JP2002007232A - Performance testing method and server testing device for www server - Google Patents

Abstract

PROBLEM TO BE SOLVED: To conduct a load test of a WWW server by reproducing a more realistic access state. SOLUTION: The WWW server, which is so programmed as to generate and send out objects responding to HTTP requests from many user agents connected via a public data communication network to the user agents, is an object server to be tested, which is connected to a server testing device via a leased data communication line replacing the public data communication network; and this server testing device sends a large number of HTTP requests, simulating many user agents to the object server to be tested simultaneously in parallel, decides whether each response includes an object specified by each HTTP request without errors, and outputs the decision result in a prescribed report format.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a performance test method for a computer system functioning as a WWW server, particularly when a large number of HTTP requests by a large number of user agents are transmitted to a WWW server concurrently. It relates to a performance test method assuming the following. It also relates to the test equipment.

[0002]

2. Description of the Related Art WWW (World Wide Web) is a server-client type information providing service. The most common use of WWW is HTML documents (Web pages)
Browsing. W that provides Web pages to clients
The WW server includes a database storing HTML document data, data such as images and sounds inserted by hyperlinks in the document, and programs such as CGI, and returns a Web page specified by the client. Also, when the client specifies and activates a program such as CGI and accesses it, the processing result is returned as HTML format data.

[0003] On the other hand, a Web server mounted on the client side
The software for viewing the page is a browser. The browser requests transmission of predetermined data or activation of a predetermined program to a predetermined WWW server.
It interprets the HTML data returned from the server and displays the result of the interpretation as a Web page.

As is well known, HTML requests and returns are HT
This is performed based on TP. HTTP is a protocol in which a session is completed by a set of requests / responses. Here, browsers related to individual sessions are referred to as “user agents”. That is, if a plurality of browsers are simultaneously started on a certain client, a plurality of HTTP requests by a plurality of user agents are transmitted from one client.

[0005] The HTTP request is an object identification (URI: U) for specifying an object to be requested, such as data to be obtained or a program to be started.
A request-line that describes a protocol (versus Resource Resource Identifier), the method of processing the object (Method), and the protocol version. The transmission date and time (Date) and the name of the software that sent the HTTP request (User- Agent) is attached as header information. When sending data to a WWW server, the data itself (Entity-Body) is also attached.

[0006] The object identification usually follows the URL description method. As methods, get objects (GET) and send data to objects (POST)
and so on.

[0007] The HTTP response returned by the WWW server in response to the HTTP request includes information (Status-Line) indicating whether the request was successful or unsuccessful, and a data body corresponding to the object specified by the object identification (Status-Line). Entity-Body), and various types of header information are attached to the subject information.

[0008] The user agent shapes the URL into a fixed HTTP request, and sends the HTTP request to an appropriate network layer. In the case of the Internet, it is transmitted by TCP / IP. Then, it obtains the HTTP response addressed to itself, appropriately processes and outputs the data itself included in the response.

[0009]

Since the commercial use of the Internet has started, the number of Internet users has been increasing explosively, and it is said that there are currently about 20 million Internet users in Japan alone. . And the number of users is expected to continue to increase in the future. For this reason, it is not uncommon recently that HTTP requests exceeding the processing capacity are concentrated on one Web site and the WWW server on which the Web site is located stops functioning.

Of course, the Internet connection service provider (provider) who manages and operates the WWW server also increases the number of operating WWW servers, or improves the processing capability of the WWW server by parallel processing by a plurality of computers. , Trying to accommodate the growing number of clients. However, there is a problem that cannot be solved fundamentally by simply increasing the hardware resources (the number of operations and the processing capacity) of the WWW server. That is, the WWW server stores the URL specified by the client.
, Not only the Web page is sent, but also an internal process such as a CGI program is executed in parallel with the sending process. Therefore, the performance of the WWW server cannot be simply determined only by the number of requests that can be accepted, but is also determined in consideration of the execution time of the program, the frequency of starting the program, and the like. It is necessary to achieve an effective performance improvement at a low cost by implementing the enhancement of the hardware resources and the improvement of the software in a well-balanced manner.

[0011] In addition to the hardware and software of the WWW server, it is necessary to consider the contents of the HTTP request. As the contents of the request, for example, requests are concentrated only on a specific Web page, or CGI
The method and form of the request corresponds to, for example, a large number of requests to a Web site having a large number of activations. This means that the performance of each WWW server differs depending on the contents of the information to be provided even if the hardware and software are the same.

[0012] Further, the client requests W once.
The operation characteristics of the WWW server also change depending on the request form reflecting the operation state of the browser on the client side, such as performing an operation such as aborting the request or retransmitting the request without waiting for a response to the transmission of the eb page. In such a request form, the WWW server generates a load due to a process of interrupting the web page sending process that has been progressing halfway or starting the sending process again. In particular, such a load frequently occurs during a time when access is concentrated. In a WWW server that provides information for mobile phones, the tendency is remarkable because access is made using a low line speed of a mobile data communication network.

As described above, the request form of the WWW server varies in various ways depending on the contents provided by itself. Therefore, it is necessary to design the hardware and software of the WWW server in accordance with the expected request form. Therefore, the present invention provides a test method for performing a load test of a WWW server by sending a large number of HTTP requests disguised as a large number of user agents to a test target WWW server which is actually operable. It is intended to provide the test apparatus.

[0014]

Means for Solving the Problems === WWW Server Performance Test Method === In order to achieve the above object, a first invention is a WWW server performance test method, which is performed via a public data communication network. HTTP from many connected user agents
A WWW server programmed to receive requests, generate an object corresponding to each request, and send the generated object to a corresponding user agent is a server to be tested. Interconnected by a dedicated data communication path instead of a data communication network, the server test equipment transmits a large number of HTTP requests imitating a large number of user agents to the server under test simultaneously and in parallel, The HTTP response from the HTTP request is individually recognized, and it is determined whether or not each response includes the object specified by the object identification included in each HTTP request without error, and the determination result is in a predetermined report format. Output.

=== Server Test Apparatus === The second to twenty-first inventions relate to the server test apparatus for performing the WWW server performance test method according to the first invention. The three invention groups can be classified.

=== First Invention Group: Second to Twelfth Inventions === The server test apparatus belonging to this invention group is mainly characterized by a hardware configuration.

<Basic Hardware Configuration> The second and third inventions relate to a basic hardware configuration of a server test apparatus, and the large amount of HTTP is stored in a server to be tested.
Request transmitting means for simultaneously transmitting a request by impersonating a large number of user agents, a response receiving means for individually recognizing the HTTP response from the server under test, and The system includes a response determination unit that determines whether the object specified by the HTTP request is included without error, and a test result output unit that outputs a determination result by the determination unit in a predetermined report format. Further, in the server test apparatus according to the second invention,
A test condition setting for arbitrarily variably setting a test condition including information defining the content of the object identification generated by the request transmitting unit and information defining the frequency of the HTTP request output by the request transmitting unit. A third aspect of the present invention is a server test apparatus having the means.

<WWW Server Performance Evaluation Parameter> The fourth to sixth inventions are server test apparatuses provided with means for measuring the time required for HTTP request / response as an index for evaluating the performance of a WWW server. Specifically, in the server test apparatus according to the second or third aspect, the request transmitting unit outputs the HTTP request until the response receiving unit receives the HTTP response to the request. A fourth aspect of the present invention is a server test apparatus comprising a response time measuring means for measuring the time of the test, and reflecting the measurement result of the measuring means on the output content of the test result output means.

A fifth invention is the server test apparatus according to the fourth invention, wherein a reference response time is set for each object identification included in the HTTP request output by the request transmitting means. The server test apparatus according to the fifth aspect of the present invention, wherein a comparison result between the response time measured by the response time measurement means and the reference response time is reflected on the output content of the test result output means. A sixth aspect of the present invention is a server test apparatus including test condition setting means for arbitrarily variably setting test conditions including the reference response time for each object identification.

<Evaluation for Variable Processing> Second or third
The server test apparatus according to the invention, wherein the request transmitting unit generates a large amount of the HTTP requests while sequentially changing specific parameters included in the HTTP requests in accordance with a predetermined rule, and sequentially outputs the generated HTTP requests. In a seventh aspect of the present invention, the server testing apparatus according to the seventh aspect, further comprising a test condition setting means for arbitrarily variably setting test conditions including data on rules for changing the parameters. And

A ninth aspect of the present invention is the server test apparatus according to the second or third aspect, wherein the test is performed while changing the output frequency of the HTTP request temporally. A request transmission schedule that defines whether to perform the request is set, and the server test apparatus operates the request transmission unit according to the schedule. A tenth invention is the server test apparatus according to the ninth invention, further comprising a test condition setting means for arbitrarily variably setting the contents of the request transmission schedule.

=== Second Invention Group: Eleventh to Seventeenth Inventions === The server test apparatus belonging to the second invention group is characterized by a test content and a report format of test results. . An eleventh invention is the server test device according to the second or third invention, wherein the determination result output means includes: a determination result by the response determination means; A server test device that generates a report output in association with the output frequency.

A twelfth aspect of the present invention is the server test apparatus according to any one of claims 4 to 6, wherein the determination result output means includes: a measurement result obtained by the response time measurement means;
The server test device generates a report output in which the measurement result is associated with the request output frequency at the time of performing the test.

Further, a thirteenth invention is directed to a ninth or tenth invention.
5. The server test apparatus according to claim 1, wherein the determination result output unit generates a report output in which the determination result by the response determination unit is associated with the request transmission schedule that has generated the determination result. And

A fourteenth invention is a server test device that satisfies any of the fourth to sixth inventions and any one of the ninth and tenth inventions, wherein the discrimination result output means includes: The server test apparatus generates a report output in which the measurement result of the measurement means is associated with the request transmission schedule that generated the measurement result.

A server test apparatus according to a fifteenth aspect is the server test apparatus according to the ninth or tenth aspect, further comprising schedule report means for outputting the request transmission schedule in an appropriate report format.

=== Third Invention Group: Sixteenth to Twenty-First Inventions === This invention group includes inventions relating to data structures processed at the time of test execution. A server test apparatus according to a sixteenth aspect of the present invention is the server test apparatus according to the second or third aspect, wherein a large number of script files respectively defining the contents of the various types of HTTP requests having different object identifications are appropriately prepared. Many kinds of test environment files that define the transmission frequency of the HTTP request and the like are stored in appropriate storage resources, and a large number of script files and a large number of the test environment files are stored. The test condition means specifies whether to execute the test using the test condition.

[0028] In the server test apparatus according to the sixth aspect of the present invention, a large number of script files defining the contents of the various HTTP requests with different object identifications are stored in appropriate storage resources. In addition, various types of test environment files that define the reference response time and the like for each object identification are stored in an appropriate storage resource, and a test can be performed using any of a number of script files and a number of the test environment files. A seventeenth aspect of the present invention is a server test apparatus for specifying whether to perform the test by the test condition means.

An eighteenth invention is the server test apparatus according to the eighth invention, wherein a large number of script files, each of which defines the contents of the various HTTP requests with different object identifications, are stored in appropriate storage resources. Numerous types of test environment files that are stored and that define rules for changing the parameters are stored in an appropriate storage resource, and a test can be performed using any of a number of script files and a number of the test environment files. The server test apparatus specifies whether the test is performed by the test condition means. The server test apparatus according to the tenth aspect, wherein a large number of script files respectively defining the contents of the various types of HTTP requests having different object identifications are stored in appropriate storage resources. Numerous types of test environment files that define a request transmission schedule and the like are stored in an appropriate storage resource, and the test condition means determines which one of a large number of script files and a large number of the test environment files is used to execute a test. The server test device specified by the above is the nineteenth invention.

A twentieth aspect of the present invention is the server testing apparatus according to any one of the sixteenth to nineteenth aspects, wherein the server testing apparatus intervenes in a data communication path between a browser executed by an appropriate computer and the server under test. To obtain an HTTP request and an HTTP response exchanged between the browser and the server under test,
Script file generating means for generating the script file based on the request / response is attached.

According to a twenty-first aspect, in the server test apparatus according to any one of the sixteenth to twentieth aspects, information for designating which script file is used for each of the test environment files is stored. The test condition setting means receives a user input for selecting an arbitrary test environment file, and determines which of a large number of script files to use for the test based on the selected test environment file. I decided to identify it.

[0032]

DESCRIPTION OF THE PREFERRED EMBODIMENTS === Test Environment === According to a WWW server performance test method according to the present invention, a WWW server (test target server) to be tested and a server test apparatus are connected to public data communication such as the Internet. The test is to be performed in an environment interconnected by a dedicated data communication path instead of a network. As a dedicated data communication path, one server test device and one server to be tested are connected.
Although a form of one-to-one connection is also conceivable, an embodiment of the method of the present invention will be described using a LAN as an example. FIG. 1 is a schematic configuration diagram of a LAN including a server test apparatus (tester) and a server to be tested.

In this embodiment, the server 10b to be tested and the tester 10a, or other computers 10c and peripheral devices (10d and 10e) are Ethernet (registered trademark).
(Ethernet (registered trademark)). The protocol used for this LAN1 is TCP /
It is an IP and has built a so-called "intranet". A general personal computer can be used for the test target server 10b and the tester 10a, and WWW server software and server test software are installed in these computers. Then, each computer executes the software to execute the WW.
Functions as a W server and tester.

An IP address is assigned to each of the computers (10a to 10c) and the peripheral devices (10d, 10e), and a specific computer or peripheral device is identified by the IP address. In this example, a private IP address is assigned to the test target server 10b. Thus, even if the LAN 1 is connected to the Internet via the router 10e, a computer outside the LAN 1 cannot directly access the test target server 10b. Of course, even if a global IP address is assigned to the test target server 10b,
If the domain name of the computer 10b is not registered with a predetermined application organization, the computer 10b can be brought into a state in which it is not substantially published on the Internet (a state in which an unspecified user is not aware of the location).

=== Outline of Test Method === The test target server 10b has an H describing a Web page.
The TML document data, image data to be inserted into the document by a link, or a database accompanying a CGI program or the like is stored. And a certain UR
When an HTTP request including L is sent from the user agent, the HTTP request including the object requested by the URL is returned.

The tester 10a is a server 10 to be tested.
Impersonate a user agent sending an HTTP request to b. Moreover, a large number of HTTP requests disguised as a large number of user agents are transmitted, and an HTTP response to each request is individually received. Then, HTTP is included in the received HTTP response.
A test is performed by determining whether the object specified by the request is included without error, measuring the response time required for the request / response, and outputting the test status in an appropriate format. Report.

The tester 10a provides H, such as the amount of traffic (the number of simultaneous accesses), the number of repetitions, and the transmission interval.
The transmission frequency of the TTP request and the test target server 10
By appropriately setting test conditions such as parameter values given to the CGI of b, a large number of various HTTP requests are generated. Then, a response time required for each HTTP request / response, presence / absence of an error for the HTTP request, and the like are checked. The tester 10a thus tests the performance of the test target server 10b from various viewpoints.

=== Disguise of User Agent === The server test software executed by the tester 10a individually manages a session based on a set of HTTP requests / responses as a session with one user agent. Therefore, if a large number of sessions occur, HTTP from a large number of user agents
The request will be impersonated.

In the present embodiment, a parent process for managing all sessions and a child process for individually managing one session are executed. The child process executes a process of transmitting an HTTP request in one session, a process of receiving an HTTP response to the process, and a process of recording a monitoring status (log) of the transmission / reception state. Then, a number of child processes having the same content are executed in parallel at the same time to set the traffic amount, or the transmission interval of the HTTP request by each child process is set to an appropriate time to appropriately change the request transmission frequency. Set. Thereby, a large number of HTTP requests are transmitted by impersonating a large number of user agents. FIG. 2 shows that many users
The impersonation state of the agent is shown as a schematic diagram. Multiple child processes are started in parallel. Each child process transmits an HTTP request including a port number (P1, P2,..., Pn) and a predetermined URL (a, b,..., X). Of course, the same URL may be specified in different child processes.

As described above, each session is executed by an individual child process. In this embodiment, the OS
(Operatiing Sysytem) assigns a port number to each child process to identify the application. Thereby, each port is perceived by the server under test as equivalent to access from a different user agent. The WWW server actually connected to the Internet also acquires the IP address and the port number as the sender of the HTTP request. Also, when the user agent accesses via a proxy server, the WWW server recognizes that the proxy server is the source of the HTTP request even if the access is from many clients (IP addresses). . Therefore, in this embodiment, the tester does not perform impersonation such that the IP address is variably set for each user agent (child process). In a LAN, the IP address of the computer used as a tester is sufficient.

Furthermore, if the server to be tested provides contents for mobile phones, a large number of user agents can be impersonated in other ways. That is, the HTT that the mobile phone sends to the WWW server
A terminal ID for individually identifying a telephone is attached to the header of the P request as information for basic authentication. Using this, the tester impersonates a large number of user agents by variably setting the terminal ID.

=== Script File === In this embodiment, for each URL received by the server to be tested, a script file describing the contents of an HTTP request including the URL and the contents of an HTTP response to the request is stored. I will prepare it. The tester runs each child process based on this script file.

FIG. 3 schematically shows the contents of an example of the script file. The script file first describes the name of this script file, the name of the server to be tested and the designation of the port number of the WWW server software, and then the syntax of the HTTP request and the test target for this request. It describes the syntax of the HTTP response to be sent by the server. In the description portion indicating this response, information that cannot be uniquely specified, such as date and time, is described in a wild card (＄ mark). Thus, the description portion of the wildcard indicates that "anything may be described".

=== Test Environment File === As described above, the script file is passed to each child process. On the other hand, information for specifying which script file the parent process uses to execute the child process is provided as a test environment file. FIG. 4 shows an outline of the data structure of the test environment file. Information for setting a time limit for the response time (reference response time), information for specifying one or more script files, and a traffic amount of a child process to be executed based on each specified script file (process loop). Number). The information described in the test environment file is not limited to these, and appropriate test conditions can be added.

In this embodiment, a large number of test conditions and U
It is assumed that a large number of test environment files and script files are stored in the database corresponding to the RL.
Of course, a test environment file or a script file may be created for each test opportunity according to a user input.

=== Process Flow === Next, the flow of a parent process and a child process in this embodiment will be described. FIG. 5 schematically shows the flow when a parent process executes a child process. Upon receiving the user's designated input of the test environment file, the tester stores the designated test environment file in a predetermined RA.
Expand to M area (work area) and execute parent process. The parent process reads the script file based on the test environment file and stores it in the work area. Then, each child process based on each script file is executed based on test conditions described in the test environment file.

The parent process monitors the number of execution loops of each child process, and executes each child process by the number of loops described in the test environment file. This number of loops corresponds to the traffic volume, and the faster the processing speed of the CPU in the tester, the more “large number of concurrent HTTP requests”
Can be disguised. Actually, the line speed in data communication is extremely slower than the processing speed of the CPU, and a sufficiently simultaneous communication environment is realized.

As for the activation sequence of each child process, for example, the next child process may be executed after the loop processing in one child process is completed, or each child process may be executed one by one in order. After that, a sequence may be adopted in which each process is sequentially executed again, and child processes executed by the number of loops are sequentially removed from the processing target. Conditions for these sequences could also be set by the test environment file.

FIG. 6 shows the flow of each process executed by a certain child process. The child process determines whether an object specified by the HTTP request is included in the HTTP response without error through transmission and reception of the HTTP request / response, and measures a response time required for the request / response. Is performed, and the determination result and the measurement result are recorded as a log.
Also, when an error occurs in various processes included in the child process, the fact is recorded.

Specifically, when the child process is started,
The transmission timeout period and the reception timeout period described in the test environment file stored in the work area are read (s11). In this example, the transmission timeout period is a time limit for the time from when a connection with the server to be tested is established by TCP / IP to when the transmission of the HTTP request is completed, and the reception timeout period is the time when the transmission of the HTTP request is completed. This is the time limit for the time from receiving the HTTP response until receiving the HTTP response.

When the time limit is set, a loop process for the number of times corresponding to the traffic amount is started. In this loop, first, the name and port number of the test target server described in the script file are read, and a connection is established with the server (s12, s1).
3). Further, the description part of the HTTP request in the script file is read and the HTTP request is transmitted to the server (s15, s16). Also, the time required for the transmission processing is measured, and when the transmission timeout time has elapsed, it is determined that an error has occurred, and that fact is recorded in a log (s17 → s29).

When an HTTP request is transmitted and an HTTP response is returned from the server to be tested, reception processing is performed. If the HTTP response is not received before the reception timeout period elapses, the fact is recorded in a log (s19 to s21 → s29). HTTP
When the response is received within the time limit, the description part of the HTTP response in the script file is read from the work area, this is compared with the actually received HTTP response, and the session with the server is terminated (s22 to s24). If the predetermined number of loops has been reached, the response time and the comparison result are recorded in a log, and the child process ends (s27 → s28). If the number of loops has not been reached, the response time and the comparison result obtained in the loop are recorded in a log, and the next loop processing is executed (S
27 → s29).

When transmitting an HTTP request including a parameter, the parameter can be dynamically changed. The parameter is an argument to be given to the CGI, or a user ID described in a header portion or a URL portion of the HTTP request.

FIGS. 7A to 7C are schematic diagrams showing an example of processing for variably setting parameters. In this example, parameters are given to a CGI program that performs an authentication procedure. The parameter is a terminal ID of the mobile phone. (A) schematically shows a script file used when the terminal ID is variably set. An HTTP request including a variable portion and an HTTP response corresponding to the fact that authentication has been obtained by the variable are described. (B) is a variable (terminal I
It shows a part of the flow of the child process including the process of variably setting D). The variable number setting process is inserted between the processes s15 and s16 in the flowchart of the child process shown in FIG. In this example, the script file referenced by the child process contains the string "parameter"
Is included, an HTTP request in which the character string is replaced with a character string “123456789012” generated by a random number is generated (s40, s41 → s42). Of course, the rule for setting the parameter value is not limited to a random number, and may be generated according to an appropriate rule such as a predetermined function. Then, an HTTP request including the set parameters is transmitted (s16). If the read script file does not include a predetermined character string, an HTTP request based on the script file is transmitted (s40, s41 → s16). (C) is s4 of (B)
2 shows the syntax of the HTTP request generated by the processing of FIG.

=== Test Contents === As described above, the server test software of the tester sets the test conditions based on the test environment file by the parent process, and stores the test conditions in the script file specified by the parent process. The child process individually executes an HTTP session with the test target server based on the child process.

Specific test contents include a load test in which an HTTP request is sent by setting a traffic amount (the number of simultaneous transmissions) for each URL, and an HTTP test for the same URL.
And an endurance test for repeatedly sending requests. In this embodiment, in order to test the server to be tested from various viewpoints, a method of changing the traffic volume of the HTTP request for each URL in a time-series manner is employed. That is, "at a certain time, an HTTP request including a certain URL is generated with a certain traffic amount."
The load test and the durability test were performed while reproducing a situation close to actual access.

=== Report output of test results === The tester compiles logs individually recorded by a large number of child processes executed through the test according to the above method.
The analysis is performed, and the analysis result is displayed or output in a predetermined report format. FIGS. 8A to 8C show examples of the report output. (A) shows the contents of the test, and the time-series change in the number of requests is displayed for each URL. In this diagram, the relationship between the time and the number of requests is shown by a bar graph for each of the three URLs (URLs: a to c). (B) shows the average value of the response time to the request for each URL as a time-series change. In this example, a continuous response time transition is represented by a line graph. (C) is a URL (U
RL: a) shows the frequency of errors for chronological order. In this example, the case where the HTTP response of the script file is the same as the actual response is “success”, and the case where they are different is “error”. The success / error in the number of requests is shown by a bar graph.

From the report output described above, for example, if the number of requests and the response time or the number of errors are substantially proportional to any URL, it can be determined that the performance of the hardware of the server to be tested should be improved. If the response time is long only for a request including a specific URL or the frequency of occurrence of an error is high, the URL
It can be seen that there is a problem in a program (CGI or the like) related to the sending processing of the object specified by L.
Further, if the response time and the frequency of occurrence of errors tend not to depend on the URL or the number of requests, it may be suspected that there is a fundamental defect in the software.

The report format (graph) shown in FIG. 8 is an example, and the graph format (bar graph, line graph, etc.) and the correspondence (vertical axis, horizontal axis) can be appropriately changed. Of course, the child process records in the log which process caused the error, so that the report content such as the frequency of the error in other processes can be arbitrarily set.

=== Other Applications / Modifications === In the above embodiment, the script file describes the contents of HTTP requests and responses by key input by the user. There is also a method of automatically generating a script file without generating it in this way. For example, a browser is started on a tester or another computer connected to the same LAN. The browser specifies the URL, transmits an HTTP request to the test target server, and receives a response to the request. The HTTP request / response transmitted / received by the browser is acquired, and the content is analyzed. And Reques
Extract predetermined items such as t-Line and Status-Line. Create a script file by applying the items to the fixed format. It is also possible to provide the tester with such an automatic script file generation function as an additional function.

The number of servers to be tested is not limited to one, and a plurality of servers to be tested may be on the same LAN. In this case, the tester may specify the server name (IP address) to specify the destination of the HTTP request. Further, there may be a form in which the WWW server software and the server test software are mounted on the same computer.

In the above embodiment, all test conditions are described in the test environment file. However, it is easy to provide the tester with a user interface for partially changing the description of the test environment file selected by the user. Alternatively, a rule (a function or the like) for changing the time limit or the parameter can be set by a user input. As a result, test conditions can be freely customized.

Also, contrary to the spirit of the present invention, there is a possibility that a user who attempts to access a WWW server on the Internet using this tester may appear. In order to cope with such a possibility, the number of routing hops may be fixedly set in the tester. As a result, the TCP / IP packet carrying the HTTP request cannot pass through more than the set number of routers and the target WWW
The server cannot be reached.

[0064]

According to the present invention, a large number of Hs impersonating a large number of user agents on a WWW server to be tested.
A load test can be performed by sending a TTP request. That is, the performance can be tested for each of the user agents accessing the test target server by identifying the specified object and accessing the server. Therefore, WWW
A faulty or faulty process can be identified from a number of internal processes executed on the server. Thereby,
Quickly respond to program changes and modifications.

Further, when making the server to be tested open to the public data communication network such as the Internet, it is possible to employ hardware of appropriate size and performance. Therefore, maximum performance can be exhibited while minimizing server installation costs.

In addition, by setting the response time required for the HTTP request / response, the accuracy of the HTTP request, and the like as test items, the performance of the WWW server can be more diversified.

By changing the object identification and the transmission frequency of the HTTP request in time series, it is possible to reproduce a situation close to an actual access. Therefore, the performance of the server to be tested can be grasped before disclosure.

The HTTP request to be sent to the server to be tested and the transmission conditions thereof may be set by user input, or a file (script file, test environment file) in which these are described in advance may be prepared and stored in those files. It may be set on the basis of. As a result, the test conditions can be finely customized, and a test can be performed in a variety of situations.

If the script file is created based on the HTTP request / response by the browser, it is not necessary to create the script file by key input or the like. Therefore, it is possible to reduce the labor of the user (the person who performs the test), shorten the period until the start of the test, and promptly disclose the WWW server.

[Brief description of the drawings]

FIG. 1 is a schematic diagram illustrating a data communication path between a server to be tested and a server test apparatus according to an embodiment of the present invention.

FIG. 2 is a schematic diagram of a child process executed by the server test apparatus of the embodiment.

FIG. 3 is a configuration diagram of a script file processed by the server test apparatus.

FIG. 4 is a schematic diagram of a test environment file processed by the server test apparatus.

FIG. 5 is a schematic diagram showing a relationship between a parent process and a child process executed by the server test apparatus and a flow of each process.

FIG. 6 shows a flowchart of the child process.

FIG. 7 is a schematic view of a parameter variable setting process executed by the server test apparatus. (A) shows an outline of a script file including parameters. (B) shows a part of the flow of the child process including the process of variably setting the parameters. (C) shows the syntax of the HTTP request including the set parameters.

FIG. 8 shows a graph displayed and output by the server test apparatus. (A) shows a report output diagram. (A)
Is a graph showing time-series changes in the number of accesses for each URL. (B) is a graph showing a response time to a request for each URL as a time-series change. (C) is a graph showing, in chronological order, the frequency of errors with respect to access to a certain URL.

[Explanation of symbols]

 1 LAN 10a server test equipment 10b server to be tested

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Satoshi Ohira 4-16-13 Nishiazabu, Minato-ku, Tokyo Within Saibird Co., Ltd. (72) Inventor Yasuo Hasegawa 222-2 Fujisawa, Fujisawa-shi, Kanagawa Pref. 72) Inventor, Kazunari Haneda 223-2 Fujisawa, Fujisawa-shi, Kanagawa Prefecture F-term within Wips Co., Ltd. (reference) BB02 DD03 EE01 EE25 GG02 GG05 HH07 JJ04 KK01 MM03

Claims (21)

[Claims] 1. A program for receiving HTTP requests from a number of user agents connected via a public data communication network, generating an object corresponding to each request, and transmitting the object to the user agent. The tested WWW server is a server to be tested, and the server to be tested and the server test apparatus are interconnected by a dedicated data communication path instead of the public data communication network, and the server test apparatus disguises a large number of user agents. A large amount of HTTP requests sent to the server under test are simultaneously transmitted to the server under test, and the HTTP responses from the server under test are individually recognized and specified by the object identification included in each HTTP request in each response. Object is incorrect To determine whether it is included
A method of testing the performance of a WWW server, wherein a result of the determination is output in a predetermined report format. 2. The server test apparatus for performing the WWW server performance test method according to claim 1, wherein the test target server impersonates a large number of HTTP requests with a large number of user agents. A request transmitting means for transmitting simultaneously and in parallel, a response receiving means for individually recognizing an HTTP response from the test target server, and each response received by this means includes an object designated by each HTTP request without error. A server test apparatus comprising: response determination means for determining whether or not a response has been made; and test result output means for outputting a determination result by the determination means in a predetermined report format. 3. The server test apparatus according to claim 2, wherein the HTTP output from the request transmitting unit is output.
A server comprising: test condition setting means for arbitrarily variably setting a test condition including information specifying the content of the object identification included in a request and information specifying a transmission frequency of an HTTP request. Testing equipment. 4. The server test apparatus according to claim 2, wherein said request transmitting means is said HTTP server.
Response time measuring means for measuring a time from when the request is output to when the response receiving means receives the HTTP response to the request, and a measurement result by the measuring means is reflected in an output content of the test result output means A server test apparatus characterized in that: 5. The server test apparatus according to claim 4, wherein said HTTP output from said request transmitting means is output.
A reference response time is set for each of the object identifications included in the request, and a comparison result between the response time measured by the response time measuring means and the reference response time is reflected in the output content of the test result output means. A server test device, characterized in that: 6. The server test apparatus according to claim 5, further comprising: test condition setting means for arbitrarily variably setting a test condition including the reference response time for each object identification. Server testing equipment. 7. The server test apparatus according to claim 2, wherein the request transmitting unit is configured to execute the HTT.
A server test apparatus, wherein a large number of HTTP requests are generated and sequentially output while changing a specific parameter included in a P request according to a predetermined rule. 8. The server test apparatus according to claim 7, further comprising: test condition setting means for arbitrarily variably setting a test condition including data on a rule for changing the parameter. Server testing equipment. 9. The server test apparatus according to claim 2, wherein a request transmission schedule is set which defines how to perform the test while temporally changing the output frequency of the HTTP request. A server test apparatus wherein the request transmitting means operates according to the schedule. 10. The server test apparatus according to claim 9, further comprising test condition setting means for arbitrarily variably setting the content of the request transmission schedule. 11. The server test apparatus according to claim 2, wherein the determination result output unit determines a determination result by the response determination unit and a request output frequency at the time of performing a test that generates the determination result. A server test apparatus for generating a report output in which a report output is associated. 12. The server test apparatus according to claim 4, wherein said discrimination result output means includes a measurement result obtained by said response time measurement means, and a test result when said test result is generated. A server test apparatus for generating a report output in which a request output frequency is associated with a request output frequency. 13. The server test apparatus according to claim 9, wherein the determination result output unit associates the determination result by the response determination unit with the request transmission schedule that has generated the determination result. A server test device for generating a report output. 14. The method according to claim 4, wherein the method comprises:
Or a server test apparatus that satisfies any one of the above (10), wherein the determination result output means generates a report output in which the measurement result of the response measurement means is associated with the request transmission schedule that generated the measurement result. A server test apparatus characterized in that: 15. The server test apparatus according to claim 9, further comprising schedule report means for outputting the request transmission schedule in an appropriate report format. 16. The server test apparatus according to claim 3, wherein a plurality of script files respectively defining the contents of the plurality of HTTP requests having different object identifications are stored in appropriate storage resources. Many kinds of test environment files that define the transmission frequency of the HTTP request and the like are stored in an appropriate storage resource, and which of a large number of script files and a large number of the test environment files is used to execute a test. A server test apparatus characterized by specifying by the test condition means. 17. The server test apparatus according to claim 6, wherein a plurality of script files respectively defining the contents of the plurality of types of HTTP requests having different object identifications are stored in appropriate storage resources. Many types of test environment files that define the reference response time and the like for each object identification are stored in an appropriate storage resource, and a test is performed using any of a number of script files and a number of the test environment files. A server test apparatus for specifying whether to perform the test by the test condition means. 18. The server test apparatus according to claim 8, wherein a plurality of script files respectively defining the contents of the plurality of HTTP requests having different object identifications are stored in appropriate storage resources. Many types of test environment files that define rules for changing the parameters are stored in an appropriate storage resource, and it is determined which one of a number of script files and a number of the test environment files is used to execute a test. A server test apparatus characterized by specifying by the test condition means. 19. The server test apparatus according to claim 10, wherein a plurality of script files respectively defining the contents of the plurality of HTTP requests having different object identifications are stored in appropriate storage resources. Many types of test environment files that define the request transmission schedule and the like are stored in an appropriate storage resource, and the test is performed to determine which one of a number of script files and a number of the test environment files is used to perform a test. A server test device characterized by specifying by a condition means. 20. The server test apparatus according to claim 16, wherein said browser is connected to a browser running on an appropriate computer via a data communication path between said browser and said server to be tested. HTTP request and HT exchanged with the server under test
A server test apparatus, which obtains a TP response and attaches a script file generating means for generating the script file based on the request / response. 21. The server test apparatus according to claim 16, further comprising information for designating which script file is to be used for each of said test environment files, wherein said test condition setting means is provided. Receives user input for selecting an arbitrary test environment file, and specifies which of a number of script files to use for testing based on the selected test environment file. Server test equipment.