X times, the number of requests N ′ = X · N, the average request execution time R ′ = X · R, and the sending time ΔT ′ = X · ΔT. Therefore, as a result, it can be understood that data of multiplicity × X can be generated.

  According to the performance measuring apparatus 20 described above, the average execution time R, frequency F0, and multiplicity M0 of the request as the current load are obtained based on the result of buffering the message for the time ΔT. Further, the request frequency Fi is obtained based on the result of actually sending the buffering message to the monitoring system 30. Then, a load is generated by interleaving the buffered message group a predetermined number of times so that the current multiplicity Mi approaches the target multiplicity Mj corresponding to the request frequency Fi, and the load is sent to the monitoring system 30 at once. Performance is measured.

  At this time, if the ratio (Mj / Mi) of the target multiplicity Mj to the current multiplicity Mi is a real number, that is, if X is a real number, requests included in the load are thinned using random numbers as follows. . For example, when a load of multiplicity × 1.5 is generated, a load of multiplicity × 2 is generated by interleaving two message groups 40 including two requests as shown in FIG. One request (for example, request 3) determined according to a predetermined rule is thinned out.

  Then, the measurement of the frequency Fi and the adjustment of the multiplicity are repeated until the absolute value of the difference between the current multiplicity Mi and the target multiplicity Mj becomes equal to or less than the predetermined value ε. That is, as shown in FIG. 17, the processing is repeated until the target multiplicity Mj corresponding to the frequency Fi obtained from the message buffering result converges within a predetermined range centered on the current multiplicity Mi.

  Therefore, since the message flowing from the business system 10 to the monitoring system 30 is adjusted by the performance measurement device 20 so that the relationship between the frequency of requests and the multiplicity is correct, the performance of the monitoring system 30 should be measured with high accuracy. Can do. Moreover, since the performance of the monitoring system 30 is measured using an actual message, for example, it is possible to cope with a change in request load accompanying a change in hardware or software of the business system 10.

  Regarding the above embodiment, the following additional notes are disclosed.

  (Supplementary note 1) Monitoring from the information system to a computer interposed between an information system that issues a plurality of messages in response to execution of a request and a monitoring system that processes messages issued by the information system online Buffering a message flowing into the system for a predetermined time in a buffer, and obtaining a correlation between the average execution time of the request, the first frequency, and the first multiplicity from the buffering result of the message; Obtaining a second frequency based on the transmission time and number of requests for sending the ringed message to the monitoring system, obtaining a second multiplicity corresponding to the second frequency based on the correlation; and Until the difference between the multiplicity of the first multiplicity and the second multiplicity is equal to or less than a predetermined value. Buffering messages flowing from the information system to the monitoring system for a predetermined time by the number of times of buffering defined by the multiplicity ratio of 2, and generating a load that interleaves each buffered message group; While substituting the second multiplicity for the first multiplicity, the second frequency is updated based on the transmission time and the number of requests for transmitting the load to the monitoring system, and the second multiplicity is determined based on the correlation. A step of repeating the process of updating the second multiplicity corresponding to the frequency, and a performance measurement program of the monitoring system for realizing.

  (Supplementary Note 2) The step of repeating the process until the difference between the first multiplicity and the second multiplicity is equal to or less than a predetermined value is to buffer a minimum integer equal to or greater than the real number when the buffering count is a real number. A message is buffered as the number of times, a request to be thinned out from the load is determined using a random number, and a message related to a request other than a request to be thinned out among requests included in the load is sent to the monitoring system The monitoring system performance measurement program according to attachment 1.

  (Supplementary note 3) The monitoring system performance measurement program according to supplementary note 2, wherein the message related to the request to be thinned is sent to the monitoring system after the second multiplicity is updated.

  (Supplementary note 4) The monitoring system performance measurement program according to any one of supplementary notes 1 to 3, wherein the mutual relationship is represented by a relational expression of multiplicity = average execution time × frequency.

  (Supplementary Note 5) The monitoring system performance measurement program according to any one of Supplementary Note 1 to Supplementary Note 4, wherein the request is specified by associating the buffered message.

  (Supplementary note 6) A computer interposed between an information system that issues a plurality of messages in response to execution of a request and a monitoring system that processes messages issued by the information system online monitors from the information system Buffering a message flowing into the system for a predetermined time in a buffer, and obtaining a correlation between the average execution time of the request, the first frequency, and the first multiplicity from the buffering result of the message; Obtaining a second frequency based on the transmission time and number of requests for sending the ringed message to the monitoring system, obtaining a second multiplicity corresponding to the second frequency based on the correlation; and Until the difference between the multiplicity of the first multiplicity and the second multiplicity is equal to or less than a predetermined value. Buffering messages flowing from the information system to the monitoring system for a predetermined time by the number of times of buffering defined by the multiplicity ratio of 2, and generating a load that interleaves each buffered message group; While substituting the second multiplicity for the first multiplicity, the second frequency is updated based on the transmission time and the number of requests for transmitting the load to the monitoring system, and the second multiplicity is determined based on the correlation. A step of repeating the process of updating the second multiplicity corresponding to the frequency, and a method for measuring the performance of the monitoring system.

  (Supplementary note 7) A message flowing from an information system that issues a plurality of messages to a monitoring system that processes a message issued by the information system in response to execution of a request is buffered for a predetermined time, and the message is buffered Based on the result, a means for obtaining a correlation between the average execution time of the request, the first frequency, and the first multiplicity, a second based on the transmission time and the number of requests for sending the buffered message to the monitoring system. Until the difference between the first multiplicity and the second multiplicity is equal to or less than a predetermined value, and means for obtaining a second multiplicity corresponding to the second frequency based on the correlation And the monitoring system from the information system for the number of times of buffering defined by the ratio of the second multiplicity to the first multiplicity. The message flowing into the buffer is buffered for a predetermined time, a load is generated by interleaving the buffered message groups, and a second multiplicity is substituted for the first multiplicity, while the load is transferred to the monitoring system. Means for updating the second frequency based on the transmitted transmission time and the number of requests, and repeating the process of updating the second multiplicity corresponding to the second frequency based on the correlation. A monitoring system performance measurement device.

DESCRIPTION OF SYMBOLS 10 Business system 20 Performance measuring apparatus 22 Control module 22A Control part 24 Buffering module 24A Buffer 24B Buffering part 24C Load generation part 24D Load sending part 24E Performance measurement part 26 Request recognition module 26A Linking part 26B Current load calculation part 30 Monitoring System 40 message group

Claims (5)

A computer interposed between an information system that issues a plurality of messages as a request is executed and a monitoring system that processes messages issued by the information system online,
A message flowing from the information system to the monitoring system is buffered in a buffer for a predetermined time, and a correlation between the average execution time of the request, the first frequency, and the first multiplicity is obtained from the buffering result of the message. Steps,
Obtaining a second frequency based on a sending time and number of requests for sending the buffered message to a monitoring system, and obtaining a second multiplicity corresponding to the second frequency based on the correlation;
Until the difference between the first multiplicity and the second multiplicity is equal to or less than a predetermined value, the number of buffering times defined by the ratio of the second multiplicity to the first multiplicity is from the information system. A message flowing into the monitoring system is buffered in a buffer for a predetermined time, and a load obtained by interleaving the buffered message groups is generated, and a second multiplicity is substituted for the first multiplicity, while the load Updating the second frequency based on the transmission time and the number of requests sent to the monitoring system, and repeating the process of updating the second multiplicity corresponding to the second frequency based on the correlation;
Monitoring system performance measurement program to realize   The step of repeating the process until the difference between the first multiplicity and the second multiplicity is equal to or less than a predetermined value is such that when the buffering count is a real number, a message with a minimum integer equal to or greater than the real number as a buffering count The buffering is performed, a request to be thinned out from the load is determined using a random number, and a message related to a request other than a request to be thinned out among requests included in the load is transmitted to the monitoring system. 1. The performance measurement program for the monitoring system according to 1.   3. The monitoring system performance measurement program according to claim 2, wherein the message related to the request to be thinned is sent to the monitoring system after the second multiplicity is updated. A computer interposed between an information system that issues a plurality of messages in accordance with execution of a request and a monitoring system that processes messages issued by the information system online,
A message flowing from the information system to the monitoring system is buffered in a buffer for a predetermined time, and a correlation between the average execution time of the request, the first frequency, and the first multiplicity is obtained from the buffering result of the message. Steps,
Obtaining a second frequency based on a sending time and number of requests for sending the buffered message to a monitoring system, and obtaining a second multiplicity corresponding to the second frequency based on the correlation;
Until the difference between the first multiplicity and the second multiplicity is equal to or less than a predetermined value, the number of buffering times defined by the ratio of the second multiplicity to the first multiplicity is from the information system. A message flowing into the monitoring system is buffered in a buffer for a predetermined time, and a load obtained by interleaving the buffered message groups is generated, and a second multiplicity is substituted for the first multiplicity, while the load Updating the second frequency based on the transmission time and the number of requests sent to the monitoring system, and repeating the process of updating the second multiplicity corresponding to the second frequency based on the correlation;
A method for measuring the performance of a monitoring system, characterized in that A message that flows from an information system that issues a plurality of messages to a monitoring system that processes the messages issued by the information system in response to the execution of the request is buffered for a predetermined time. Means for determining a correlation between average execution time, first frequency and first multiplicity;
Means for determining a second frequency based on a transmission time and the number of requests for transmitting the buffered message to a monitoring system, and determining a second multiplicity corresponding to the second frequency based on the correlation;
Until the difference between the first multiplicity and the second multiplicity is equal to or less than a predetermined value, the number of buffering times defined by the ratio of the second multiplicity to the first multiplicity is from the information system. A message flowing into the monitoring system is buffered for a predetermined time, a load is generated by interleaving the buffered message groups, and a second multiplicity is substituted for the first multiplicity, while the load is monitored. Means for updating the second frequency based on the transmission time and the number of requests sent to the system, and repeating the process of updating the second multiplicity corresponding to the second frequency based on the correlation;
An apparatus for measuring the performance of a monitoring system, comprising:

Images