JP2010118072A - Trace processing method, program and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To rapidly specify a bottle neck while minimizing the overhead of trace collection. <P>SOLUTION: When a Web server receives a request telegram from a client 24, a trace condition embedding part 28 delivers a relay code to the next application while embedding a trace collection condition therein. Trace collection parts 30-1 to 30-3 process the relay code, upon receipt of request telegram, while discriminating presence/absence of trace collection. The trace collection condition includes a function ID which indicates a trace range in application units, and a relay ID which records the passed applications. The trace condition embedding part 28 further includes intermittent switch information which instructs a trace collection ratio to the receiving frequency of the request telegram in an instruction state with trace collection by the switch information, and embeds a trace collection condition for switch-on to the relay code according to the ratio of the intermittent switch information. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

The present invention relates to a trace processing method, program, and apparatus for a distributed application system that implements an application service via a network, and in particular, a trace processing method that collects a trace and identifies a bottleneck location that caused a response deterioration, The present invention relates to a program and an apparatus.

  In a network-compatible distributed application system, a plurality of servers are configured in multiple layers, and applications that operate on the servers are combined in a complex manner, and each service operates in cooperation with each other to realize one service.

  In such a complex distributed application system, in order to identify the bottleneck part that caused the response deterioration, investigation and analysis are performed by arranging trace information that records the behavior of the application in time series. ing.

  The collection of trace information is usually the entry point when the executed application (including a program or a module in the program) is called and the exit point when the processing ends, and the time when each point is passed. Recorded and stored with information.

  FIG. 16 is a conventional collection of trace information. The web server 200, the application server 202, the database server 204, and the database 206 are arranged in a hierarchical configuration, and a request message from a client is processed by executing an application of each server. A response is made by searching the database 206 or the like.

  Trace information is collected at each of the points P1 to P12 which are the entrance and exit of the executed application. Here, when the transaction response t1 deteriorates, the bottleneck portion is identified by comparing the transaction responses w1, w2, a1, a2, d1, and d2 of each server obtained from the trace information with those at the normal time. be able to.

  Further, in order to facilitate the connection of trace information for examining invocation relations related to complicated applications, not only the time information of the entrance and exit, but also FIG. As shown in FIG. 4, there is a technique for recording information such as recording information of an application to be called next. FIG. 17A shows an example of transaction distribution. Corresponding to this, call hierarchy information of each application is generated as shown in FIG.

JP-A-1-277940 JP-A-4-84234 JP 2001-318809 A

  However, the conventional distributed application system trace processing method has the following problems. First, the conventional trace processing method has too much overhead. Information such as entry information, exit information, and call relation information of each application to be traced in a distributed application system is a very large amount of information, and the overhead to the system cannot be ignored.

  For this reason, it is virtually impossible to always collect traces in preparation for unforeseen circumstances, and operations such as collecting traces within a range that will not affect the system as much as possible after a trouble occurs. However, since it can only be done, it takes time to pursue the cause of the bottleneck.

  The conventional trace processing method is complicated to operate in a distributed application system. In FIG. 18, when collecting trace information for a distributed application system, switches 208, 210, and 212 for instructing the presence / absence of tracing are provided for each system and each application in each server 200, 202, and 204. For this reason, when trying to collect traces after a trouble has occurred, the administrator needs to turn on a number of switches, and it is necessary to determine which switch should be turned on. It has been.

An object of the present invention is to provide a trace processing method, program, and apparatus for a distributed application system that can quickly identify a bottleneck while minimizing the overhead of trace collection.

FIG. 1 is a diagram illustrating the principle of the present invention. The present invention has a plurality of applications that operate on a plurality of hierarchically arranged servers, and when a request message is received from the client 24, responds by executing processing while sequentially transferring relay codes between the applications. For the trace processing method of the distributed application system,
A trace condition embedding step of embedding a preset trace collection condition including switch information instructing whether or not to collect a trace into a relay code and passing it to the next application when a request message is received;
A trace collection step that is provided in each of the applications and that determines and processes the presence or absence of trace collection based on the switch information of the trace collection condition embedded in the relay code when the request message is received.

  Here, the trace condition embedding step further includes, as the trace collection condition, a function ID that instructs a range of collecting the trace for each application and a relay ID that records the passed application.

  The trace condition embedding step generates a unique relay ID when the switch information indicates the presence of trace collection, and the trace collection step connects the identifier of the application to be called next to the relay ID.

  The trace condition embedding step includes, as a trace collection condition, a ratio of intermittent switch information, including intermittent switch information that indicates a ratio of collecting a trace with respect to the number of times a request message is received in an instruction state with trace collection by switch information. Follow the procedure to embed trace collection conditions in the relay code.

  The intermittent switch information can also identify the route of the request message and indicate the ratio of collecting the trace with respect to the number of times the request message has been received for each identified route. The intermittent switch indicates the ratio of collecting the trace in the range of 0% to 100% with respect to the number of receptions of the request message.

  The analysis display step extracts the trace information of the route according to the relay ID from the trace information collected by each application in the trace collection step and sorts it in order of time. Based on the sorting result, the relationship between the route and each application is processed. Generate and display.

  Specifically, in the analysis display step, the processing time from when the request message is received until the next delivery is displayed is displayed corresponding to each application serving as a passing point. In the analysis display step, the ratio of the processing time until each application receives the request message and delivers it to the next is the passing point with respect to the total processing time until the application at the entrance receives the request message and responds. You may display corresponding to each application.

The present invention provides a program executed by a computer constituting a server having a first layer application that first processes a request message from a client. This program
A trace condition embedding step of embedding a preset trace collection condition including switch information instructing whether or not to collect traces in the relay code when the request message is received from the client and passing it to the next application;
The trace collection condition switch embedded in the relay code of the request message when the request message is received from another application is processed by determining whether or not the trace is collected from the switch information of the trace collection condition embedded in the relay code. A trace collection step for determining whether or not trace collection is performed based on information, and processing,
Is executed.

  The present invention provides a program executed by a computer constituting a server including an application in the second and subsequent layers that processes a request message from a client for the second time and thereafter. When this program receives a request message from another application, it executes a trace collection step that determines whether or not trace collection is performed based on the switch information of the trace collection condition embedded in the relay code of the request message and processes it. It is characterized by.

  The present invention has a plurality of application processing units that operate on a plurality of servers arranged in a hierarchy, and when a request message is received from the client 24, the relay codes are sequentially arranged between the application processing units 26, 34, and 36. Targeting a distributed application trace processing device that responds by executing a process while passing it to the relay, when a request message is received, a preset trace collection condition including switch information indicating whether or not to collect the trace is used as a relay code. Trace condition embedding unit 28 that is embedded and passed to the next application, and application processing units 26, 34, and 36, and traces based on the switch information of the trace collection condition embedded in the relay code when the request message is received Trace collection units 30-1 to 30-3 for determining the presence or absence of collection and processing And said that there were pictures.

The details of the program and the trace processing apparatus of the present invention are basically the same as those of the trace processing method.

  According to the present invention, in a distributed application system, overhead due to trace collection can be minimized and localized, so that operation for always collecting traces can be realized, and bottlenecks can be quickly identified when trouble occurs.

In addition, since the distributed application system is complicated and has a complicated configuration, there is also a problem that it is difficult to know where and what application is operating. Even for such a problem, the present invention can easily grasp how the application operates in conjunction with the application.

Principle explanatory diagram of the present invention Illustration of a distributed application system to which the present invention is applied Block diagram of functional configuration of trace processing according to the present invention Block diagram of the functional configuration of the trace condition embedding unit provided in the web server of FIG. Illustration of trace collection condition and indirect switch information embedded in relay code Illustration of relay ID in the present invention Explanatory drawing of the trace processing according to the present invention to which a relay code in which a trace collection condition is embedded in a request message is added and delivered Flowchart of trace condition embedding process according to the present invention Flow chart of trace collection processing according to the present invention Flowchart of trace information analysis display processing according to the present invention Illustration of trace information arranged for analysis display Explanatory drawing of the example of a display of the analysis result obtained from the arrangement information of FIG. Illustration of other trace information arranged for analysis display Explanatory drawing of the example of a display of the analysis result obtained from the arrangement information of FIG. Explanatory drawing of the other example of a display of the analysis result obtained from the organization information of FIG. Illustration of trace collection points in a conventional distributed application system Explanatory drawing of the inheritance relationship of applications collected by conventional trace processing Explanatory diagram of switch operation that determines the presence or absence of trace required for each application in a conventional distributed application system

  FIG. 2 is an explanatory diagram of a distributed application system to which the trace processing of the present invention is applied. In FIG. 2, the distributed application system 10 has a plurality of servers arranged in a hierarchy. In this example, web servers 12-1, 12-2, 12-3, and 12-4 are provided on the entrance side of a system that receives a request message from a client 24 via the Internet 22, and the next hierarchy Are provided with application servers 14-1, 14-2, 14-3, 14-4, and further provided with database servers 16-1, 16-2 in the third hierarchy, with respect to the database servers 16-1, 16-2. Databases 18-1 and 18-2 are connected to each other.

  The web servers 12-1 to 12-4, the application servers 14-1 to 14-4, and the database servers 16-1 and 16-2 arranged in a hierarchy in the distributed application system 10 are input / output connected to each other. For example, in the search of the databases 18-1 and 18-2, a response to the request message is executed through a complicated processing path that traces the hierarchically arranged application.

  In a plurality of hierarchically arranged servers that construct the distributed application system 10, when a request message is received from the client 24 via the Internet 22 by the web servers 12-1 to 12-4 on the entrance side, the application It works while passing keywords called relay codes in order.

  A management server 20 is provided for the hierarchically arranged servers constituting the distributed application system 10. The management server 20 sets, for the web servers 12-1 to 12-4 located on the entrance side, a trace collection condition that is added to the request message from the client 24 and embedded in the relay code passed between applications. Further, the management server 20 performs a process of collecting, analyzing and displaying the trace information accumulated in each server as necessary.

  FIG. 3 is a block diagram of a functional configuration of the trace processing according to the present invention, and shows one system from the client 24 to the database 18 in the distributed application system 10 of FIG.

  In FIG. 3, the request message from the client 24 is received by the web server 12 located on the entrance side via the Internet 22, and the application of the web server 12, the application of the next application server 14, and the application of the database server 16 are also received. After execution, the database 18 is accessed, and the access result is returned to the client 24 side via the application of the database server 16, the application of the application server 14, and the application of the web server 12.

  For this reason, each of the web server 12, the application server 14, and the database server 16 is provided with application processing units 26, 34, and 36 that function as application programs.

  The web server 12 located at the entrance of the distributed application system 10 is provided with a trace condition embedding unit 28. When the trace condition embedding unit 28 receives a request message from the client 24, the trace condition embedding unit 28 embeds a trace collection condition set in advance by an operation of the management console 38 of the management server 20 in a relay code passed between applications. The data is transferred to the application processing unit 34 of the application server 14.

  The trace collection units 30-1, 30-2, and 30-3 determine the presence / absence of trace collection based on the trace collection condition embedded in the relay code, and if there is a trace instruction, the application processing unit 26 based on reception of the request message. , 34, and 36 are collected and stored in the trace files 32-1, 32-2, and 32-3. If there is no instruction to collect a trace, the trace is not collected, that is, the trace is ignored.

  FIG. 4 is a block diagram of a functional configuration of the trace condition embedding unit 28 provided in the web server 12 of FIG. In FIG. 4, the trace condition embedding unit 28 includes a trace collection condition storage unit 42, an indirect switch processing unit 44, a relay ID generation unit 45, a trace collection condition code holding unit 46, and a trace collection condition code embedding unit 48. It is done.

  In the trace collection condition storage unit 42, a switch 66, a function ID 68, and a relay ID 70 are stored as trace collection conditions by operating the management console 38 provided in the management server 20 of FIG. The switch 66 stored in the trace collection condition storage unit 42 instructs to turn on or off the trace, and when the application processing unit 26 in FIG. 3 receives the request message from the client 24 and executes the processing according to this instruction, the switch 66 is stored. Whether to collect traces or not.

  FIG. 5 is an explanatory diagram of the trace collection conditions and indirect switch information set in the trace collection condition storage unit 42 of FIG. 4 by an instruction from the management console 38. The trace collection condition 64 shown in FIG. 5A includes three switches 66, a function ID 68, and a relay ID 70 as embedded information.

  The switch 66 has a size of 4 bytes, the function ID 68 has the same size of 4 bytes, and the relay ID 70 has a variable length. The switch 66 instructs on / off whether to collect a trace or not. As shown in FIG. 3, the switch 66 sets the trace on / off only for the web server 12 having the application processing unit 26 for executing the application of the first layer, and once for the web server 12 for collecting the trace. If the switch is turned on, the trace collection condition including the switch 66 is embedded in a relay code passed between applications and sent to the subsequent second and subsequent applications.

  Therefore, in the subsequent application processing units 34 and 36, trace collection is automatically performed based on the instruction of the switch 66 of the trace collection condition 64 embedded in the relay code without performing the switch operation from the management console 38 side. Is called. For this reason, the switch 66 functions as an interlock switch for collecting traces between applications.

  The function ID 68 determines the trace range so that a trace is collected only for a specific function. Therefore, it is possible to specify trace collection only for a specific application in a series of application routes accompanying reception of a request message.

  The relay ID 70 is information for connecting application calling relationships at the time of analysis of trace information. By following the relay ID 70, it is possible to analyze what application route a certain request message has passed. As the relay ID 70, an ID unique in the distributed application system, for example, an ID obtained by adding a process ID to a time stamp is created as an initial value. Try to connect to.

  FIG. 6 is an explanatory diagram of the relay ID in the trace processing according to the present invention. The application 74 is located in the first hierarchy on the entrance side, the applications 75 and 76 are located in the second hierarchy, and the application 78 is located in the third hierarchy. , 80 and 82 are located. When a request message is received for the first layer application 74, a relay code in which the trace collection condition of FIG. 5A is embedded is added. The relay ID at this time is “123” as a unique ID for the distributed application system. ".

  Subsequently, a request message to which a relay code is added is delivered from the first layer application 74 to the second layer application 75 or application 76. When the request message is delivered to the application 75, the first relay ID “123” is assigned. The relay ID “123.b1” is added with the identifier “b1” of the application 75. In addition, when the data is transferred to the application 76, the relay ID “123.b2” is added with the identifier “b2” of the application 76.

  Similarly, when the relay ID is transferred from the second layer application 75 to the third layer application 78, the relay ID is “123.b1.c1”, and is transferred from the second layer application 76 to the third layer application 80. The relay ID is “123.b2.c2”. Further, when the application is transferred from the second layer application 76 to the third layer application 82, the relay ID is “123.b2.c3”.

  Such a relay ID is stored together with trace data that accompanies trace collection in the application of each layer. By tracing the relay ID when the trace information is collected by the management server, it is possible to determine what route a request message has. You can analyze whether you have passed each application.

  Referring again to FIG. 4, the indirect switch processing unit 44 reads the switch 66 storing the ON state stored in the trace collection condition storage unit 42 every time a request message reception count is obtained, and receives an instruction from the management console 38. Based on the stored indirect switch information, for example, if it is “1% switch”, the trace collection condition code holding unit is set to switch on one time out of 100 request message reception counts and switch off for the remaining 99 times. 46 is written as a switch 66-1.

  FIG. 5B is an explanatory diagram of intermittent switch information set in the indirect switch processing unit 44 by the management console 38. In the intermittent switch information 72, an intermittent switch is set as setting information, and the value of n is designated as “n% switch” as setting contents. The “n% switch” in the intermittent switch information 72 can arbitrarily set the value of n in the range of 0 to 100%.

  For example, “1% switch” or “0.1% switch” is set as the “n% switch”. In the case of “1% switch”, the trace collection condition that is switched on at a rate of once per 100 times of reception of the request message is embedded in the relay code and transferred to the next application.

  In the case of “0.1% switch”, the trace collection condition in which the switch is turned on at a rate of once per 1000 receptions of the request message is embedded in the relay code and transferred to the next application. By setting the intermittent switch information 72 as described above, it is possible to minimize the overhead applied to the distributed application system by reducing the ratio of the trace collection to the request message, and to always collect the trace even during the normal operation. realizable.

  Since “n% switch” in the intermittent switch information 72 can arbitrarily set the value of n in the range of 0 to 100%, no trace is collected when “0% switch” is set. Become operational. When “100% switch” is set, trace information is collected for all request messages, which corresponds to the function of the indirect switch processing unit 44 being eliminated.

The following two methods are used to collect traces using the “n% switch”.
(1) Collecting traces at regular intervals (2) Collecting random traces For example, taking “0.5% switch” as an example, trace collection is performed with the trace switch turned on for 5 times out of 1000 requests. In collecting traces at equal intervals, traces are collected at the 200th, 400th, 600th, 800th and 1000th times.

  On the other hand, in random trace collection, trace collection is performed, for example, at the 142nd, 262th, 438th, 626th, and 854th times using a random number with respect to the request message reception count of 1000. This random trace collection is more complicated than the regular interval trace collection, but has the advantage of reducing the periodic dependence of trace information.

  Furthermore, in the distributed application system, the application is executed sequentially with different routes depending on the contents of the request message. If the application route is different, the intermittent switch information 72 is set for each route. . The intermittent switch information setting for each route is determined by the type of request message, so the type of request message is determined, and the trace collection condition that switches on at the ratio of the corresponding intermittent switch information is embedded in the relay code. That's fine.

  As a result, even in a distributed application system where various request messages with different routes are mixed, the ratio of collecting traces for each route is set by indirect switch information, and the overhead for each route is minimized. It is possible to always collect traces during normal operation.

  Referring to FIG. 4 again, the relay ID generation unit 45 adds the identifier of the application to be transferred next to the unique relay ID 70 to the system stored in the trace collection condition storage unit 42 and sends it to the trace collection condition code holding unit 46. Write as relay ID 70-1. The function ID 68 stored in the trace collection condition storage unit 42 is written in the trace collection condition code holding unit 46 as it is.

  The trace collection condition code embedding unit 48 inputs a relay code to be added to the request message when the application processing unit 26 that has received the request message from the client executes the process and delivers it to the next application. The trace collection condition code configured by the switch 66-1, the function ID 68, and the relay ID 70-1 held in the code holding unit 46 is embedded in the relay code and output.

  Referring to FIG. 3 again, in the application server 14 and the database server 16 following the web server 12 at the entrance, trace collection units 30-2 and 30-3 and trace files 32-2 and 32-3 are provided. For trace collection, the trace condition embedding unit 28 of the web server 12 on the entrance side takes out the trace collection condition embedded in the relay code, and trace collection is executed based on this. Further, when the trace collection unit 30-1 delivers the data to the next application, the identifier of the next application is added to the relay ID and sent.

  FIG. 7 is an explanatory diagram of trace processing when a request message is received from the client 24 in the distributed application system 10 of FIG.

  In FIG. 7, first, the web server 12 is provided with a component 52 as an application, the application server 14 is provided with components 54 and 56 as applications, and the database server 16 is provided with components 58 and 60 as applications. ing.

  In the distributed application system 10, when a request message is received from the client 24, transactions are sequentially executed in each component 52, 54, 56, 58, 60, and a relay code is relayed when the request message is delivered to the next application. As shown by the codes 62-1 to 62-10, they are added and delivered.

  In response to such a request message from the client 24, the components 52, 54, 56, 58, 60 are routed when the database 18 is accessed and responded via the web server 12, the application server 14, and the database server 16. The passing positions in these are trace collection points P1 to P10.

  In the process of connecting a plurality of applications, in the present invention, the component 52 that is the application of the web server 12 at the entrance that receives a request message from the client 24 is traced by an instruction from the management console 38 of the management server 20. If the slace switch 50 is turned on for collection, the instruction content of the trace switch 50 is embedded in the relay code 62-1 that is passed to the next component 54 in addition to the request message. , 56, 58, and 60 can be instructed by a single switch operation.

  The function of the trace switch 50 is specifically stored as the switch 66 in the trace collection condition storage unit 42 of FIG. Further, the specific component in the route of the trace collection points P1 to P10 is determined by determining the trace range so as to collect the trace of the specific component by the function ID 68 stored in the trace collection condition storage unit 42 of FIG. Trace collection can also be performed only in.

  The ratio of embedding the switch-on trace collection condition 64 in the relay code 62-1 in the web server 12 is set by setting the indirect switch information 72 in FIG. 5B from the management console 38 in the indirect switch processing unit 44 in FIG. For example, if “1% switch” is set, the trace collection condition 64 that is switched on is embedded in the relay code 62-1 only once out of 100 received request messages. In this case, the overhead of the distributed application system can be minimized and the trace collection can be realized during normal operation by limiting the trace collection ratio to 1% with respect to the reception of the request message.

  FIG. 8 is a flowchart of the trace condition embedding process according to the present invention, which is executed by the trace condition embedding unit 28 provided in the web server 12 of FIG. In the trace condition embedding process of FIG. 8, in response to an instruction from the management console 38 in step S1, the trace collection switch is turned on. To turn on the trace collection switch, specifically, the switch on is written to the switch 66 in the trace collection condition storage unit 44 of the trace condition embedding unit 28 in FIG.

  Next, in step S2, the trace collection ratio based on the intermittent switch condition and the trace range condition based on the function ID are stored as the trace collection condition. Specifically, the intermittent switch condition and the function ID are also written in the indirect switch information to the indirect switch processing unit 44 of the trace condition embedding unit 28 in FIG. 4 and the trace range condition for the function ID 66 in the trace collection condition storage unit 42. Realized by writing.

  Subsequently, in step S3, reception of a request message from the client is awaited. When the request message is received, the process proceeds to step S4, and it is checked whether or not the trace is collected based on the intermittent switch condition stored in step S2. For example, in the case of “0.1% switch”, it is determined that the trace collection condition is satisfied when the count of the received request message is 100th, and the process proceeds to step S5.

  In step S5, the switch set in the ON state is embedded in the relay cord. In step S6, the function ID and the relay ID are embedded in the relay code. In step S8, a relay code is added to the request message, and in step S9, the message is delivered to the next application.

  On the other hand, if the trace is not collected based on the intermittent switch condition in step S4, the relay code is embedded in the relay code in step S7, and then the relay code is added to the request message in step S8. Pass to the next application.

  FIG. 9 is a flowchart of the trace collection process according to the present invention. In the trace collection process of FIG. 8, whether or not a request message has been received is checked in step S1, and when the request message is received, the content of the relay code added to the request message is confirmed and embedded in the relay code. If the switch in the current trace collection condition is on, the process proceeds to step S3 and a trace is collected.

  Subsequently, in step S4, the identifier of the next application is added to the relay ID, and in step S6, it is transferred to the next application. On the other hand, if the switch is in the OFF state in the relay code trace collection conditions, the process proceeds to step S5, and the trace is not collected, and is passed to the next application in step S6.

  The trace collection unit 30-1 provided in the web server 12 of FIG. 3 does not perform the processing of the flowchart of FIG. 9 when receiving a request message from the client 24, and stores the trace collection condition of FIG. If the switch 66 of the unit 42 is switched on, a trace is collected. If the switch 66 is switched off, a process of not collecting a trace is performed. Further, when the trace collection unit 30-1 receives the relay code added to the response message from the application server 14, the trace collection unit 30-1 performs the trace collection processing according to the flowchart of FIG.

  FIG. 10 is a flowchart of the trace information analysis display process according to the present invention, which is executed by the analysis display unit 40 of the management server 20 shown in FIG. In the analysis display process of FIG. 10, in accordance with an instruction from an operator or the like, trace data is collected from the trace file stored on the server side in step S1.

  Next, in step S2, the collected data is arranged so that the collected data are arranged in the order of time of the trace collection points according to the route by the relay ID, for example. Subsequently, in step S3, a route analysis is performed on the collected data for each request message, and the time and time ratio required for processing of each application on the route are obtained, and the analysis result is displayed in step S4.

  FIG. 11 is an explanatory diagram showing an example of trace information arranged for analysis display. The trace information 84 in FIG. 11 is organized by items of time information 86, IP address 88, process ID 90, relay ID 92, collection point 94, component name 96, in-component identification information 98, and method name 100.

  FIG. 12 is an explanatory diagram of a display example of the analysis result obtained from the organization information of FIG. In the display example of the analysis result of FIG. 12, the time axis is taken in the vertical axis direction with respect to the passing points on the route composed of the components 52, 54, 56, 58, 60 and the database 18, and each execution is executed. Times 102, 104, 106,... 122 are shown according to the length of the processing time, and for each of the execution times 102 to 122, 10 ms, 10 ms, 12 ms,. It is displayed. Furthermore, the total execution time in the components 52, 54, 56, 58, 60 and the database 18 is displayed at the lower end.

  By observing the display of the analysis result of the trace information in FIG. 12, it is possible to determine at a glance which component is the bottleneck on the route of processing of the request message by the distributed application system.

  FIG. 13 is an explanatory diagram of other trace information arranged for analysis display. Similar to the case of FIG. 11, time information 86, IP address 88, process ID 90, relay ID 92, sampling point 94, component name 96, component The items of the internal identification information 98 and the method name 100 are sorted and arranged in time order.

  FIG. 14 is an explanatory diagram of a display example analyzed from the organization information of FIG. In the analysis display about the route of the request message in FIG. 14, in contrast to FIG. 12, in addition to the components 52, 54, 56, 58, 60 and the database 18, the component 126 is newly added as a passing point. I understand that. Also in this case, with respect to the passing points of the components 52, 54, 56, 58, 60, the database 18 and the component 126 shown on the horizontal axis, the execution times 128 to 156 shown on the vertical axis and the time values for the respective execution times, Furthermore, by displaying the total value of the execution times at the lower end, it is possible to easily determine at which position of the route for processing the request message a bottleneck has occurred.

  FIG. 15 is another explanatory view of the display example of the analysis result obtained from the organization information of FIG. 13. In this display example, the execution time of the application is set to the response time of the component 52, that is, the total processing time. It is characterized by the time ratio (percentage). Even if the numerical value of the execution time is shown as a percentage of the time ratio, it is possible to easily determine at which component position on the route the bottleneck is the same as in the display example of FIG. In the actual bottleneck determination, the numerical display of the execution time in FIG. 14 can make a more accurate determination than the percentage display in FIG.

  The present invention provides a program for trace processing in the web server 12, application server 14 and database server 16 of FIG. These servers are realized by computer hardware resources.

  As for computer hardware resources, RAM, a hard disk, a communication board, etc. are connected to a CPU bus, and a program necessary for the trace processing of the present invention is loaded on the hard disk. A necessary program is read out, developed on the RAM, and executed by the CPU.

  As a program for the trace processing in the web server 12 located on the entry side of the request message to the client 24, a trace condition embedding program according to the flowchart of the trace condition embedding process of FIG. 8 and a trace collecting process of FIG. The program according to the flowchart is stored.

  On the other hand, for the application server 14 and the database server 16 in the second and subsequent stages, only the trace collection program according to the flowchart of the trace collection process in FIG. 9 is stored and executed. Furthermore, in the management server 20, the analysis display program shown in the flowchart of the analysis display process in FIG. 10 is loaded and executed.

  The present invention includes appropriate modifications that do not impair the objects and advantages thereof, and is not limited by the numerical values shown in the above embodiments.

    Here, the features of the present invention are enumerated as follows.

(Appendix)
(Appendix 1)
Trace of a distributed application system that has multiple applications that operate on multiple servers arranged in a hierarchy and responds by executing processing while sequentially receiving relay codes between applications when receiving a request message from a client In the processing method,
A trace condition embedding step of embedding a preset trace collection condition including switch information instructing whether or not trace collection is received in the relay code when the request message is received;
A trace collection step that is provided in each of the applications and determines and processes the presence or absence of trace collection by switch information of a trace collection condition embedded in the relay code when the request message is received;
An analysis display step for collecting and analyzing and displaying traces collected by each application;
A trace processing method comprising: (1)

(Appendix 2)
In the trace processing method according to attachment 1, in the trace condition embedding step, as the trace collection condition, a function ID that indicates a range in which the trace is collected for each application, and a relay ID that records the passed application The trace processing method characterized by including these.

(Appendix 3)
In the trace processing method described in Appendix 1,
The trace condition embedding step generates a unique relay ID when the switch information indicates the presence of trace collection.
In the trace collection step, an identifier of an application to be called next is connected to the back of the relay ID.

(Appendix 4)
In the trace processing method according to attachment 1, in the trace condition embedding step, a trace is further collected as the trace collection condition with respect to the number of times the request message is received in an instruction state with trace collection by the switch information. A trace processing method comprising: intermittent switch information indicating a ratio, and embedding a trace collection condition in the relay code according to the ratio of the intermittent switch information. (2)

(Appendix 5)
The trace processing method according to appendix 4, wherein the intermittent switch information identifies a route of a request message, and indicates a ratio of collecting a trace with respect to the number of reception times of the request message for each identified route. Trace processing method to be performed.

(Appendix 6)
The trace processing method according to appendix 4 or 5, wherein the intermittent switch indicates a ratio of collecting a trace with respect to the number of times the request message is received in a range of 0% to 100%.

(Appendix 7)
In the trace processing method according to attachment 1, in the analysis display step, the trace information having the same relay ID is extracted from the trace information collected by each application in the trace collection step, and is sorted in time order. A trace processing method characterized by generating and displaying a relationship between a route and each application based on the result.

(Appendix 8)
The trace processing method according to appendix 7, wherein the analysis display step displays a processing time from receipt of the request message to the next delivery corresponding to each application serving as a passing point. Processing method.

(Appendix 9)
In the trace processing method according to attachment 7, in the analysis display step, each application receives a request message and then delivers it to the total processing time from when the application at the entrance receives the request message until it responds. The trace processing method is characterized by displaying the ratio of the processing time up to each application corresponding to the passing point.

(Appendix 10)
In a computer constituting a server having a first-tier application that first processes a client request message,
A trace condition embedding step of embedding a preset trace collection condition including switch information instructing whether or not to collect traces in the relay code when the request message is received from the client and passing it to the next application;
Collecting traces embedded in the relay code of the request message when the request message is received from another application, and processing is performed by determining the presence or absence of trace collection based on the switch information of the trace collection condition embedded in the relay code. Trace collection step for processing by determining presence / absence of trace collection based on condition switch information,
A program characterized by having executed. (3)

(Appendix 11)
In a computer that constitutes a server having an application of the second hierarchy or later that processes a client request message for the second time or later,
When the request message is received from another application, a trace collection step is executed to determine whether or not to collect a trace based on the switch information of the trace collection condition embedded in the relay code of the request message and to process it. Program. (4)

(Appendix 12)
The program according to appendix 10 or 11, wherein the trace condition embedding step further includes a function ID for instructing a range for collecting a trace for each application unit and a relay ID for recording the passed application. A program characterized by

(Appendix 13)
In the program described in Appendix 12,
The trace condition embedding step generates a unique relay ID when the switch information indicates the presence of trace collection.
In the trace collection step, an identifier of an application to be called next is connected to the back of the relay ID.

(Appendix 14)
In the program according to attachment 10 or 11, in the trace condition embedding step, a trace is further collected as the trace collection condition with respect to the number of times the request message is received in an instruction state with trace collection by the switch information. A program comprising intermittent switch information indicating a ratio, and embedding a trace collection condition in the relay code according to the ratio of the intermittent switch information.

(Appendix 15)
The program according to claim 14, wherein the intermittent switch information identifies a route of a request message, and indicates a ratio of collecting a trace with respect to the number of reception times of the request message for each identified route. .

(Appendix 16)
The program according to claim 10 or 11, wherein the intermittent switch indicates a ratio of collecting a trace with respect to the number of reception times of the request message within a range of 0% to 100%.

(Appendix 17)
It has a plurality of application processing units that operate on a plurality of hierarchically arranged servers, and when it receives a request message from a client, it responds by executing the processing while sequentially passing relay codes between the application processing units In the trace processing device for distributed applications,
When receiving the request message, a trace condition embedding unit that embeds a preset trace collection condition including switch information indicating whether or not to collect a trace and passes it to the next application,
A trace collection unit that is provided in each of the applications, and determines and processes the presence or absence of trace collection based on the switch information of the trace collection condition embedded in the relay code when the request message is received;
A trace processing apparatus comprising: an analysis display unit that collects and analyzes and displays a trace collected by each application processing unit. (5)

(Appendix 18)
The trace processing device according to appendix 17, wherein the trace condition embedding unit further includes a function ID for instructing a range for collecting a trace for each application, and a relay ID for recording a passed application as the trace collection condition. A trace processing apparatus comprising:

(Appendix 19)
In the trace processing apparatus according to appendix 18,
The trace condition embedding unit generates a unique relay ID when the switch information indicates that there is a trace collection,
The trace processing unit, wherein the trace collection unit connects an identifier of an application to be called next to the relay ID.

(Appendix 20)
The trace processing device according to appendix 17, wherein the trace condition embedding unit further collects a trace for the number of times the request message has been received in the instruction state with trace collection by the switch information as the trace collection condition. A trace processing apparatus comprising intermittent switch information indicating a ratio, and embedding a trace collection condition in the relay code according to the ratio of the intermittent switch information.

(Appendix 21)
The trace processing device according to attachment 20, wherein the intermittent switch information identifies a route of a request message, and indicates a ratio of collecting a trace with respect to the number of reception times of the request message for each identified route. Trace processing device to perform.

(Appendix 22)
The trace processing device according to appendix 20 or 21, wherein the intermittent switch indicates a ratio of collecting a trace with respect to the number of times of reception of the request message within a range of 0% to 100%.

(Appendix 23)
In the trace processing device according to attachment 17, the analysis display unit extracts the trace information of the route according to the relay ID from the trace information collected by each application by the trace collection unit, and sorts the information in time order, A trace processing apparatus that generates and displays a relationship between a route, each application, and processing based on a sorting result.

(Appendix 24)
The trace processing apparatus according to appendix 17, wherein the analysis display unit displays a processing time from receipt of a request message to the next delivery corresponding to each application serving as a passing point. Processing equipment.

10: Distributed application system 12, 12-1 to 12-4: Web server 14, 14-1 to 14-4: Application server 16, 16-1, 16-2: Database server 18, 18-1, 18- 2: Database 20: Management server 22: Internet 24: Clients 26, 34, 36: Application processing unit 28: Trace condition embedding unit 30-1 to 30-3: Trace collection unit 32-1 to 32-3: Trace file 38: Management console 40: Analysis display unit 42: Trace collection condition storage unit 44: Indirect switch processing unit 45: Relay ID generation unit 46: Trace collection condition code holding unit 48: Trace collection condition code embedding unit 50: Trace switch 52 , 54, 56, 58, 60, 126: components 62-1 to 62-10: Rekodo 64: trace collection conditions 66: switch 68: function ID
70: Relay ID
72: Intermittent switch information 74, 75, 76, 78, 80, 82: Application field 590: Subsystem second destination field 592: Second destination designation flag

Claims (6)

A trace collection method for an application system in which a computer processes a message request by linking multiple applications.
When the message request is received, information specifying the presence / absence of trace collection in the trace collection condition storage unit and the identifier of the message request are added to the message request,
When the electronic message request to which the information and the identifier are added is received, the computer executes to record the trace information including the identifier at the time of executing the application based on the information. .
The trace processing method according to claim 1,
The trace collection condition storage unit includes a process of adding a relay ID for recording a function ID that instructs a range of collecting traces for each application.
The trace processing method according to claim 1,
The trace collection condition storage unit includes intermittent switch information instructing a ratio of collecting a trace based on the number of receptions of the telegram request in an instruction state with trace collection by switch information, and the telegram according to the ratio of the intermittent switch information A trace processing method comprising adding a trace collection condition to a request.
In the trace processing method of Claim 3,
The intermittent switch information is information for identifying a route of a message request, and indicates a ratio of collecting trace information based on the number of times of reception of the message request for each route based on the information for identifying the route. Trace processing method to be performed.
A computer is a program that processes a telegram request by linking multiple applications together,
A trace condition adding step of adding information specifying the presence / absence of trace collection in the trace collection condition storage unit and the identifier of the message request to the message request when receiving the message request;
A trace information recording step of recording trace information including the identifier at the time of application execution based on the information when the electronic message request to which the information and the identifier are added is received;
That causes the computer to execute the program.
A trace information processing apparatus of an application system in which a computer processes a message request by linking a plurality of applications,
A trace condition storage unit for adding information specifying the presence or absence of trace collection and an identifier of the message request to the message request when receiving the message request;
A trace information processing apparatus comprising: a trace information storage device that records trace information including the identifier when an application is executed based on the information when the electronic message request to which the information and the identifier are added is received apparatus.

Images