JP4731960B2 - Pooling method, system and program - Google Patents

Description

  The present invention relates to a technique effective when applied to a pooling technique for efficiently processing data by pooling and reusing a generated object or program.

  In recent years, there has been an increasing demand for a system that mainly provides Web-based services via the Internet or an intranet. When constructing such a system, a technique of developing a user application that operates on an application server using middleware called an application server is becoming widespread. The application server provides a security management function, a transaction management function, a resource access function, a component management function, and the like. Therefore, a large-scale and highly reusable system can be constructed relatively easily by using the application server.

  The application server sometimes performs pooling of query expression objects and query result objects in order to realize efficient access to the database. Object pooling is a technique for reducing the cost of object generation by holding an object with a high generation cost in a memory without discarding it after use and reusing it. Pooling control is realized by profiling the use of query expressions and query results by user applications. Profiling is a technology that stores, for example, the frequency of use of query expressions and query results, and finds frequently used objects. Based on the results of profiling, control is performed such as preferentially pooling frequently used objects. As an object pooling method, a method as shown in Patent Document 1 has been proposed.

  By the way, a system may be constructed by combining a plurality of application servers distributed horizontally or vertically. The reason why the system has such a configuration is that the system can be expanded in a scalable manner in accordance with changes in demands on the system. Scale-out refers to expanding a system by adding a new application server to a system that has already been constructed and operated.

JP 2004-78651 A

  When scaling out (adding) an application server (which may be a computer, logical server, or virtual machine), the application server to be newly added performs user application processing with the query expression object and query result object not being pooled. Will start. Therefore, there is a problem that the response to the use of these objects by the user application decreases after the scale-out until the query expression object or query result object in the database is pooled.

  Even when the application server is restarted, the processing of the user application is started in a state where the query expression object and the query result object are not pooled. Therefore, as in the case of scaling out, there is a problem that the response in the use of the query expression object and the query result object by the user application is reduced until the object is pooled.

  An object of the present invention is to solve the above problems and provide a technique capable of realizing a good response immediately after scale-out or immediately after restart in the use of an object or program in an added or restarted information processing apparatus. There is.

  The present invention scales profiling information obtained by monitoring use of a query expression object or query result object (may be a program) in a database in a data processing system composed of a single or a plurality of application servers. It is reused when it is out or when the application server is restarted.

  In the data processing system of the present invention, the application server profiles use of the query expression and query result of the database by the user program, and controls the pooling of the query expression object and query result object using the profiling information. At the same time, the application server holds profiling information in memory or persistent storage means.

  When the application server is scaled out, the profiling information acquired in the already operating application server and held in the memory or the permanent storage means is distributed to the newly added application server. The newly added application server uses the distributed profiling information to restore the object pool of the query expression object and the query result object in the database. In this way, by reusing the profiling information acquired on the application server that is already in operation on the newly added application server, it is possible to eliminate the waste of acquiring profiling information again. In the use of the query expression object and the query result object by the user application, a satisfactory response can be realized.

  Further, when the application server is restarted, the object pool of the query expression object and the query result object in the database is restored using the profiling information acquired before the restart and stored in the permanent storage means. In this way, by reusing the profiling information acquired before the restart after the restart, it is possible to eliminate the waste of acquiring the profiling information again. In using the query result object, a good response can be realized.

  According to the present invention, in the use of an object or program in an information processing apparatus that has been added or restarted, a good response can be realized immediately after scale-out or immediately after restarting.

A pooling system according to the first embodiment that reduces the number of times an object is generated by pooling and reusing the generated object will be described below.
In the data processing system of this embodiment, when scaling out an application server (a computer or an information processing apparatus, or a program or object that performs such processing), the profiling information acquired by the operating application server Are reused in a newly added application server, and a process of restoring a query expression object pool and a query result object pool of the database is performed. In the present embodiment, an object is described, but it may be applied to a program.

  FIG. 1 is a diagram showing the overall structure of the data processing system of this embodiment. As shown in FIG. 1, in the present embodiment, an operating application server 0001 from which a user application is executed and profiling information is acquired, application servers 0004 and 0005 added at the time of scale-out, and a database from which the user application makes an inquiry 0003 are connected by a network 0002, and the profiling information 0006 and 0007 can be transmitted from the operating application server 0001 to the application servers 0004 and 0005 to be added, respectively.

  FIG. 2 is a diagram illustrating an internal configuration example of the application server 0001 in operation according to the present embodiment. As shown in FIG. 2, the operating application server 0001 includes a memory 0008, a CPU 0009, an input device 0010, an output device 0011, a communication device 0012, a magnetic disk device 0013, and a CD-ROM device 0014.

  An application container 0016 for executing a user application in the operating application server 0001 is stored in a magnetic disk device 0013 by a CD-ROM device 0014 from a recording medium such as a CD-ROM, and then loaded into the memory 0008 for execution. Shall be. The application container 0016 may be used by being stored in a magnetic disk device 0013 from a recording medium such as a CD-ROM via a network. The recording medium for recording the application container 0016 may be a recording medium other than the CD-ROM.

  In the application server 0001 in operation, the user application 0015 is stored in the magnetic disk device 0013 by the CD-ROM device 0014 from a recording medium such as a CD-ROM and then loaded into the memory 0008 by the application container 0016 and executed. And The user application 0015 may be stored in the magnetic disk device 0013 from a recording medium such as a CD-ROM via a network, or the application container 0016 may store the user application from a recording medium such as a CD-ROM device in a memory via the network. You can load it.

  In the application container 0016, there is an object management unit 0040 that manages the profiling information 0017 and the object pool.

  The object management unit 0040 monitors the use of the query expression object and the query result object in the database by the user application 0015, and holds the profiling information 0017 obtained by the monitoring in the memory 0008.

  In addition, the object management unit 0040 uses the profiling information 0017 to control the pooling of objects to the query expression object pool 0018 and the query result object pool 0019.

  Further, the object management unit 0040 periodically records the content of the profiling information 0017 in the magnetic disk device 0013. The operation of periodically recording here may be performed at regular time intervals, or may be performed in response to profiling information, the number of reference of the object pool, or the number of updates.

  The object management unit 0040 has a function of transmitting / receiving profiling information to / from the added application servers 0004 and 0005.

  FIG. 3 is a diagram showing an example of the internal configuration of the application servers 0004 and 0005 added at the scale-out of this embodiment. As shown in FIG. 3, the added application servers 0004 and 0005 do not have the profiling information 0042, the query object pool 0043, and the query result object pool 0044. If the database is accessed in this state, the response to the use of the query expression object or query result object by the user application is reduced until the object is pooled.

  Therefore, prior to starting access to the database, the object management unit 0041 obtains the profiling information 0017 from the operating application server 0001 via the communication device 0024, and stores the profiling information 0042 in the memory 0020. At the same time, a query expression object pool 0043 and a query result object pool 0044 are created using the profiling information 0042.

  Thereafter, by starting database access from the added application servers 0004 and 0005, it is possible to realize access with a good response using the pooled object from the beginning of the access.

  FIG. 4 is a diagram showing an example of the profiling information 0017 of this embodiment. The profiling information 0017 includes a table that associates the query expression character string 0030 with the query expression object use frequency 0031 and a table that associates the query result character string 0032 with the query result object use frequency 0033.

  The object management unit 0040 manages the object pool using the query object usage frequency 0031 and the query result object usage frequency 0033. Specifically, when there is a request by a new query expression and it is necessary to generate a query expression object corresponding to the request, but there is no space in the object pool, refer to the profiling information 0017 and refer to the least frequently used query. After the expression object is discarded and removed from the object pool, a query expression object corresponding to a new query expression is generated and stored in the object pool.

  FIG. 5 is a flowchart showing the processing procedure of the object management unit 0041 of this embodiment. FIG. 5 shows the processing contents performed by the object management unit 0041 before the database access is started in the application servers 0004 and 0005 to be added.

  In step 0050, the object management unit 0041 obtains the profiling information 0017 from the operating application server 0001 via the communication device 0024. In step 0051, the obtained profiling information 0017 is arranged on the memory 0020 as profiling information 0042. In step 0052, a query type object pool 0043 and a query result object pool 0044 are created in the memory 0020 from the arranged profiling information 0042.

  The query expression object pool 0043 and the query result object pool 0044 are created by inputting the query expression character string 0030 and the query result character string 0032 recorded in the profiling information 0042 and inputting corresponding objects in descending order of use frequency. It is to create and store in the object pool. Thus, the application servers 0004 and 0005 to be added can have the same object pool as that of the application server 0001 in operation.

  Here, when profiling information 0017 is obtained from a plurality of operating application servers 0001, such as when performing scale-out in a system in which a plurality of operating application servers 0001 operate, it is executed by the added application servers 0004 and 0005. Profiling information corresponding to the user application 0027 to be acquired may be obtained.

  That is, when the profiling information 0017 is created by each active application server 0001, identification information for identifying the user application 0015 executed by the active application server 0001 is added to the profiling information 0017, When obtaining the profiling information 0017 from the middle application server 0001, the identification information of the user application 0015 added to the profiling information 0017 is compared with the identification information of the user application 0027 executed by the added application servers 0004 and 0005. Then, the profiling information 0017 with the identification information that matches the identification information of the user application 0027 is obtained. As a result, it is possible to realize access with a good response corresponding to the user application 0027 executed by the added application servers 0004 and 0005.

  Further, when the profiling information 0017 is obtained from the application server 0001 in operation, the profiling information corresponding to the specifications such as the sizes of the object pools of the application servers 0004 and 0005 to be added may be obtained.

  That is, when the profiling information 0017 is created by the application server 0001 in operation, the profiling information of the object discarded when the object pool is not empty is separately held, and the object pools of the application servers 0004 and 0005 to be added are stored. If the size is larger than the size of the object pool of the operating application server 0001, the profiling information 0017 and the separately stored profiling information are obtained from the operating application server 0001. As a result, the application servers 0004 and 0005 to be added have a good response corresponding to the size of the object pool even when the size of the object pool is larger than the size of the object pool of the application server 0001 in operation. Can be realized.

  The data processing system according to the second embodiment will be described below in which, when the application server is restarted, the profiling information before restart is read from the permanent storage unit, and the database query expression and the query result object pool are restored. .

  As described in the first embodiment, the object management unit 0040 periodically records the content of the profiling information 0017 in the magnetic disk device 0013. The operation of periodically recording here may be performed at regular time intervals, or may be performed in response to profiling information, the number of reference of the object pool, or the number of updates.

  When the operating application server 0001 terminates abnormally or when it is necessary to restart for some reason, the profiling information 0017 on the memory, the query object pool 0018, and the query result object pool 0019 are lost.

  Before restarting access to the database after restarting the application server 0001 in operation, the profiling information 0017 recorded in the magnetic disk device 0013 is read, and the query-type object pool 0018 and the query result are read from the profiling information. By recovering the object pool 0019, it is possible to realize access with a good response using the pooled object from the beginning of database access after restart.

  FIG. 6 is a flowchart showing the processing procedure of the object management unit 0040 of this embodiment. FIG. 6 shows the contents of processing performed by the object management unit 0040 between the restart of the application server 0001 in operation and the start of database access.

  In step 0061, the object management unit 0040 obtains the recorded profiling information from the magnetic disk device 0013. In step 0061, the obtained profiling information is arranged on the memory 0008 as profiling information 0017. In step 0062, a query expression object pool 0018 and a query result object pool 0019 are created on the memory 0008 from the arranged profiling information 0017.

  That is, the query expression character string 0030 and the query result character string 0032 recorded in the profiling information 0017 are input, and corresponding objects are generated in advance in descending order of use and stored in the object pool. Thereby, even when the application server 0001 in operation is restarted, it is possible to realize access with a good response using the pooled object from the beginning of the database access after the restart.

  Here, when a plurality of application programs are being executed on the application server 0001 in operation and some of the application programs that have ended abnormally are not executed, such as changing the application to be executed after the restart, When the profiling information 0017 recorded on the magnetic disk device 0013 is acquired at the time of restart, the profiling information corresponding to the user application 0015 executed by the operating application server 0001 to be restarted may be acquired.

  That is, when the profiling information 0017 is recorded on the magnetic disk device 0013 by the operating application server 0001, identification information for identifying the user application 0015 executed by the operating application server 0001 is added to the profiling information 0017. When the profiling information 0017 is obtained from the magnetic disk device 0013 at the time of restart, the identification information of the user application 0015 added to the profiling information 0017 is compared with the identification information of the user application executed after the restart. Then, the profiling information 0017 to which the identification information matching the identification information of the user application executed after the restart is added is obtained from the magnetic disk device 0013. As a result, when the application server 0001 in operation is restarted, access with a good response corresponding to the user application 0015 executed after the restart can be realized.

  As described above, according to the present embodiment, in the application server to be added or the application server after restart, the pool of objects or programs from the profiling information obtained from other application servers or the profiling information saved by the own application By generating, it becomes possible to realize database access with a good response using the pooled object immediately after addition and restart.

  As described above, according to the pooling system of the present embodiment, the profiling information obtained by the information processing apparatus in operation is reused at the time of scale-out or restart. It is possible to realize a good response immediately after the scale-out or restart after using the object or program.

It is a figure which shows the whole structure of the data processing system of this embodiment. It is a figure which shows the internal structural example of the application server 0001 in operation of this embodiment. It is a figure which shows the internal structural example of the application servers 0004 and 0005 added at the time of the scale-out of this embodiment. It is a figure which shows an example of the profiling information 0017 of this embodiment. It is a flowchart which shows the process sequence of the object management part 0041 of this embodiment. It is a flowchart which shows the process sequence of the object management part 0040 of this embodiment.

Explanation of symbols

0001 ... Application server in operation, 0002 ... Network, 0003 ... Database, 0004 and 0005 ... Additional application server, 0006 and 0007 ... Profiling information, 0008 ... Memory, 0009 ... CPU, 0010 ... Input device, 0011 ... Output device, 0012 ... Communication device, 0013 ... Magnetic disk device, 0014 ... CD-ROM device, 0015 ... User application, 0016 ... Application container, 0017 ... Profiling information, 0018 ... Inquiry object pool, 0019 ... Inquiry result object pool, 0040 ... Object Management unit, 0020 ... memory, 0021 ... CPU, 0022 ... input device, 0023 ... output device, 0024 ... communication device, 0025 ... magnetic Disk device, 0026 ... CD-ROM device, 0027 ... user application, 0028 ... application container, 0041 ... object management unit, 0042 ... profiling information, 0043 ... query object pool, 0044 ... query result object pool, 0030 ... query character Column, 0031 ... frequency of use of query expression object, 0032 ... query result character string, 0033 ... frequency of use of query result object.

Claims (10)

In a pooling method that reduces the number of times an object or program is generated by pooling and reusing the generated object or program,
The added information processing apparatus obtains the frequency of use of the object or program by the communication apparatus from the information processing apparatus in operation;
A step of added information processing apparatus, to place the frequency of use of objects or program the obtained, the added information processing devices that are going,
Added information processing apparatus, features and to pulp-ring method further comprising the steps of: creating a pool of objects or programs on the storage device in response to the frequency of use of the object or program was arranged. When the added information processing device obtains the usage frequency of the object or program from the information processing device in operation, the usage frequency of the object or program corresponding to the object or program executed by the added information processing device is obtained. The pooling method according to claim 1, wherein: When the added information processing device obtains the usage frequency of the object or program from the information processing device in operation, the information processing device must obtain the usage frequency of the object or program corresponding to the pool size of the added information processing device. The pooling method according to claim 1, wherein the pooling method is characterized. In a pooling method that reduces the number of times an object or program is generated by pooling and reusing the generated object or program,
The restarted information processing apparatus obtains the usage frequency of the object or program recorded in the nonvolatile storage device during operation from the nonvolatile storage device upon restarting;
A step of the restarted information processing apparatus, to place the frequency of use of objects or program the obtained, the restarted information processing devices that are going,
A pooling method , wherein the restarted information processing apparatus includes a step of creating a pool of objects or programs on a storage device in accordance with the frequency of use of the arranged objects or programs . Corresponding to the object or program executed by the restarted information processing device when the restarted information processing device obtains the usage frequency of the object or program recorded in the nonvolatile storage device during the operation. 5. The pooling method according to claim 4, wherein the usage frequency of the object or program is obtained from the nonvolatile storage device. The restarted information processing apparatus, a frequently used objects or program among objects or program shows the query result string and the query expression string that is recorded in advance generated and the frequency of use of the object or program 6. The pooling method according to claim 1, wherein a pool of the object or program is created on a storage device by storing in the pool. In pooling system to reduce the number of times to generate an object or program by the generated object or program to pooling and re-use,
Obtain the usage frequency of the object or program from the information processing device in use by the communication device,
The frequency of use of objects or program the obtained was placed in the added information processing devices that are going,
A pooling system comprising: a management unit that creates a pool of objects or programs on a storage device in accordance with the frequency of use of the arranged objects or programs . In a pooling system that reduces the number of times an object or program is created by pooling and reusing the created object or program,
Obtain the usage frequency of the object or program recorded in the nonvolatile storage device during operation from the nonvolatile storage device when restarting,
Wherein the frequency of use of objects or program was obtained, it placed restarted information processing devices that are going, management unit to create a pool of objects or programs on the storage device in response to the frequency of use of the object or program arranged A pooling system comprising: In a program for causing a computer to execute a pooling method for reducing the number of times an object or program is generated by pooling and reusing the generated object or program,
Obtaining the usage frequency of the object or program from the information processing device in operation by the communication device;
Placing the frequency of use of objects or program the obtained, the added information processing devices that are going,
A program causing a computer to execute a step of creating a pool of objects or programs on a storage device in accordance with the frequency of use of the arranged objects or programs. In a program for causing a computer to execute a pooling method for reducing the number of times an object or program is generated by pooling and reusing the generated object or program,
Obtaining from the non-volatile storage device the frequency of use of the object or program recorded in the non-volatile storage device during operation;
Placing the frequency of use of objects or program the obtained, the restarted information processing devices that are going,
A program causing a computer to execute a step of creating a pool of objects or programs on a storage device in accordance with the frequency of use of the arranged objects or programs.

Images