JP2002082926A - Distributed application test and operation management system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make performable test and operation management by distributed application units by specifying reliability, performance evaluation, and a bottle neck by server process or component units being the constituting elements of the distributed AP. SOLUTION: Quality measuring parts (313, 316, and 319) for measuring the performance data of components are embedded in the source code file group of a distributed application, and those source code files are transmitted through a complier, and a server execution file necessary for starting and operating a server process is prepared and operated. Then, a quantity data collecting and analyzing part(306) collects the performance data of the respective components from the quality measuring parts (382-384), and collects data provided with normal operability from an application life cycle managing part (305).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

TECHNICAL FIELD The present invention relates to a servlet [J] in the field of various information and communication services (hereinafter, referred to as services) such as an electronic commerce service and an information distribution service.
ava Servlet API specification, SUN Microsystems In
c., http://java.sun.com/products/servlet/index.htm
l], CORBA (Common Object Request Broker Architect
ure) [Object Management Group, "The Common Request
Broker: Architecture and Specification ", Rev.2.3,
ftp://omg.org/pub/docs/formal/99-12-01.pdf], EJB
(Enterprise Java Beans) [Enterprise Java Beans Sp
ecification Version 1.1, Sun Microsystems, http: //
java.sun.com/products/ejb/index.html], distributed object technology and component technology (hereinafter C0R
A distributed object represented by a BA object and a component represented by a servlet or EJB are collectively referred to as a "component"), or a component-to-component communication platform that executes communication between a component or a process containing the component. Or, it is related to the inter-process communication platform technology, and in particular, when testing or operating a distributed application (hereinafter, referred to as “AP”), individual components or server processes constituting the AP software (hereinafter, components and server processes are collectively referred to as servers). Measurement and collection of performance-related data such as reliability of normal operation and processing execution time, processing execution frequency, and processing response time, and evaluation and verification of the reliability and performance of the entire AP based on the collected data. Do, A Reliability and performance of P
The present invention relates to an AP test / operation management technique for identifying a bottleneck (hereinafter, collectively referred to as reliability and performance as quality) and performing an operation measure to secure an appropriate AP quality during operation.

[0002]

2. Description of the Related Art In recent years, in the field of e-commerce services on the Internet and in-house systems on an intranet, the need for timely development of service application software and quick response to changes in user needs has led to the need for components. There is an increasing demand for a tool that supports an AP development method of assembling software to build an application (hereinafter, simply referred to as “AP”) and further supports easy introduction of the developed AP software into an operation environment. In response to this, WWW (Wor1dWide W
eb), servlets, EJB, CORBA, and other technologies are attracting attention, and development support tools and AP server products based on these technologies are being introduced.

FIG. 7 shows a typical configuration of a system constructed by applying the Web, servlets, EJBs, CORBAs, and the like.
The standard distributed object communication protocol in CORBA is I
IOP (Intenet Inter-0RB Protocol) (Java Servlet API
specification, SUN Microsystems Inc., http: // java.
sun.com/products/servlet/index.html], and J2EE (Java2 Enterprise
Edition) uses RMI (Remote Method Invocation) [Remote Method
Invocation, Sun Microsystems Inc., http://java.su
n.com/products/rmi/index.html] or RMI over I
IOP. FIG. 8 shows an inter-component communication method for all components in IIOP, RMI, and RMI over IIOP.
In the case of communication between JB components, an inter-component communication method is shown in FIG.

[0004] Generally, in inter-component communication across processes and hosts, a processing request sent from one component is sent to the other component via a stub, ORB, and skeleton, and the processing result is reversed on the other hand. Is returned to the component that issued the processing request via the stub. This stub, ORB (Object Req
uest Broker), communication via the skeleton is called the statically activated communication method. In CORBA, there is another communication method called dynamic activation, but from the viewpoint of development efficiency and performance, the static communication method is generally used more often.

When a processing request is sent to an EJB component, the processing request is sent via an EJB container in addition to the skeleton.

As shown in FIG. 7, the application is:
It is composed of a server group (111-118, 141-149) that provides one service such as an e-commerce service. Each server, which is an individual component of the distributed AP, receives a processing request from another server, and as a result of executing the processing, a service is provided to a service user. A typical AP is a web browser (10
Web server process (111, 115) that receives a service request (hereinafter, referred to as an HTTP request) (121-125) via the HTTP protocol from (1-105) and performs user authentication as necessary.
EJB server process (112,113,116), CORBA server process (113), database (114,118), Web server process or servlet (142,146,148) that operates in the EJB server process and performs processing according to the contents of the HTTP request received by the Web server process , An EJB component (143, 147, 149) that operates in the EJB server process and executes functions necessary for business logic and various services, and a CORBA object (144) that operates in the CORBA server process. The Web server process receives the HTTP request from the browser, the servlet started in the Web server process or the EJB server process proceeds, and the RMI, RMI over IIOP, or IIOP protocol is used between the servlet and the EJB component or between the EJB components. Process requests (hereinafter, RMI / IIOP requests) via the server (131-136), and as a result, the entire distributed Web server process, servlet, EJB components, etc. are processed and the service user proceeds. Providing services.

[0007] In addition, in order to ensure the reliability and response performance of services in a system or a data center of a provider that generally provides electronic commerce and the like, a host CPU, which is currently managed as a system management, Monitoring of the normality and usage of hardware resources such as memories and hard disks is mainly performed. In addition, as AP management, monitoring the normal operation of packaged products that are provided as a package and are not distributed, such as groupware, a database system, and ERP, are mainly performed.

[0008]

[Problems to be Solved by the Invention] However, the quality such as the reliability and performance of a distributed AP that is constructed by applying technologies such as Web, servlet, EJB, and CORBA, and deployed and deployed and operated is measured, When processing time or response time deteriorates, identify the server that is the bottleneck from the distributed AP components to support evaluation and verification of distributed AP quality, and further improve implementation and operation. A distributed AP test and operation management mechanism for supporting quality assurance by coping has not yet been sufficiently established.

In order to differentiate itself from other businesses in highly competitive fields such as e-commerce, the quality must be evaluated and verified by testing before providing the service, and if necessary, the quality must be improved through implementation improvements and operational measures. It is effective to provide a service user with a service whose quality is ensured by making improvements. In addition, it is necessary to measure whether or not a predetermined quality is secured when a user uses the service after introducing the service, and to improve the implementation and take operational measures as necessary to secure the service quality. In the future, it is expected that the AP development method of combining distributed components such as EJB components and constructing an AP that provides the desired service by additionally expanding as necessary will be the mainstream. A mechanism for specifying a component or a server process that is a main cause of quality deterioration when the quality of an AP as viewed from a service user when using the service is deteriorated has not yet been established. At present, even if AP can be developed quickly using component technology,
Much time and effort is still spent identifying bottlenecks inherent in the AP software when detecting the quality degradation. Measures the quality of the servers that make up the AP when service quality deteriorates during AP testing and operation, and quickly provides accurate information to determine which components and server processes should be improved for implementation and operational measures. A mechanism (system) to provide is desired.

An object of the present invention is to measure the quality of each server constituting an AP, collect data, evaluate and verify the quality as an AP (entire server group as a constituent element),
By quickly and accurately identifying the server that is the bottleneck in ensuring the quality of P, and taking automatic action as necessary, A
The purpose of the present invention is to realize a mechanism that supports improvement and securing of the quality of services provided through the execution of P.

[0011]

The present invention provides a distributed object technology and a component technology (hereinafter, a distributed object and a component are collectively referred to as a component).
One or more hosts (computers) that use an inter-component communication platform or an inter-process communication platform to execute communication between a component or a process containing the component (hereinafter, the component and the process are collectively referred to as a server) based on the With respect to a set of one or more servers (defined as “distributed application” in the present specification) that are distributed and deployed in the same manner as above (defined as “distributed application”), a processing request is passed to the server from another component or process. When the processing result is returned from the server, the processing performance data including the start time and end time of the processing, the processing name, and the name of the server that executed the processing or, if the server is a component, its class name are acquired. And keep the server When a server requests processing and receives the processing result, the processing request issue time and the processing result reception time, the name of the requested processing, and the name of the server to which the processing request is issued or, if the server is a component, Obtain and hold the response performance data including the class name, calculate the processing execution time, processing execution frequency, processing request response time, etc. on each server from the obtained processing performance data and response performance data, and And a quality measurement unit that generates an event message when a predetermined threshold value is exceeded, and processing execution time, processing execution frequency, processing request response time, etc. relating to each server constituting one arbitrarily selected distributed application. Data collected from the quality measurement unit group that holds the data, and provided. And the location of the bottleneck in ensuring the performance quality of the distributed application from the dependency information between servers, and furthermore, the abnormal termination or abnormal termination that occurred in each server that constitutes one arbitrarily selected distributed application Time, process ID, and the like related to abnormal events such as transition to abnormal state
An individual server that constitutes an AP by collecting and providing information such as a component name and a component class name, and using a quality data collection / analysis unit that specifies a bottleneck that ensures a reliable quality of a distributed application. It measures the quality of the application and evaluates the quality as an application, and quickly and accurately identifies the server that is the bottleneck in ensuring the quality of the AP.

According to yet another aspect of the present invention,
A test / operation management execution condition management and test / operation management coordinator unit are provided so as to deal with automatic operation as necessary.

[0013]

Embodiments of the present invention will be described below.

FIG. 1 is a diagram showing the relationship between the components of the distributed AP test / operation management apparatus according to the present invention, and FIG. 2 is a diagram showing the process configuration during operation. A distributed AP test / operation management device according to an embodiment of the present invention has the following components. In FIGS. 1 and 2, the same components are denoted by different reference numerals for convenience of description, but this does not mean that the components are different.

(1) GUI unit (FIG. 1, 201, FIG. 2, 301) Upon receiving an AP operation request from an AP test executor or an operation manager,
Displays test result data and operation management status data.

(2) AP life cycle management unit (FIGS. 1, 202, FIG. 2, 305, 307, 308, 311, 314, 317) The following functions are provided.

(2-1) Holding of AP configuration information and deployment information (231 in FIG. 1) The component configuration information of the AP is stored. In addition, for each component, deployment information about which host is operating (and is operable) is held.

(2-2) Server entry and AP entry information
(FIG. 1, 232) Information necessary for starting and stopping the server process is stored as a server entry, and information necessary for starting and stopping the AP which is a set of servers is also stored. It is held as an AP entry. The server entry contains
It includes the operation host name of the server process, the absolute path name of the executable file, and the server process start command. A
The P entry is provided as a set of server entries,
It is used to start and stop a server group that is a component of the AP.

(2-2) (Re) starting and stopping the server process (2-3) Monitoring the operating status of the server process The operating status of the server process is monitored, and
An event message is generated and sent when a state change such as abnormal end or abnormal state is detected. The activation of the process is detected when the AP life cycle management unit receives a message transmitted when the quality measurement unit described later is activated.
The normal termination of the process is detected using a signal or a system call provided by the operating system. The abnormal termination of the process is detected by using the application life cycle management unit to perform polling abnormality to the server process or by using a signal or a system call provided by the operating system. The abnormal state of the process is detected from the polling abnormality to the server process. In addition, the number of times an abnormal termination or abnormal state occurs in any server process of an application component per unit time is periodically measured as the frequency of application abnormalities, and the low frequency of abnormalities is regarded as operational normality. deal with.

(2-4) (Re) start and stop of the AP (2-5) Start and stop of the test client (FIGS. 1, 213, FIG. 2, 310) Upon receiving a test operation instruction from the operation management coordinator unit (FIGS. 1, 202, 2, and 304), the test / operation management execution condition management unit
(FIG. 1, 203, FIG. 2, 303) held test execution condition information
(Fig. 1, 237) According to the test client unit (Fig. 1, 21
(3, Fig. 2, 310) and start the Web server process instead of the browser (Fig. 2, 212, Fig. 331) used by the service user.
(FIGS. 1, 241, 244 and FIGS. 2, 341) generate a test request. In addition, the test client unit started after the test ends is stopped and terminated.

(3) Test / operation management execution condition management unit (FIG. 1, 203, FIG. 2, 303) The following data is stored and provided to the requester.

(3-1) Component management option (233 in FIG. 1) Which component is polled, and how much time interval is required for performance data such as the process execution time and process execution frequency of which component What is the permissible limit value (threshold) for performance, such as the collection or the processing execution time and processing execution frequency of each component, and the processing execution time and processing execution frequency of each component exceed the predetermined threshold Data on whether an event message should be generated or sent when the event occurs.

(3-2) Host management option (234 in FIG. 1) Which host collects data on the usage rate of hardware resources such as CPU, memory, and hard disk,
For each host, what is the permissible limit (threshold) of their respective resource usage rates?

(3-3) Test / operation management policy (235 in FIG. 1) Receives an event message generated and transmitted during AP test / operation management, and automatically performs a desired AP operation countermeasure according to the content of the event message. Data used for automatically executing the apparatus according to the present invention. It is represented as a set of policy elements composed of a detected event type, a target host and a target server related to the event content, and a coping procedure.

The following are examples of the detected event type.

Specification of hardware resources such as component startup, component normal termination / abnormal termination, component polling failure, component performance threshold option exceeding thresholds, exceeding a specified number of times, CPU usage, etc. Exceeded the specified value, failure to execute the automatic response procedure according to the policy.

The coping procedure is defined by combining coping procedure elements. Examples of coping procedure elements include the following. AP (re) start, AP stop, AP block / block release (pause / resume), dependent component stop, server process (re) start, server process stop, server execution file specified host specified directory Deployment, component blocking / blocking release (pause /
Restart), notification of the event contents to the components of the device according to the present invention, details of the executed coping procedure and notification of coping / success execution of coping procedures to the components of the device according to the present invention,
Select the host with the lowest CPU usage.

(3-4) Report format (236 in FIG. 1) Processing execution time, processing execution frequency, various thresholds, processing performance guarantee rate processed and stored in text format and HTML format
Specify the display format of quality measurement result data such as (the number of processes whose processing execution time is equal to or less than the threshold value / the total number of processes) and the guaranteed response performance (the number of processing requests whose response time is equal to or less than the threshold value / the total number of processing requests) data.

(3-5) Test execution conditions (FIG. 1, 237) Host group for operating test clients that issue HTTP requests to the Web server process at the time of test execution, and test for operating on each of those hosts Data that defines test environment setting conditions for executing a test, such as the number of clients and the request generation time interval for each test client.

(3-6) Component Dependency (238 in FIG. 1) Defined based on the process call relationship between components, and represented by a set of pairs of “dependent component name-dependent component name”. It can be defined in a form that depends on a specific AP or a form that does not depend on the AP. If the definition is made dependent on a specific AP, the dependency is applied only when the AP is activated. If the two components have a dependency, the dependent component is forcibly stopped when the stop of the dependent component is detected.

(4) Test / operation management coordinator
(FIGS. 1, 202, 2, and 304) Upon receiving a test execution instruction from the AP test performer, instructing the AP life cycle management unit (FIGS. 1, 202, 2, and 308) to start the test client unit. The quality data collection and analysis unit (FIGS. 1, 202, 2, and 306) collects data on reliability and performance from each component constituting the AP.
And proceed with the test procedure. Also, the AP life cycle management unit (Fig. 1, 202, Fig. 2, 30)
8) and a quality measuring unit (FIG. 1, 205-210, FIG. 2, 313, 316,
319), the test / operation management execution condition management unit (FIGS. 1, 203, 2, 30) receives an event message indicating that an abnormality has occurred from another component of the system according to the present invention.
With reference to the test / operation management policy information (235 in FIG. 1) held in 3), a countermeasure procedure according to the contents of the relevant policy element is executed.

(5) Test input data generation unit (FIG. 1, 204, FIG. 2, 302) H handled by the Web server process (FIG. 1, 241, 244, FIG. 2, 341)
Search and extract URL (Unified Resource Locator) related to HTTP request issuance from the contents of TML file group,
Further, a list of test URLs (251 in FIG. 1) is generated with reference to the input argument data given by the test executor.

(6) Test Client Unit (FIGS. 1, 213, FIG. 2, 310) The test URL list (FIG. 1, 251) generated by the test input data generation unit (FIG. 1, 204, FIG. 2, 302) , An HTTP request is issued to the Web server process (FIG. 1, 241, 244, FIG. 2, 341).

(7) Quality Measurement Unit (FIGS. 1, 205-210, FIG. 2, 313, 316, 319) A test HTTP request is issued from the test client unit (FIG. 1, 213, FIG. 2, 310), and then the AP is executed. When a processing request is passed to a server such as a Web server process, a servlet, or an EJB component, and when a processing result is returned (measurement timing depends on the location where the preprocessor described later embeds the source code for quality measurement) And obtains and holds processing performance data such as a processing start time, a processing end time, a processing execution server name, and a processing name (method name) for each processing executed by the server. At this time, the processing execution time is calculated by subtracting the processing start time from the processing end time.

Further, the number of processes started to be executed per unit time is periodically calculated as a process execution frequency. When the processing execution time or the processing execution frequency exceeds a predetermined threshold, an event message is generated and transmitted.

In executing the application, the Web
When a processing request is issued to another server such as a server process, servlet, or EJB component, and when a processing result is received (measurement timing depends on the location where the preprocessor described later embeds the source code for quality measurement). Processing request issue time for the processing to be performed,
Response performance data such as a processing result reception time, a request processing name, and a processing request destination server name are acquired and held, and a value obtained by subtracting the processing request issue time from the processing result reception time is calculated as the processing response time. When the processing response time exceeds a predetermined threshold, an event message is generated and transmitted.

The quality measuring unit (FIGS. 1, 205-210, FIG. 2,
313, 316, 319) at the time of its own activation, the AP life cycle management unit (FIG. 1, 202, FIG. 2, 305, 307, 308, 311, 314, 317)
To the quality data collection / analysis unit (FIGS. 1, 202, FIG. 2, 306). Thereafter, it becomes possible for the AP life cycle management unit to poll the quality measurement unit for status confirmation, and for the quality data collection / analysis unit to collect performance data from the quality measurement unit.

(8) Quality Data Collection / Analysis Unit (FIGS. 1, 202, 2, 306) From the data acquired by the quality measurement unit, the processing execution time (history and average) of each server, the processing execution frequency, etc. Create data for Further, the process call relationship between the AP component servers is examined using the AP configuration information and the component dependency relationship information, and a set of components, that is, a bottleneck point of quality such as reliability and performance of the entire AP is determined using the relationship information. To identify.

(9) Report Creation Unit (FIGS. 1, 202, 2, and 306) A report is created in text format or HTML format based on the data on the reliability and performance of the AP collected by the quality data collection unit.

(10) Host operation data acquisition section (FIG. 1, 202,
(FIGS. 2, 309, 312, 315, 318) The usage rates of host resources such as CPU, memory, and hard disk are periodically checked. If any of those resource usages exceeds a predetermined threshold, an event message is generated and transmitted.

(11) Host operation data collection unit (FIGS. 1, 202, 2, and 307) Resources such as a CPU, a memory, and a hard disk in each host periodically or in response to a request from another component of the present invention. Is collected from the host operation data acquisition unit (FIGS. 1, 202, 2309, 312, 315, 318) and provided.

(12) Preprocessor (FIG. 1, 211) The above-mentioned quality measuring unit (FIG. 1, 205-210, FIG. 2, 313, 316, 319)
Into the various server executable files created by the application developer.

The above object is achieved by using the above components.

[Operation of the Embodiment] The AP test / operation manager performs the test / operation management of the distributed AP according to the following procedure.

(1) Incorporation of the Quality Measurement Unit into the Server Executable File When the AP developer or the AP test / operation manager performs the AP test / operation management, first, the source code file group or the Embed the quality measurement unit in the source code files automatically generated by the distributed AP platform or AP server, and then pass those source code files through a compiler to create the server executable required to start and operate the server process I do. As an example, FIG. 3 shows a procedure for embedding a quality measuring unit in the source code of a servlet or an EJB component created in the Java language and creating a server executable file for enabling the quality measurement of the component. FIG. 4 shows an example of a source code embedded for acquiring a processing start time or a processing request issue time in the server, and FIG. 5 shows an example of a source code for acquiring a processing end time or a processing result reception time. . Also,
FIG. 6 shows examples of two source code embedding methods.

When embedding a quality measuring unit for measuring the processing performance of a servlet according to this procedure, (a) AP
Insert the source code for acquiring the processing start time at the beginning of each method description in the servlet source code created by the developer, and insert the source code for acquiring the processing end time at the end (The method of “pattern 1” in FIG. 6) or (b) Insert the source code for acquiring the processing start time into the part immediately before the skeleton source code calls the code created by the AP developer. Then, the source code for acquiring the processing end time is inserted into the part immediately after (the method of “pattern 2” in FIG. 6). After that, a server executable file is created using a compiler.

When embedding a quality measuring section for measuring the processing performance of an EJB component according to this procedure, (c) the EJB container description in the source code file created by the tool attached to the application server product Insert the source code for acquiring the processing start time immediately before the description of invoking the method of the component, and insert the source code for acquiring the processing end time immediately after (the method of “Pattern 2” in FIG. 6), or ( d) Insert the source code for acquiring the processing start time in the part immediately before the skeleton source code calls the EJB container or the EJB component, and insert the source code for acquiring the processing end time in the part immediately after (see FIG. 6) "Pattern 2" inside
Method) or (e) the source code for acquiring the processing start time at the beginning of each method description in the source code of the EJB component created by the AP developer, and the processing end time at the last part (The method of “pattern 1” in FIG. 6). After that, a server executable file is created using a compiler.

When embedding a quality measuring unit for measuring the response performance related to the processing requested by the servlet according to this procedure, (a) immediately before starting the processing request with the servlet source code created by the AP developer, A source code for acquiring the processing request issue time is embedded, and a source code for acquiring the processing result reception time is embedded immediately thereafter (method of “pattern 2” in FIG. 6).
Or (b) Insert the source code for acquiring the processing start time at the beginning of each method description in the source code of the stub called by the EJB component and the source code for acquiring the processing end time at the last part (The method of “pattern 1” in FIG. 6). After that, a server executable file is created using a compiler.

When embedding a quality measuring unit for measuring the response performance related to the processing requested by the EJB component according to this procedure, (a) immediately before starting the processing request with the source code of the EJB component created by the AP developer. Either embed the source code for acquiring the processing request issue time and the source code for acquiring the processing result reception time immediately after (the method of “pattern 2” in FIG. 6), or (b)
The source code for acquiring the processing start time is inserted at the beginning of each method description in the source code of the stub called by the EJB component, and the source code for acquiring the processing end time is inserted at the last part (see FIG. 6). "Pattern 1" method). After that, a server executable file is created using a compiler.

(2) Setting of Server Entry and AP Entry The AP test / operation developer registers the server entry and the AP entry in the AP life cycle management unit via the GUI unit (361 in FIG. 2).

(3) Setting of Component Monitoring Condition and Performance Data Acquisition Condition The AP test / operation developer registers the component management option information in the test / operation management execution condition management unit via the GUI unit (361 in FIG. 2). ).

(4) Setting of Test / Operation Management Policy The AP test / operation developer registers the test / operation rational policy information in the test / operation management execution condition management unit via the GUI unit (361 in FIG. 2). .

(5) AP Startup The AP test / operation developer specifies an AP entry name in the GUI unit, and sends the A to the AP life cycle management unit existing in the domain manager through the test / operation management coordinator unit.
P activation is requested (361 in FIG. 2). The AP life cycle management unit existing in the domain manager refers to the AP entry with the designated name, confirms individual server entries constituting the AP entry, and, for each server entry, an AP existing in the server manager. Request the life cycle management unit to start the server process (Fig. 2, 36
3, 364, 366, 368). AP existing in server manager
The life cycle management unit starts the server process using the designated server execution file on the designated host (FIG. 2, 377-379). When the server process is started, all the servers constituting the AP are started by this procedure, and as a result, the AP is started.

(6) Test Client Activation In performing the AP test, the test executor requests the test / operation management coordinator unit to activate the test client via the GUI unit (361 in FIG. 2). The test / operation management coordinator unit obtains the test execution conditions from the test / operation management execution condition management unit, and based on the contents, activates the specified number of test clients on the specified host and the server manager. (3, 363). The activated test client sends a test HTTP request to the Web server process at specified time intervals (380 in FIG. 2).

(7) Performance Data Acquisition When the Web server process receives an HTTP request from a test client (during a test) or a browser (during operation) used by a service user (FIG. 2, 370, 380), the Web server process or the EJB server process An RMI / IIOP request is sent from the servlet started in step (3) to the EJB component (371 in FIG. 2) or from the EJB component to another EJB component (372 in FIG. 2), and as a result, the service execution process is performed on the entire component. When the process requested by each component is executed, the quality measurement unit performs processing start time, process end time, process name (method name), process execution server name or the server of the process executed by each component. If it is a component, the processing performance data including the class name is obtained (FIG. 2, 373-375), and the processing execution time (history,
Data such as (average) and processing execution frequency are calculated and stored. Also, when each component requests the other component to execute processing and receives a processing result, the quality measuring unit issues a processing request issued time from each component, a processing request receiving time, a processing result receiving time, and a processing name. (Method name) and the response performance data including the name of the server to which the processing request is issued or, if the server is a component, the class name thereof are acquired (FIG. 2, 37).
3-375), data such as response time (both history and average) is calculated and accumulated based on the data.

(8) Automatic Handling When an Error Occurs During an AP test or operation, an error such as an abnormal stop of a process or an excess of a processing execution time or a processing execution frequency of a component exceeds a threshold value in a life cycle management unit (FIG. 2, 309, 312, 315, 318), an event message to that effect is sent to the test / operation management coordinator unit (FIG. 2, 365, 367, 369 if the process is abnormally stopped; 2,389
-391). The test / operation management coordinator refers to the test / operation management policy information held by the test / operation management execution condition management unit and executes a coping procedure.

As an example, an execution example of the automatic handling of the abnormal process stop is as follows. First, the AP life cycle management unit existing in the server manager detects a process abnormal stop in a Web server process, an EJB server process, or the like (FIG. 2, 377-379). Next, the AP life cycle management unit existing in the server manager is tested and tested through the AP life cycle management unit existing in the domain manager.
An event message to that effect is sent to the operation management coordinator unit (365, 367, 369 in FIG. 2). The policy element data related to the event (in this case, “server abnormal termination”) generated from the test / operation management policy information held by the test / operation management execution condition management unit is searched. Here, if the corresponding policy element data is not found, no coping procedure is executed. If the relevant policy element data is found, the coping procedure is advanced according to the contents described in the policy element.

For example, the content of the coping procedure is as follows (a)-
In the case of (c), the procedure is executed in the following procedure.

(A) Selection of the Host with the Lowest CPU Usage The CPU usage information of each operating host is collected from the host operation data collection unit, and the host with the lowest usage is selected.

(B) Deployment of Server Executable File to Selected Host From the host specified by the AP test / operation manager,
Deploy the server executable file to a predetermined directory on the host selected by the above procedure.

(C) Starting the server process on the selected host The server process is started using the server execution file in a predetermined directory on the selected host.

(9) Reliability / Performance Data Aggregation and Report Creation When the AP test / operation management policy is applied to test and operate the AP, the quality data collection unit The performance data of each of the constituent components is collected (385-388 in FIG. 2), and the abnormality frequency data of the server process forming the AP is collected from the AP life cycle management unit (392 in FIG. 2). Further, the report creating unit creates a bow reliability and performance report based on the data collected by the quality data collection unit.

[0063]

As described above, according to the present invention,
Testing and operation management can be performed for each AP, which is an aggregate of servers. In addition, reliability and performance can be measured and evaluated in units of server processes and components that are components of the distributed AP, and server processes and bottlenecks that ensure the reliability and performance of applications can be measured. The component can be specified.

Further, by using the test input data generation unit and the test client unit, the test can be automatically executed, and the load of the test execution is reduced as compared with the case where the test is performed manually.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the relationship between components of the present invention.

FIG. 2 is a diagram illustrating which component of the present invention operates on which process.

FIG. 3 is a diagram illustrating an example of a procedure for incorporating a quality measuring unit, which is one component of the present invention, into a server execution file.

FIG. 4 is a diagram showing an example of a source code embedded as a part of a quality measuring unit for obtaining a processing start time and for obtaining a processing request issuance time.

FIG. 5 is a diagram illustrating an example of a source code embedded as part of a quality measuring unit for acquiring a processing end time and for acquiring a processing result reception time.

FIG. 6 is a diagram illustrating an example of a method of embedding a source code of a quality measuring unit in a server execution file.

FIG. 7 is a diagram showing a typical configuration of a system constructed by applying Web, servlet, EJB, CORBA, and the like.

[Fig. 8] In the system where components such as COBRA object, EJB component and servlet communicate with each other,
FIG. 4 is a diagram illustrating a case of static startup.

[Fig. 9] In the system where components such as COBRA object, EJB component, servlet communicate with each other,
FIG. 4 is a diagram illustrating a case of static startup.

[Explanation of symbols]

101-105 ... Browser 111,115 ... Web server process 112,116,117 ... EJB server process 113 ... CORBA server process 114,118 ... Database server 121-125 ... HTTP request 131,132,134,135 ... IIOP, RMI or RMI over II
Processing request using OP 133,136… Processing request using communication protocol for database access 141,142,145,146,148… Servlet 144… CORBA object 143,147,149… EJB component 151… Distributed application 201… GUI part 202… Test and operation management coordinator part, application life cycle management Department, quality data collection and analysis department
/ Report creation unit, host operation data acquisition unit / host operation data collection unit 203… test / operation management execution condition management unit 204… test input data generation unit 205-210… quality measurement unit 211… preprocessor 212… browser 213… for testing Client part 241,244… Web server process 242,243,245… EJB server process 246… CORBA server process 251… Test URL list 301… GUI part 302… Test input data generation part 303… Test / operation management execution condition management part 304… Test / operation management Coordinator part 305,308,311,314,317… AP life cycle management part 306… Quality data collection analysis part / report creation part 307… Host operation data collection part 309,312,315,318… Host operation data acquisition part 310… Test client part 313,316,319… Quality measurement part 331… Browser 341… Web server process 342,343… EJB server process

Claims (18)

[Claims] An inter-component communication platform or a distributed application operating environment including one or more inter-process communication platforms for executing communication between components which are software components or processes including the components, the system includes: It stores information necessary to start and stop a server, such as a process including its components, as a server entry, and can be used for starting a distributed application unit, which is a set of one or more servers that provide one service. A set of server entries required for stopping is maintained as an application entry, and the start and stop of the distributed application are controlled using the application entry. An application lifecycle management unit that holds timed configuration information, and when a processing request is passed from another component or process to the server, and when a processing result is returned from the server, the start time and end time of the processing, and the processing name , And the server name that executed the process or, if the server is a component, obtains processing performance data including the class name, and when the server requests processing from another server and receives the processing result, the processing request A performance data acquisition unit for acquiring response time data including an issue time and a process result reception time, a name of a requested process, and a server name of a process request issuance destination or, if the server is a component, a class name thereof; From the processing performance data and response performance data acquired by the performance data acquisition unit, A quality measurement unit that calculates a processing execution time, a processing execution frequency, a processing request response time, and the like, and generates an event message when a predetermined threshold is exceeded for each of them, and one arbitrarily selected distributed application. Process execution time, process execution frequency,
Data such as processing request response time is collected and provided from the quality measurement unit group that holds the data, and it becomes a bottleneck in ensuring the performance quality of distributed applications based on the data and application configuration information and dependency information between servers. The time and process I related to abnormal events such as abnormal termination or transition to an abnormal state that occurred in each server constituting one arbitrarily selected distributed application are identified.
Collects and provides information such as D, component name, and component class name from the application life cycle management unit and quality measurement unit, and specifies and analyzes quality data to identify bottlenecks in ensuring the reliability and quality of distributed applications. A distributed application test / operation management system comprising: 2. The distributed application test / operation management system according to claim 1, wherein the quality measurement unit is an EJB (Enterprise Java).
va Beans) In collecting the processing performance data of the component, the performance data acquisition unit which is a part of the quality measurement unit is included in the inter-component communication platform or the inter-process communication platform, and the part of the quality measurement unit is E.
It operates as an interceptor that executes an arbitrary process immediately before a processing request reaches the JB component and immediately after a processing result is output from the EJB component, and operates immediately before the processing request reaches the EJB component (processing start time). The performance data acquisition unit acquires the processing performance data including the time immediately after the processing result is output (processing end time), the name of the executed processing, and the name of the EJB component that executed the processing or the class name of the EJB component. A distributed application test and operation management system, characterized in that the quality measurement unit holds the application. 3. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects processing performance data of the EJB component, a performance data acquisition unit, which is a part of the quality measurement unit, includes a skeleton. The time at which a processing request passed from another component or process to the EJB component passes through the performance data acquisition unit in the skeleton (processing start time), and the processing result returned from the EJB component to the other component is stored in the skeleton The time (processing end time) passed through a certain performance data acquisition unit, the name of the executed process, and the E that executed the process
A distributed application test / operation management system, wherein a performance data acquisition unit acquires processing performance data including a JB component name or a class name of the EJB component, and the quality measurement unit holds the performance data. 4. When the quality measurement unit collects processing performance data of an EJB component while using the distributed AP operating environment, a performance data acquisition unit which is a part of the quality measurement unit is embedded in an EJB container. The time at which a processing request passed from another component or process to the EJB component has passed through the performance data acquisition unit in the EJB container (processing start time), and the processing result returned from the EJB component to the other component is E
Processing performance data including the time (processing end time) passed through the performance data acquisition unit in the JB container, the name of the executed process, the name of the EJB component that executed the process, or the class name of the EJB component is stored in the performance data acquisition unit. 3. The distributed application test / operation management system according to claim 2, wherein the information is acquired by the quality measurement unit. 5. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects processing performance data of the EJB component, the performance data acquisition unit, which is a part of the quality measurement unit, includes an EJB data collection unit. Embedded in the component, the time immediately after the processing request is passed to the EJB component from another component or process (processing start time),
The performance data acquisition unit acquires the processing performance data including the time immediately before the processing result is returned from the EJB component (processing end time), the name of the executed processing, the name of the EJB component that executed the processing, or the class name of the EJB component. A distributed application test and operation management system, characterized in that the quality measurement unit holds the application. 6. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects the processing performance data of the servlet, a performance data acquisition unit that is a part of the quality measurement unit includes an An interceptor that exists in the communication platform or inter-process communication platform and executes any processing before the processing request reaches the servlet and after the processing result is output from the servlet, and operates as an interceptor immediately before the processing request reaches the servlet (processing Start time), the time immediately after the processing result is output from the servlet (processing end time), the name of the executed processing, and the processing performance data including the name of the servlet that executed the processing or the class name of the servlet. Acquired by the department and maintained by the quality measurement department Distributed application test and operation management system. 7. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects processing performance data of the servlet, a performance data acquisition unit, which is a part of the quality measurement unit, transmits the servlet to the servlet. Embedded, the time immediately after the processing request is passed to the servlet from another component or process (processing start time), the time immediately before the processing result is returned from the servlet (processing end time),
A distributed application test / operation characterized in that the performance data acquisition unit acquires processing performance data including the name of the executed process, the name of the servlet that executed the process, or the class name of the servlet, and the quality measurement unit holds it. Management system. 8. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects the processing performance data of the servlet, the performance data acquisition unit, which is a part of the quality measurement unit, includes a skeleton. Embedded, the time when the processing request passed from the other component or process to the servlet passed through the performance data acquisition unit in the skeleton (processing start time), and the processing result returned from the servlet to the other component or process is in the skeleton The time when the data passed through the performance data acquisition unit (processing end time),
A distributed application test / processing system characterized in that the performance data acquisition unit acquires processing performance data including the name of the executed process, the name of the servlet that executed the process, or the class name of the servlet, and the quality measurement unit holds the data. Operation management system. 9. The distributed application test / operation management system according to claim 1, wherein the system transmits the quality measurement unit, a performance data acquisition unit included in the quality measurement unit, and an event message from the performance data acquisition unit. It has a test and operation management coordinator unit that can receive, and for each process executed by the server, when the performance data acquisition unit acquires its process start time and process end time,
The quality measurement unit calculates the processing execution time of the process, and when the event that the processing execution time exceeds the specified time occurs a specified number of times within the specified time range, the quality measurement unit performs the test and operation. A distributed application test / operation management system, wherein an event message to that effect is sent to the management coordinator, and the test / operation management coordinator that has received the event message proceeds with a coping procedure with predetermined contents. 10. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects response performance data of the EJB component, a performance data acquisition unit that is a part of the quality measurement unit includes a component. Inter-process communication platform or inter-process communication platform, operates as an interceptor that executes arbitrary processing immediately after a processing request is issued from the EJB component and immediately before a processing result reaches the EJB component, and a processing request is issued from the EJB component Time immediately after the execution (processing request issuance time), the time immediately before the processing result reaches the EJB component (processing result reception time), the name of the executed processing, the name of the EJB component that executed the processing, or the class of the EJB component A distributed application test / operation management system, wherein a response data including a name is acquired by a performance data acquisition unit and held by a quality measurement unit. 11. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects response performance data of the EJB component, a performance data acquisition unit that is a part of the quality measurement unit includes a stub. The time at which a processing request issued by the EJB component to another server passes through the performance data acquisition unit in the stub (processing request issuance time), and the processing result received by the EJB component from another server is in the stub The response performance data including the time at which the data passed through the performance data acquisition unit (processing result reception time), the name of the requested process, and the name of the server to which the processing request was issued or, if the server is a component, its class name, A distributed application characterized by being acquired by an acquisition unit and held by a quality measurement unit test·
Operation management system. 12. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects response performance data of the EJB component, the performance data acquisition unit, which is a part of the quality measurement unit, includes an EJB data collection unit. Embedded in the component, the time immediately before the EJB component issues a processing request to another server (processing request issue time), and the time immediately after the EJB component receives a processing result from another server (processing result reception time) The performance data acquisition unit acquires the name of the requested process, the name of the server to which the process request is issued, or the response performance data including the class name if the server is a component, and the quality measurement unit holds it. Features a distributed application test and operation management system. 13. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects response performance data of the servlet, a performance data acquisition unit that is a part of the quality measurement unit includes an It exists in the communication platform or inter-process communication platform and operates as an interceptor that executes any processing after the processing request is issued from the servlet and before the processing result reaches the servlet, and immediately after the processing request is issued from the servlet. Response performance data including the time (processing request issuance time), the time immediately before the processing result reaches the servlet (processing result reception time), the name of the executed processing, and the servlet name or the servlet class name of the executed processing The performance data acquisition unit acquires the data and the quality measurement unit A distributed application test and operation management system, characterized by the fact that it is maintained. 14. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects response performance data of the servlet, a performance data acquisition unit, which is a part of the quality measurement unit, transmits the response data to the servlet. Embedded, the time immediately before the servlet issues a processing request to another server (processing request issuance time), the time immediately after the servlet receives a processing result from another server (processing result receiving time), the requested processing The performance data acquisition unit acquires response performance data including the name of the server and the name of the server to which the processing request is issued or, if the server is a component, the class name, and the quality measurement unit holds the response performance data. Application test and operation management system. 15. The distributed application test / operation management system according to claim 1, wherein when the quality measurement unit collects the response performance data of the servlet, the performance data acquisition unit, which is a part of the quality measurement unit, includes a stub. Time at which the processing request issued by the servlet to another server passes through the performance data acquisition unit in the stub (processing request issuance time), and the processing result received by the servlet from the other server is acquired in the stub The performance data acquisition unit outputs the response performance data including the time at which the processing request was passed (processing result reception time), the name of the requested processing, and the name of the server to which the processing request was issued or the class name if the server is a component. A distributed application test and operation management system that is acquired and held by a quality measurement unit. 16. The distributed application test / operation management system according to claim 1, wherein the system transmits the quality measurement unit, a performance data acquisition unit included in the quality measurement unit, and an event message from the performance data acquisition unit. It has a receivable test / operation management coordinator unit, and for each processing request issued by the server to another server, when the performance data acquisition unit acquires the processing request issue time and the processing result reception time, The quality measurement unit calculates the response time of the processing request, and when the response time exceeds the specified time and the specified number of times occurs within the specified time range, the quality management unit An event message to that effect is sent to the coordinator, and the test / operation management coordinator receiving the event message Wherein the advancing deal procedure with the specified contents, distributed application testing and operation management system. 17. The distributed application test / operation management system according to claim 1, wherein the HT requests the servlet to be started on the Web server based on the contents of the HTML file returned from the Web server to the browser.
A test input data generation unit that lists a URL description portion corresponding to the TP request, and a test client unit that generates a test HTTP request using the obtained URL list and argument data given by the test executor A distributed application test and operation management system, characterized in that: 18. The distributed application test / operation management system according to claim 1, wherein any two servers depend on a specific application (a server of a dependence source and a server of a dependence destination). Are both active and dependencies are valid only when a specific application is running), or do not depend on a specific AP (if both the dependent and dependent servers are active) A test / operation management coordinator that sets dependencies in the (dependency valid) form and, if a dependency is set, terminates the dependent server when the termination of the dependent server is detected. A distributed application test and operation management system, comprising: