Apparatus and method for monitoring service quality, and service quality monitoring program

【0001】

[Technical field to which the invention belongs]

The present invention relates to a service quality monitoring device and a method thereof, and a service quality monitoring program, which monitor the service quality of an information communication terminal connected to a network by using a model theory.

【0002】

[Conventional technology]

The conventional service quality control technology only measures the evaluation index of the resource of the information communication terminal for information transmission and displays the current status. In addition, regarding forecasts, trends are analyzed by a statistical method based on resource information, and long-term forecasts (when a transition of a certain value is expressed in chronological order, the overall trend is increasing or decreasing. It was used to predict trends).

【0003】

On the other hand, in order to make a short-term service quality prediction, the information used in the prediction is the cause system of the service quality (an item closer to the cause of the change in the target value, and if the target is the response time, the response time. It is necessary to approach the CPU load, which is a factor of increase / decrease, or the number of accesses, which is an element item). Therefore, a "WWW site performance monitoring device" using the number of accesses using model theory has been proposed (see Patent Document 1).

【0004】

[Patent Document 1]

JP-A-2002-268922 [0005]

[Problems to be Solved by the Invention]

By the way, in the conventional system implemented by the above-mentioned model theory, it is impossible to obtain the initial state at the start of calculation (the state in the model system at the start of simulation) because there is no data before the start of calculation. Therefore, the calculation could not be performed unless the measurement was started from the state where nothing existed in the system. In addition, for the method of evaluating the accuracy of model implementation and the adjustment with the coefficient for the model element (created in comparison with the actual system configuration, performance, operating principle, etc.), the value representing the target service quality. I could only verify it. Therefore, it was difficult to start the calculation from the middle of the system operation and improve the accuracy of the model.

【0006】

The present invention has been made in view of such circumstances, and an object thereof is to obtain an initial state at the start of calculation for a system, and further, to verify the validity of a model and adjust a model system by using virtual resource log information. By doing so, it is an object of the present invention to provide a service quality monitoring device and a method thereof, and a service quality monitoring program, which can start calculation from the middle of system operation and improve prediction accuracy.

【0007】

[Means for solving problems]

In order to solve the above-mentioned problems, the present invention is a service quality monitoring device that monitors the service quality of an information and communication terminal connected to a network by using a model theory, and is a resource output from the information and communication terminal. Virtual resource log information for a model in the same format as log information is created based on the simulation of the model, and when creating the virtual resource log information, the variable group used in the program in the model corresponds to the virtual resource log information. Created as a snapshot to be attached, the validity of the model is evaluated by comparing the virtual resource log information and the corresponding snapshot with the resource log information output from the information communication terminal, and the parameters for the model are modified. It is characterized by having a service quality monitoring unit that modifies the model.

【0008】

Further, in the present invention, the service quality monitoring unit creates virtual resource log information for a model in the same format as the resource log information output from the information communication terminal based on the simulation of the model. A corresponding snapshot creation / saving means for creating and saving a group of variables used in the program in the model as a snapshot associated with the virtual resource log information, the virtual resource log information and the corresponding snapshot, and the information. It is characterized by including a model evaluation adjustment unit that evaluates the validity of the model by comparing it with the resource log information output from the communication terminal, modifies the parameters for the model, and modifies the model.

【0009】

Further, in the present invention, the model evaluation adjustment unit is characterized in that the initial information of the model is selected and used from the resource log information of the information communication terminal by using the stored information of the virtual resource log information and the corresponding snapshot. And.

【0010】

According to the above configuration, the service quality monitoring unit creates virtual resource log information for the model in the same format as the resource log information output from the information communication terminal based on the model simulation, and when creating the virtual resource log information. , Create a group of variables used in the program in the model as a snapshot that associates with the virtual resource log information, and compare the virtual resource log information and the corresponding snapshot with the resource log information output from the information communication terminal to model the model. It is possible to adjust the model more accurately by evaluating the validity of the model, modifying the parameters for the model, and modifying the model. Also, by setting the initial information of the model, the model theory It is possible to improve the accuracy of short-term prediction by the method using.

【0011】

Here, "model theory" is to create a model that imitates an actual system, input the same data as the actual system to this model system, and perform a simulation. For example, a website. It refers to the method of calculating the target value such as the response time of.

【0012】

In order to solve the above-mentioned problems, the present invention is a service quality monitoring method used in a service quality monitoring device that monitors the service quality of an information communication terminal connected to a network by using a model theory. The quality monitoring device creates virtual resource log information for the model in the same format as the resource log information output from the information communication terminal based on the simulation of the model, and at this time, a group of variables used in the program in the model. Is created as a snapshot associated with the virtual resource log information, and the validity of the model is evaluated by comparing the virtual resource log information and the corresponding snapshot with the resource log information output from the information communication terminal. It is characterized in that the parameter for the model is modified to modify the model.

【0013】

In order to solve the above-mentioned problems, the present invention is a service quality monitoring program used in a service quality monitoring device that monitors the service quality of an information communication terminal connected to a network by using a model theory, and is the information communication. A step of creating virtual resource log information for a model in the same format as the resource log information output from the terminal based on the simulation of the model, and variables used in the program in the model when creating the virtual resource log information. The validity of the model is evaluated by comparing the step of creating the group as a snapshot associated with the virtual resource log information, the virtual resource log information and the corresponding snapshot, and the resource log information output from the information communication terminal. It is characterized by having a computer execute a step of evaluating, modifying parameters for the model, and modifying the model.

【0014】

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, the service quality monitoring device according to the embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a schematic block diagram showing a configuration of a service quality monitoring system that uses model theory, which is an embodiment of the present invention, and measures response time as service quality.

【0015】

In this figure, the log acquisition agent 4 that sends the resource log information of the server is operating on the Web server 1 connected to the service access line such as the Internet, and similarly, the DB connected to the Web server 1 It is assumed that the log acquisition agent 4 also operates on the server 2. The service quality monitoring device 3 connected to the same LAN 6 (Local Area Network) as the Web server 1 and the DB server 2 receives the data from the log acquisition agent 4 and models the hardware resources of the Web server 1 and the DB server 2. It has a model manager 5 that runs the simulation as. According to the present invention, the service quality monitoring unit 50, which will be described later, is newly implemented in the model manager 5.

【0016】

2 and 3 are flowcharts showing the schematic operation of the model manager 5 shown in FIG. This flowchart corresponds to the processing procedure of the service quality monitoring program in the present invention. Hereinafter, the schematic operation of the model manager shown in FIG. 1 will be described with reference to the flowcharts shown in FIGS. 2 and 3.

【0017】

In the flowchart shown in FIG. 2, first, in response to the resource log information request from the model manager 5 (S21), the log acquisition agent 4 receives the access information from the client on the Internet and the operating status of the Web server 1 and the DB server 2. The resource log information representing is sent to Model Manager 5 (S22). The model manager 5 calculates the virtual resource log information by simulation using the hardware resource as a model based on the resource log information and access information received from the log acquisition agent 4 (S23), and calculates the variable value in the model at that time. Associate as a snapshot and additionally output to a file (S24).

【0018】

Here, in the past, the response times calculated by introducing the client's request into the model system were compared with each other to evaluate the validity of the model and adjust the model system. By test-execution of the load pattern and quantitatively comparing each element of the resource log information acquired from the log acquisition agent 4 that can be obtained at that time with the virtual resource log information calculated in the model and its coefficient (S25), It is possible to evaluate the validity of the model and adjust the performance of the model system (S26). That is, in S25, the values of the resource log information and virtual resource log information by each workload pattern test are evaluated (validity evaluation), and if it is NG, the elements and coefficients of the S26 model are reexamined and adjusted. repeat. If the result of the validity evaluation is OK, the virtual resource information and the corresponding snapshot at the time of executing each load pattern test are saved (S27: save work), and the process shown in Fig. 3 is executed.

【0019】

In the flowchart shown in FIG. 3, when the model manager 5 starts the prediction (S28), it acquires the latest resource log information from the log acquisition agent 4 and compares it with the virtual resource log information already saved (S29). Select the closest virtual resource log information by any matching process (S30, S31).

【0020】

Here, "S30: Matching process" includes (1) a method of finding the closest statistically, (2) a method of repeating until the same thing appears, and (3) statistically if there is no same thing. There is a way to find the closest one. The flowchart shown in FIG. 3 is realized when (3) is adopted as the matching processing method, and the flow differs depending on the method for realizing the matching processing described above. That is, when the method shown in (1) is adopted as the matching process, the difference between all the elements and a certain data group is obtained, and the one with the smallest total absolute value of the difference is the current value. As an example, a method of determining that the state is close can be considered. In this case, there is no NG path for S30. Further, when the method shown in (2) is adopted, the matching process (S30) is executed again by returning to the information collecting process shown in S29. If the method shown in (2) is practiced the specified number of times and the result is NG, the method shown in (1) is practiced. The method (3) is as shown in the flowchart. By using the snapshot associated with the virtual resource log information selected in this way as the initial value of the variable value in the model, it is possible to calculate the initial value at the start of the model system calculation.

【0021】

Hereinafter, an embodiment of the present invention will be described in detail with reference to FIGS. 4 and later. FIG. 4 is a block diagram showing the internal configuration of the service quality monitoring device shown in FIG. 1 with expanded functions. The service quality monitoring unit 50 shown in FIG. 4 is implemented in the model manager 5 of the service quality monitoring device shown in FIG. 1. In FIG. 4, only the configuration of the service quality monitoring unit 50 is extracted and shown. It is composed of a virtual resource log information creation unit 51, a virtual resource log information and corresponding snapshot 52 (corresponding snapshot creation / saving unit), and a model evaluation adjustment unit 53.

【0022】

The virtual resource log information creation unit 51 creates virtual resource log information for a model in the same format as the resource log information (55) output from information communication terminals such as Web server 1 and DB server 2 based on model simulation. Has a function. The virtual resource log information creation unit 51 also creates a group of variables used in the program in the model as a snapshot associated with the virtual resource log information, and saves the variables as the virtual resource log information and the corresponding snapshot 52. An example of the virtual resource log information and the data structure of the corresponding snapshot 52 is shown in Fig. 8 as <Table 3>.

【0023】

The model evaluation adjustment unit 53 compares the virtual resource log information and the corresponding snapshot 52 with the resource log information (55) output from the information communication terminal, evaluates the validity of the model, and corrects the parameters for the model. It has a function to modify the model. The model evaluation adjustment unit 53 also has a function of selecting and using the initial model information from the resource log information of the information communication terminal by using the stored information of the virtual resource log information and the corresponding snapshot.

【0024】

Figure 5 shows an example of implementing a resource model in the model manager. As one embodiment of the resource model, here, the CPUs of Web server 1 and DB server 2 are used as model elements, and the actual access status is simulated based on the access log information acquired by the log acquisition agent 4. , The response time in the model system can be calculated. At this time, the CPU model of Web server 1 has two CPUs with the same performance. Also, the processing time for processing one request from the client is 2 seconds. On the other hand, the CPU model of DB server 2 has four CPUs with the same performance, and the processing time for processing one request from the client is 4 seconds. Also, it is assumed that each CPU can process only one request at a time. The model simulation is performed by importing the access log information acquired by the log acquisition agent into Model Manager 5.

【0025】

When the access log information of Web server 1 is in the state of <Table 1> shown in Fig. 6, the requests that occurred at 0 seconds are designated as a, and thereafter, if the requests are numbered in the order of occurrence, the requests from a to d are used. Is occurring. The request is first processed by the CPU of Web server 1 of the model system. After that, it is processed by the CPU of DB server 2 of the model system, and the process ends. Both Web server 1 and DB server 2 collect resource log information every second from the start of processing to the end of processing. In the above model system, when the requests a to d are simulated as shown in <Table 1> shown in Fig. 6, the snapshot of the model when the time transition is made every second is shown in Fig. 7 as <Table 2>. Can be represented.

【0026】

In addition, the CPU usage rate of each server can be acquired from the resource log information collected by the Web server 1 and the DB server 2. This value is the percentage of the time that the CPU is running on each server within the unit time divided by the unit time. Therefore, the virtual CPU usage rate (value of the virtual resource log) of each of the Web server 1 and the DB server 2 calculated by the model system can be calculated by, for example, the following calculation formula (1).

【0027】

[Number 1]<img file="JP2004241800A_D0001.tif" /> 【0028】

Therefore, when the virtual resource log information is calculated from the time transition of the model shown in &lt;Table 2&gt; in Fig. 7 by the following calculation formula, it is as shown in the virtual resource log information shown in &lt;Table 3&gt; in Fig. 8. Also, take the corresponding snapshot as shown in &lt;Table 3&gt;.

【0029】

The model evaluation adjustment unit 53 uses the virtual resource log information obtained by the above calculation formula (1) and the log acquisition agent when a request is generated from the client to the actual system as shown in &lt;Table 1&gt; in Fig. 6. The resource model can be expressed more accurately by comparing it with the resource log information acquired by and adjusting the processing time per request of the CPU model of Web server 1 and DB server 2 and other program values. It becomes possible.

【0030】

As a result, if the resource log information acquired from the log acquisition agent 4 at the start of model operation is Web server CPU = 50% and DB server CPU = 75%, a snapshot (Web server CPU = 1) is taken from the virtual resource log information. , DB server CPU = 3) turns out to be good to use as the initial value of the program. By using the virtual resource log information in this way, it is possible to infer the resource model state at the start of the model.

【0031】

If the value of the resource log information acquired from the log acquisition agent 4 at the start of model operation is Web server CPU = 50% and DB server CPU = 20% and does not match the value of any virtual resource log information, After a certain period of time, the resource log information is acquired again from the log acquisition agent 4 and compared with the value of the virtual resource log information. This means that the virtual resource log information and the resource log information acquired from the log acquisition agent 4 are repeatedly repeated until they match, and the calculation based on the model theory is started using the snapshot when they match. It should be noted that matching is not essential, and it is possible to consider similar things as matching.

【0032】

As described above, the present invention obtains the initial state at the start of calculation for the system, and further performs calculation from the middle of system operation by verifying the validity of the model and adjusting the model system using virtual resource log information. To improve the prediction accuracy. A computer-readable record of the procedures executed by each of the virtual resource log information creation unit 51, the virtual resource log information and the corresponding snapshot 52 (corresponding snapshot creation and storage unit), and the model evaluation adjustment unit 53 shown in FIG. It is assumed that the service quality monitoring device of the present invention is realized by recording on a medium, loading the program recorded on the recording medium into a computer system, and executing the program. The computer system referred to here includes hardware such as an OS and peripheral devices.

【0033】

In addition, the "computer system" shall include the homepage providing environment (or display environment) if the WWW system is used. Further, the program may be transmitted from a computer system in which this program is stored in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the "transmission medium" for transmitting a program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. Further, the above program may be for realizing a part of the above-mentioned functions. Further, a so-called difference file (difference program) may be used, which can realize the above-mentioned functions in combination with a program already recorded in the computer system.

【0034】

Although the embodiments of the present invention have been described in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs and the like within a range that does not deviate from the gist of the present invention.

【0035】

[Effect of the invention]

As described above, according to the present invention, it is possible to infer the initial state (snapshot) at the start of calculation, which was not obtained in the conventional system implemented by the model theory, and to verify the validity of the model and to verify the validity of the model. By adjusting the model system, it is possible to provide a service quality monitoring device that can start the calculation from the middle of the system operation and improve the prediction accuracy.

[Simple explanation of drawings]

FIG. 1 is a schematic block diagram showing a configuration example of a service quality monitoring system using model theory.

FIG. 2 is a flowchart cited for explaining the schematic operation of the service quality monitoring device shown in FIG.

FIG. 3 is a flowchart cited for explaining the schematic operation of the service quality monitoring device shown in FIG.

FIG. 4 is a block diagram showing the internal configuration of the service quality monitoring device shown in FIG. 1 with expanded functions.

[Fig. 5] Fig. 5 is a diagram cited to explain an example of realizing a resource model in the model manager shown in Fig. 1.

FIG. 6 is a diagram cited for explaining the operation of the embodiment of the present invention, and is a diagram showing the time transition of the access status in a table format.

FIG. 7 is a diagram cited for explaining the operation of the embodiment of the present invention, and is a diagram showing the time transition of the model in a table format.

FIG. 8 is a diagram cited for explaining the operation of the embodiment of the present invention, and is a diagram showing an example of virtual resource log information and corresponding snapshots in a table format.

[Explanation of symbols]

1 ... Web server, 2 ... DB server, 3 ... Service quality monitoring device, 4 ... Log acquisition agent, 5 ... Model manager, 6 ... LAN, 50 ... Service quality Monitoring department, 51 ... Virtual resource log information creation department, 52 ... Virtual resource log information and corresponding snapshots, 53 ... Model evaluation adjustment department, 54 ... Access log, 55 ... Resource log, 56 ... workload pattern