JP2009140358A - Performance arithmetic device and performance arithmetic processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To generate many tabulation section data for parameter identification of a performance model of an object device in a fixed time while maintaining the accuracy of the tabulation section data constant. <P>SOLUTION: A tabulation section calculation part 5 sets start time of a tabulation section i, and acquires usage rate information in a fixed time from the start time in measurement data. A processing time calculation part 4 estimates resource processing time information using the resource usage rate information and a no-load time processing time. The tabulation section calculation part 5 calculates time for the tabulation section i using processing time information R and an allowable error rate, and calculates end time of the tabulation section i. The calculation part 5 transmits the start time and end time of the tabulation section i and measurement data in between to a data generation part 6. The data generation part 6 generates tabulation section data based on the start time and end time of the tabulation section i and the measurement data in between by tabulating resource use time between the start time and end time of the tabulation section i and a generation quantity of each type of processing. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a performance calculation device and a performance calculation processing program for a computer system such as an Internet system or an intranet system.

Conventionally, in a computer system, the performance tends to gradually deteriorate as the amount of processing increases due to operator learning and the increase in the number of users, the amount of data increases, and the cache miss rate increases accordingly. Therefore, as disclosed in, for example, Patent Document 1 and Patent Document 2, an apparatus for monitoring the performance of the system is used.
JP 2007-26303 A JP 2005-157933 A

  In order to prevent system performance problems in advance, it is necessary to monitor the margin and processing amount of the system, catch the signs of performance degradation, and respond in advance.

  For each type of process, if the resource usage when the process is executed once can be estimated, the resource usage rate can be estimated according to the ratio at which each process is executed when the process is at its peak. it can. By periodically estimating the resource utilization rate, measures such as resource enhancement can be implemented before the performance deteriorates, and the performance deterioration can be prevented in advance.

  By the way, in a computer system in operation, it is difficult to stop the operation and measure the resource usage when one process is executed. Therefore, it is necessary to estimate the resource usage when one process is executed using data that can be measured during operation. In an operating system, a plurality of types of processes are executed simultaneously, and the number of processes executed for each type and the resource utilization rate can be measured within a certain period of time.

  If the resource usage when one process is executed is always a constant value, the resource usage for each process can be calculated from data that can be measured during operation by solving simultaneous equations. However, there is a problem in that it cannot be solved by simultaneous equations because the resource usage fluctuates due to a measurement error or a difference in parameters of processing.

  Therefore, for each type of processing, the resource usage when the processing is executed once is expressed by a probability distribution such as a normal distribution, and the computer system in operation using the average value and standard deviation as parameters. A method of parameter identification based on the number of occurrences for each type of processing and the resource utilization rate, which can be measured by the above method, can be considered.

  If this parameter, that is, the probability distribution of resource usage when each type of processing is executed once, can be successfully identified, the probability of resource usage will be adjusted to the rate at which each processing is executed at the peak processing amount. The distribution can be estimated.

  The number of occurrences and the resource usage rate for each type of processing are measured at predetermined time intervals. A collection of one or a plurality of measurement data at a certain time interval is used as aggregated section data and used for parameter identification calculation.

  In order to increase the parameter identification accuracy, it is necessary to accurately count the generation amount for each type of processing occurring in the counting section and the resource usage time used by those processing. In addition, as a result of previous parameter identification experiments, it has been found that the use of a large number of total interval data is indispensable for improving the accuracy of parameter identification. It is necessary to generate the total interval data.

  If the total section is simply divided in a short time, there is a problem that the section is divided while the generated process is using the resource, and the accuracy of the total section data is lowered. On the other hand, if the totaling section is divided for a long time, the ratio of the processing that occurs during the use of resources decreases and the accuracy of the totaling section data increases, but the totaling section that can be generated within the measurement data for a certain period Since the number of data becomes small, it is necessary to divide by an appropriate length.

  In addition, when the resource usage rate of the computer where the target system is installed is high, a waiting time occurs, so the processing time of each process becomes long, and the aggregation interval is divided by the same time as when the resource usage rate is low. If this is done, the accuracy of the aggregated section data will deteriorate.

  Accordingly, an object of the present invention is to provide a performance calculation device capable of generating a large amount of total interval data within a predetermined time while maintaining the accuracy of the total interval data for parameter identification of the performance model of the target device. And providing a performance calculation processing program.

  In other words, the performance calculation device according to the present invention is executed in a state where the measured value of each resource used by each process executed by the target device and the target device uses a specific resource and the resource does not compete with other resources. The processing time of each process executed by the target device is estimated based on the processing time information of each process to be performed, and based on the estimated processing time, the time of the aggregation section for parameter identification of the performance model of the target device Is calculated so that the time outside the section of the sum of the processing times of the processes generated in the section satisfies the allowable condition, and the utilization rate of each resource within the calculated time is totaled.

  According to the present invention, it is possible to generate a large amount of total section data within a predetermined time while keeping the accuracy of the total section data for parameter identification of the performance model of the target device constant.

Embodiments of the present invention will be described below with reference to the drawings.
(First embodiment)
First, a first embodiment of the present invention will be described.
FIG. 1 is a block diagram showing an example of the configuration of a performance calculation device according to the first embodiment of the present invention.
FIG. 2 is a diagram showing an example of an input / output relationship of various data handled by the performance calculation device according to the first embodiment of the present invention.
The performance calculation device according to the first embodiment of the present invention estimates the processing time of each process estimated from the resource utilization rate during operation of the computer on which the target system is mounted and the processing time at no load. In this case, it has a function of generating aggregate section data.

  As shown in FIG. 1, the performance calculation device according to the first embodiment includes a measurement data input unit 1, a no-load processing time input unit 2, an allowable error rate input unit 3, a processing time calculation unit 4, a totaling interval. A calculation unit 5, a data generation unit 6, a total section data output unit 7, a storage device 8, an input device 9 such as a keyboard and a mouse, and a display device 10 such as a display device are connected to each other via a bus 11.

  The storage device 8 is a storage medium configured by hardware such as a hard disk drive or a nonvolatile memory device, for example, and includes a measurement data input unit 1, a no-load processing time input unit 2, an allowable error rate input unit 3, and a processing time. In addition to storing a control program for operation by the calculation unit 4, the total interval calculation unit 5, the data generation unit 6, and the total interval data output unit 7, the measurement data storage unit 21, the processing time storage unit 22, and the error rate storage unit 23. And a total section data storage unit 24. The measurement data input unit 1, the no-load processing time input unit 2, the allowable error rate input unit 3, the processing time calculation unit 4, the totaling interval calculation unit 5, the data generation unit 6, and the totaling interval data output unit 7 are stored in the storage device 8. The process is executed by reading the stored control program.

  The measurement data input unit 1 receives performance data from the outside, measured by the measurement tool program, such as the CPU of the computer on which the target system is mounted, the utilization rate of resources such as disks, and the processing amount by processing type, This data is transmitted to the total section calculation unit 5.

The no-load processing time input unit 2 externally outputs the processing time of each process, that is, the processing time of each process measured in a state where the resources of the computer on which the target system is mounted do not compete with other processes. Is transmitted to the processing time calculation unit 4.
The allowable error rate input unit 3 receives the allowable error rate of the aggregate interval data to be generated from the outside, and transmits this to the aggregate interval calculation unit 5.

  The processing time calculation unit 4 receives the no-load processing time of each process from the no-load processing time input unit 2 and holds it in the processing time storage unit 22 of the storage device 8. The resource usage rate is received from the totaling interval calculation unit 5, the processing time of each process is calculated based on the received resource usage rate, and the calculated processing time is transmitted to the totaling interval calculation unit 5.

  The total section calculation unit 5 receives the measurement data from the measurement data input unit 1 and stores it in the measurement data storage unit 21 of the storage device 8. The storage device 8 uses the initial time of the stored measurement data as the start time of the total section. And the resource utilization rate of the computer on which the target system is mounted at this start time is transmitted to the processing time calculation unit 4.

  Further, the total section calculation unit 5 receives the allowable error rate from the allowable error rate input unit 3 and holds it in the error rate storage unit 23 of the storage device 8, and receives the processing time of each process from the processing time calculation unit 4. Based on the processing time received from the processing time calculation unit 4 and the allowable error rate received from the allowable error rate input unit 3, the time of the total section is calculated.

  Further, the total section calculation unit 5 obtains the total section end time by adding the time of the total section calculated as described above to the start time of the total section held in the storage device 8 as described above. It is confirmed whether measurement data up to this end time is held.

  Then, when holding the measurement data described above, the tabulation section calculation unit 5 transmits the start time and end time of the tabulation section and measurement data therebetween to the data generation unit 6, and calculates the end time as the next tabulation. The section start time is stored in the storage device 8, and the resource utilization rate within a predetermined time from this start time is transmitted to the processing time calculation unit 4.

When the data generation unit 6 receives the start time and end time of the total interval and the measurement data between them, from the total interval calculation unit 5, the data generation unit 6 totals the resource usage time and the processing amount therebetween, and the total interval data output unit as the total interval data 7 to send.
The total section data output unit 7 receives the total section data from the data generation unit 6 and outputs it to the display device 10.

Next, the operation | movement concerning total area data generation by the performance arithmetic unit comprised as mentioned above is demonstrated.
FIG. 3 is a flowchart showing an example of the processing operation of the performance calculation device according to the first embodiment of the present invention.
First, the measurement data input unit 1 inputs measurement data from the outside and transmits it to the totaling interval calculation unit 5. The totaling interval calculation unit 5 holds the received measurement data in the measurement data storage unit 21 of the storage device 8 (step S1).

  The no-load processing time input unit 2 inputs the no-load processing time from the outside and transmits it to the processing time calculation unit 4. The processing time calculation unit 4 holds the no-load processing time received from the no-load processing time input unit 2 in the processing time storage unit 22 of the storage device 8 (step S2).

  The allowable error rate input unit 3 inputs the allowable error rate of the tabulation interval data from the outside, and transmits this to the tabulation interval calculation unit 5. The tabulation section calculation unit 5 holds the allowable error rate received from the allowable error rate input unit 3 in the error rate storage unit 23 of the storage device 8 (step S3).

The aggregation section calculation unit 5 sets a variable i representing the number of the aggregation section data to 1 as an initial setting (step S4).
Time window calculation unit 5 sets the initial time T ₀ of the measurement data acquired in the process of step S1 to the start time Ts _i aggregation interval i (step S5), and from among the measurement data within a predetermined time after the start time Resource utilization rate information U (Ts _i ) is acquired and transmitted to the processing time calculation unit 4.

When the processing time calculation unit 4 receives the resource utilization rate information U (Ts _i ) from the total interval calculation unit 5, the processing time calculation unit 4 uses the no-load processing time from the allowable error rate input unit 3 to obtain the resource utilization rate information U ( ts _i) processing time when the information to estimate the R (Ts _i) (step S6), and transmits it to the time window calculation unit 5.

When receiving the processing time information R (Ts _i ) from the processing time calculation unit 4, the totaling interval calculation unit 5 uses the allowable error rate held in the error rate storage unit 23 of the storage device 8 and uses the allowable error rate. The time Tb _i is calculated, and the end time Te _i of the counting section i is calculated according to the following equation (1) (step S7).

Te _i = Ts _i + Tb _i (1)
This equation (1) means that the time after the time Tb _i from the start time Ts _i of the counting section i becomes the end time Te _i .

Time window calculation unit 5, if the end time Te _i aggregation interval i and compared to the last time T _f of the measurement data, later than the last time T _f of the end time Te _i measurement data time window i ( The processing in step S8 is completed.

On the other hand, if the end time Te _i of the total section i is before the final time T _{f of the} measurement data (NO in step S8), the total section calculation unit 5 starts the start time Ts _i and the end time Te _{i of the} total section _i. And the measurement data between them are transmitted to the data generation unit 6.

The data generation unit 6 aggregates the measured data received from the aggregation interval calculation unit 5 by aggregating the resource usage time between the start time Ts _i and the end time Te _{i of} the aggregation interval _i and the generation amount for each type of processing. Section data is generated (step S9), and is transmitted to the total section data output unit 7.
The total section data output unit 7 displays the total section data received from the data generation unit 6 on the display device 10 (step S10).

Then, the total section calculation unit 5 adds 1 to the number i of the total section data and updates it (step S11).
Time window calculation unit 5 sets the end time _{Te i-1} of the time window i-1 to the start time Ts _i aggregation interval i (step S12), the process returns to step S6.

  As described above, the performance calculation device according to the first embodiment of the present invention includes the pre-measured processing time of each process of the target system when there is no load, and the resource usage rate of the computer on which the target system is mounted. Then, the processing time of each process at the time of the resource utilization rate is estimated, and the aggregate interval data is determined using the estimated processing time of each process and the allowable error rate input from the outside. Therefore, while keeping the accuracy of the total section data constant, a lot of total section data can be generated within a certain time, and the accuracy of parameter identification of the performance model can be maintained high.

(Second Embodiment)
Next, a second embodiment of the present invention will be described. The configuration of the performance calculation apparatus according to the present embodiment is basically the same as that shown in FIG.
The performance calculation device according to the second embodiment has a function of generating aggregate section data when the processing time during operation of the target system is measured.
FIG. 4 is a block diagram showing an example of the configuration of the performance calculation device according to the second embodiment of the present invention.
FIG. 5 is a diagram showing an example of an input / output relationship of various data handled by the performance calculation device according to the second embodiment of the present invention.
As shown in FIG. 4, the performance calculation device according to the second embodiment of the present invention is compared with the performance calculation device according to the first embodiment of the present invention. While the processing time calculation unit 4 is not provided, a processing time input unit 31 is further provided.
The processing time input unit 31 receives the processing time of each process at the time of operation of the target system from the outside, and transmits it to the totaling interval calculation unit 5.
The total section calculation unit 5 receives the measurement data from the measurement data input unit 1 and holds it in the measurement data storage unit 21 of the storage device 8, and receives the processing time of each process during operation from the processing time input unit 31. The allowable error rate is stored in the processing time storage unit 22 of the storage device 8, received from the allowable error rate input unit 3, and stored in the error rate storage unit 23 of the storage device 8.

  Moreover, the total section calculation unit 5 stores the initial time of the stored measurement data in the storage device 8 as the start time of the total section, searches the processing time of each process at the start time, and the processing time and the allowable error rate Based on the above, the time of the total section is calculated.

  The total section calculation unit 5 obtains the total section end time by adding the calculated total section time to the start time of the total section held in the storage device 8 and stores the measurement data up to the end time. If it is held in the device 8, if it is held, the start time and end time of the totaling section and the measurement data between them are transmitted to the data generation unit 6, and the end time is started at the next totaling section. The time is held, and the resource utilization rate at that time is transmitted to the processing time calculator 4.

Next, the operation | movement concerning total area data generation by the performance arithmetic unit comprised as mentioned above is demonstrated.
FIG. 6 is a flowchart showing an example of the processing operation of the performance calculation device according to the second embodiment of the present invention.
First, as in the first embodiment, the measurement data input unit 1 inputs measurement data from the outside and transmits it to the totaling interval calculation unit 5. The totaling interval calculation unit 5 holds the received measurement data in the measurement data storage unit 21 of the storage device 8 (step S21).

When the processing time input unit 31 inputs the processing time of each process at the time of operation from the outside, the processing time input unit 31 transmits this to the totaling interval calculation unit 5. The total section calculation unit 5 holds the processing time information received from the processing time input unit 31 in the processing time storage unit 22 of the storage device 8 (step S22).
Thereafter, the processing from steps S3 to S5 described in the first embodiment is performed (steps S23 to S25).

Then, the counting section calculation unit 5 searches the processing time information R (Ts _i ) of the start time Ts _i of the counting section i, and uses the searched processing time and the retained allowable error rate to calculate the counting section i. Time Tb _i is calculated, and the end time Te _i of the counting section i is calculated according to the above-described equation (1) (step S26).
Thereafter, the processing from steps S8 to S12 described in the first embodiment is performed (steps S27 to S31).

  As described above, the performance calculation device according to the second embodiment of the present invention determines the aggregate interval data using the processing time of each process during the operation of the target system and the allowable error rate input from the outside. By doing so, while maintaining the accuracy of the total interval data, a large amount of the total interval data can be generated within a predetermined time, and the accuracy of parameter identification of the performance model can be maintained high.

  Examples of each embodiment will be described below.

  Example 1 will be described first. Example 1 is an example of the above-described first embodiment, and describes a total interval data generation method for performance model parameter identification.

  The target system described in this embodiment includes a server group such as a Web server, an application server, and a database server, and a client. The target system measures the following information as measurement data from these server groups and clients and inputs the measurement data to the measurement data input unit 1.

  The measurement data is resource utilization rate and processing amount information. The resource usage rate is a usage rate of resources such as a CPU and a disk of the server machine, and is measured at regular time intervals such as 10 seconds. The server group can be measured and output by a program tool such as a system monitor program performance monitor (perfmon.exe) of Microsoft Corporation.

  The processing amount information is a usage history obtained by the log function of the Web server or application server. There are two types of usage history: Each of these indicates the amount of processing given to the system at regular intervals.

  The first usage history, like an access log, is a record of the type of processing started in response to a request from the client and the time of occurrence. In the second usage history, the number of occurrences for each process in the section is recorded at regular intervals.

FIG. 7 is a diagram illustrating an example of measurement data according to the first embodiment in a table format.
The measurement data shown in FIG. 7 is obtained by measuring the processing amount of each process and the utilization rate of each resource at intervals of 10 seconds from midnight to 24:00.
In the following description, it is assumed that the resource utilization rate of the server computer and the processing amount of each process are obtained as measurement data from the initial time T ₀ to the final time T _f at regular time intervals ΔT.

As described in the first embodiment, the measurement data input to the measurement data input unit 1 is transmitted to the totaling interval calculation unit 5 and held in the totaling interval calculation unit 5. The tabulation section calculation unit 5 sets an initial value “1” for a variable i representing the number of the tabulation section. Then, the tally interval calculation unit 5 sets the initial time T ₀ of the acquired measurement data to the start time Ts _i of the tally interval i, and sets the resource utilization rate information U (Ts _i ) at that time to the processing time calculation unit 4. Send.

The resource utilization rate information U (Ts _i ) is expressed by the following equation (2).

U (Ts _i ) = {U ¹ (Ts _i ), U ² (Ts _i ),..., U ^K (Ts _i )} Equation (2)
Further, the utilization factor of the resource k (k = 1, 2,..., K; K represents the number of resources) between the start time Ts _i and the fixed time ΔT is represented as U ^k (Ts _i ).

The processing time calculation unit 4 uses the resource utilization rate information U (Ts _i ) received from the total section calculation unit 5 and the no-load processing time information R ₀ received from the no-load processing time input unit 2. resource utilization information to estimate U (Ts _i) the processing time information of each process in the case R (Ts _i).

The no-load processing time information _R0 is expressed by the following equation (3).

In each embodiment, it is assumed that the following expression (4) holds.

R _jk0 in equation (4) is _required for the process j (j = 1, 2,..., J) to be executed using the resource k (k = 1, 2,..., K) in the no-load state. It's time. J in Equation (3) is the number of types of processing.

As an example of the estimation method of the processing time information, the estimation formulas are shown in the following formulas (5) and (6).
R (Ts _i ) = {R ¹ (Ts _i ), R ² (Ts _i ),..., R ^J (Ts _i )} Equation (5)

R ^j (Ts _i ) in equation (6) is the processing time of processing that starts within a predetermined time ΔT from the start time Ts _i of processing j. Little official of these formulas queuing theory applied for each resource, an approximate expression to calculate the processing time of the process starting from the start time Ts _i in the processing j the addition to the predetermined time within [Delta] T.

The start time Ts _i vector by arranging the processing time of the processing for starting a predetermined time in the [Delta] T from R (Ts _i), referred to herein as processing time information of the processing starting from the start time Ts _i within a predetermined time [Delta] T.

The processing time calculation unit 4 transmits the calculated processing time information R (Ts _i ) to the total interval calculation unit 5.
When receiving the processing time information R (Ts _i ) from the processing time calculation unit 4, the totaling interval calculation unit 5 uses the allowable error rate ρ held in the error rate storage unit 23 of the storage device 8 to calculate the totaling interval i. And the end time of the counting section i are calculated.

The formula for calculating the total interval time is represented by the following formula (7) and the above-described formula (1).

The right side of Expression (7) is a value obtained by dividing the maximum value of {R ¹ (Ts _i ), R ² (Ts _i ),..., R ^J (Ts _i )}, which is the right side of Expression (5), by 2ρ. is there.
In this example, the time of the total section is calculated in accordance with the process having the longest processing time in each process, but the time of the total section may be obtained from the average value of the processing times of each process.
The derivation method of Expression (7) will be described in relation to the allowable error rate described later and the time of the totaling section.

Aggregate time window calculation unit 5 compares the final time T _f of the end time Te _i and the measurement data of the time window i, if the end time Te _i aggregation interval i is later than the last time T _f of the measurement data End the section data generation.

If the end time Te _i of the total section i is earlier than the final time T _{f of the} measurement data, the total section calculation unit 5 generates data of the start time Ts _i and end time Te _{i of} the total section _i and the measurement data therebetween. Transmit to unit 6.

From the received measurement data, the data generation unit 6 uses the resource utilization time between the start time Ts _i and the end time Te _{i of} the counting section i according to the following formulas (8), (9), and (10): Aggregate the amount of each process.

  The left side of Expression (8) is the total section data of the total section i, the left side of Expression (9) is the usage time of the resource k in the total section i, and the left side of Expression (10) is the processing j in the total section i. The amount of processing.

The data generation unit 6 transmits the total interval data S _i to the total interval data output unit 7. The total section data output unit 7 outputs this total section data to the display device 10.
Then, the counting section calculation unit 5 adds 1 to the number i of the counting section data and updates it, and sets the end time Te _i-1 of the counting section i _-1 as the start time Ts _i of the counting section i.
Thereafter, the process is repeated from the point where the resource utilization rate information R (Ts _i ) is transmitted to the processing time calculation unit 4, and aggregated section data is sequentially generated.

FIG. 8 is a diagram illustrating an example of the tabulation section data in the first embodiment in a table format.
The aggregate section data shown in FIG. 8 indicates the aggregate section number, the start time and end time of the aggregate section, the resource usage time (seconds) of each resource, and the processing amount of each process.

Next, the relationship between the allowable error rate ρ and the time Tb _{i of the} total section will be described.
In order to simplify the explanation, it is assumed that the processing time of each process is equal, that is, the following expression (11) holds.

R ^j (t) = R (j = 1, 2,..., J) (11)
FIG. 9 is a diagram illustrating an example of the relationship between the counting interval and the processing time.
In the example shown in FIG. 9, the time Tb _{i of the} counting section is three times the processing time R. Here, a section obtained by dividing the total section by the processing time interval of processing is referred to as a sub section, and sub sections of the total section i are referred to as sub sections A, B, and C as illustrated.

  The arrow in FIG. 9 represents the process, and the left end of the arrow is the process occurrence time, and the right end is the process end time. As can be seen from FIG. 9, the processing that occurred in the sub-intervals A and B ends within the aggregation interval i. On the other hand, as shown in FIG. 9, the processing that occurred in the sub-section C distinguished by diagonal lines in FIG. 9 is that the end time is after the end time of the counting section i. In other words, a part of the processing is the counting section i. It is out of the range.

  When the processing occurs uniformly, in the processing occurring in the sub-section C, on average, half of the processing time protrudes from the counting section i. Accordingly, the ratio of the resource usage time of all processes occurring in the total section i that protrudes from the total section i, that is, the error rate is 1/3 × 1/2 = 1/6. Similarly, when the time of the totaling section is n times the processing time, the error rate is 1 / 2n.

When the allowable error rate ρ is given, the relationship between the total time Tb _i for satisfying the allowable error rate and the processing time R is expressed by the following equations (12) and (13).
ρ ≧ 1 / 2n = R / 2Tb _i (12)
Tb _i ≧ R / 2ρ Equation (13)
That is, it can be seen that the time Tb _{i of the} totaling section must be at least 1 / 2ρ times the processing time R. In this description, the processing time R of each process is set to a constant value for simplicity of explanation, but in this embodiment, the totaling interval calculation unit 5 performs processing time of each process as shown in Expression (7). Are different, the total section is calculated using the maximum processing time among these processing times.

Next, the relationship between the resource usage rate and the time of the total section will be described.
FIG. 10 is a diagram illustrating a first example of the relationship between the processing time and the total section error rate.
The example shown in FIG. 10 is an example in which the time of the totaling section is set to three times the processing time, similarly to the example shown in FIG.

  In this example, the arrows drawn in the sub-section C represent the average start time and average end time of the processing that occurs in the sub-section C. In addition, half of the processing time of the processing that occurred in the sub-section C distinguished by diagonal lines in FIG. Therefore, the error rate of the resource usage measured in the total section i is 1/3 × 1/2 = 1/6.

Next, consider the case where a lot of processing occurs and the resource utilization rate becomes high.
FIG. 11 is a diagram illustrating a second example of the relationship between the processing time and the total section error rate.
In this example, since the resource utilization rate is high, the processing time of each processing is twice the processing time of a. In this case, the processing time of the processing occurring in the sub-intervals B and C distinguished by diagonal lines in FIG. The arrows represent the average start time and average end time of processing that occurs in the sub-sections B and C.

  One half of the processing time of the processing occurring in the sub-intervals B and C protrudes from the totaling interval i. Therefore, the error rate of the resource usage measured in the total section i is 2/3 × 1/2 = 1/3. Thus, it can be seen that as the resource utilization rate increases, the processing time of each process increases, and for this reason, the error rate of the aggregated section data increases.

FIG. 12 is a diagram illustrating a third example of the relationship between the processing time and the total section error rate.
In this example, when the processing time of each process is double the processing time of a due to a high resource utilization rate, the time of the totaling section is also doubled according to the processing time. In this case, the processing time of the processing that occurred in the sub-section C distinguished by diagonal lines in FIG. The arrows represent the average start time and average end time of processing that occurs in sub-section C.

  1/2 of the processing time of the processing that occurred in the sub-interval c protrudes from the aggregation section i. Therefore, the error rate of the resource usage amount measured in the total interval i is 1/3 × 1/2 = 1/6, which is equal to the error rate obtained in the example shown in FIG. Thus, when the processing time of each process fluctuates due to a change in the resource usage rate, it can be understood that the error rate of the total section data can be kept constant by adjusting the time of the total section in accordance with the processing time. .

  In general, it is difficult to directly measure the processing time of each process, so in this example, the processing time during operation is estimated from the resource usage rate and the processing time of each process at no load, and the time of the aggregation section is calculated. Has been decided.

In the present embodiment, the processing time calculation unit 4 receives the resource utilization rate information U (Ts _i ) at the start time of the counting section i and the no-load processing time information R ₀ received from the no-load processing time input unit 2. The processing time of each process is estimated, but if the change in resource utilization rate is large, not only the start time but also the average value of resource utilization rates at multiple time points from the start time, The processing time of each process can also be estimated.

  Further, in this embodiment, the processing time estimation method for each process is described on the assumption that the usage time for each resource of each process can be measured as the no-load processing time information. It is also conceivable that the processing time is measured. In that case, the processing time of each process can be estimated using the processing time of the entire resource of each process and the highest resource utilization rate among the resources. Alternatively, when there is a resource that occupies most of the processing time, the processing time of each process can be estimated using the utilization rate of the resource.

  In addition, the data generation unit 6 compares the processing amount as a result of totaling the resource usage time and processing amount of the totaling section with a preset value, and when the processing amount exceeds the set value, The total section data may be discarded and not transmitted to the total section data output unit 7. By performing this processing, it is possible to eliminate the total section data that causes a decrease in the parameter identification accuracy and generate the total section data.

  Next, Example 2 will be described. Example 2 is an example of the above-described second embodiment, and describes an aggregated section data generation method for performance model parameter identification. Further, detailed description of the same processing as in the first embodiment is omitted. Since the target system described in the present embodiment is the same as that in the first embodiment, description thereof is omitted.

Hereinafter, as in the first embodiment, it is assumed that the resource utilization rate of the server computer and the amount of processing of each process are obtained as measurement data at a constant time ΔT interval from the initial time T ₀ to the final time T _f .

  The measurement data input to the measurement data input unit 1 is transmitted to the total section calculation unit 5 and held in the measurement data storage unit 21 of the storage device 8. Further, the processing time data input to the processing time input unit 31 is transmitted to the total section calculation unit 5 and held in the processing time storage unit 22 of the storage device 8.

FIG. 13 is a diagram illustrating an example of processing time data according to the second embodiment.
In the example shown in FIG. 13, the start time and end time of the measurement section and the processing time average value of the processing that occurred between them are shown for each type of processing.

As described in the above-described embodiment, the totaling interval calculation unit 5 sets an initial value “1” to the variable i representing the number of the totaling interval.
The counting section calculation unit 5 sets the initial time T ₀ of the acquired measurement data as the start time Ts _i of the counting section i, and processing time information R (Ts _i ) = {R ¹ (Ts _i ), R at that time. ² (Ts _i ),..., R ^J (Ts _i )}, and by using the searched processing time and the allowable error rate ρ, the time Tb _i of the totaling interval _i is calculated and totaled. The end time Te _i = Ts _i + Tb _i of the section i is calculated. Since this calculation method is the same as that of the first embodiment, detailed description thereof is omitted. The subsequent processing is the same as in the first embodiment.

  The present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. Various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be omitted from all the components shown in the embodiment. Furthermore, you may combine suitably the component covering different embodiment.

The block diagram which shows an example of a structure of the performance calculating apparatus according to the 1st Embodiment of this invention. The figure which shows an example of the input / output relationship of the various data which the performance calculating apparatus according to the 1st Embodiment of this invention handles. The flowchart which shows an example of the processing operation of the performance calculating apparatus according to the first embodiment of the present invention. The block diagram which shows an example of a structure of the performance calculating apparatus according to the 2nd Embodiment of this invention. The figure which shows an example of the input / output relationship of the various data which the performance calculating apparatus according to the 2nd Embodiment of this invention handles. The flowchart which shows an example of the processing operation of the performance calculating apparatus according to the 2nd Embodiment of this invention. The figure which shows an example of the measurement data in Example 1 in a table format. The figure which shows an example of the total area data in Example 1 in a table format. The figure which shows an example of the relationship between a total area and processing time. The figure which shows the 1st example of the relationship between processing time and a total area error rate. The figure which shows the 2nd example of the relationship between processing time and a total area error rate. The figure which shows the 3rd example of the relationship between processing time and a total area error rate. FIG. 10 is a diagram illustrating an example of processing time data according to the second embodiment.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Measurement data input part, 2 ... Unload processing time input part, 3 ... Allowable error rate input part, 4 ... Processing time calculation part, 5 ... Total section calculation part, 6 ... Data generation part, 7 ... Total section data Generation unit, 8 ... storage device, 9 ... input device, 10 ... display device, 11 ... bus, 21 ... measurement data storage unit, 22 ... processing time storage unit, 23 ... error rate storage unit, 24 ... aggregate interval data storage unit 31 ... Processing time input unit.

Claims (5)

Based on the measured value of the utilization rate of each resource by each process executed by the target device and the processing time information of each process executed by the target device in a state where the resource does not compete with other resources using the specific resource , Processing time estimation means for estimating the processing time of each process executed by the target device;
Based on the processing time estimated by the processing time estimation means, the time of the aggregation section for parameter identification of the performance model of the target device is out of the section of the total processing time of each process generated in the section. An aggregate interval calculation means for calculating so that the time to satisfy the allowable condition,
A performance calculation apparatus comprising: a totaling unit that totals the usage rates of each resource within the time calculated by the totaling interval calculation unit based on the measured value of the usage rate of each resource. An acquisition means for acquiring a measurement value of a processing time of each process executed by the target device;
Based on the processing time acquired by the acquisition means, the time of the aggregation section for parameter identification of the performance model of the target device is the time outside the section of the total processing time of each process generated in the section An aggregate interval calculation means for calculating so as to satisfy the allowable condition;
And a totaling unit that totalizes the utilization rate of each resource within the time calculated by the totaling interval calculation unit based on a measured value of the utilization rate of each resource by each process executed by the target device. Performance calculator. The counting means includes
The generation amount of each process within the time of the total section calculated by the total section calculation unit is further totaled based on the measured value of the processing amount of each process executed by the target device. 2. The performance calculation device according to 2. The counting means includes
The utilization rate of each resource in the said time is totaled when the generation amount of each process in the total period calculated by the said total section calculation means does not exceed a reference value. Performance calculator. Computer
Based on the measured value of the utilization rate of each resource by each process executed by the target device and the processing time information of each process executed by the target device in a state where the resource does not compete with other resources using the specific resource , Processing time estimation means for estimating the processing time of each process executed by the target device,
Based on the processing time estimated by the processing time estimation means, the time of the aggregation section for parameter identification of the performance model of the target device is out of the section of the total processing time of each process generated in the section. Aggregation interval calculation means for calculating a certain period of time so as to satisfy an allowable condition, and aggregation for calculating the utilization rate of each resource within the time calculated by the aggregation interval calculation means based on the measured value of the utilization rate of each resource A computer-readable performance calculation processing program which is made to function as a means.

Images