JP5071518B2 - Database processing method, database processing system, and database management program - Google Patents

Description

  The present invention relates to a database processing method, a database processing system, and a database management program, and more particularly to a database processing method, a database processing system, and a database processing system that are effective when applied to a database management system (DBMS) having a system switching function. It relates to a database management program.

  In the world of Internet business, where the suspension of service is directly linked to the loss of a significant business opportunity, a robust system that continues to operate 24 hours a day, 365 days a year is required. In particular, localization of the impact at the time of failure and rapid system recovery are important. Conventionally, in a database (DB) system, a standby machine is prepared separately from an execution machine for service execution in order to quickly recover a system when a failure occurs, and the standby machine executes services when a failure occurs. A technique called “system switching” has been used.

  For example, Non-Patent Document 1 (Jim Gray and Andreas Reuter: "TRANSACTION PROCESSING: CONCEPTS AND TECHNIQUES", Morgan Kaufmann Publishers, 1993) discloses hot standby non-stop operation as a DB failure countermeasure by HA (High Availability) system configuration. ing.

  On the other hand, Non-Patent Document 2 ("Parallel Database Systems: The Future of High Performance Database Systems", COMMUNICATIONS OF THE ACM, Vol.35, No.6, 1992, P.85-P.98) An architecture is disclosed in which a load is distributed to a plurality of processors and executed in parallel. In the Shared Everything, Shared Disk (shared type) architecture described in the above prior art, all computers that execute DB processing can access all data. In the Shared Nothing (non-shared type) architecture, Only data stored on a disk connected to the computer can be accessed.

  The Shared Nothing (non-shared) architecture has fewer shared resources among the structural units that execute DB processing than the shared architecture, and is extremely superior in terms of scalability. Even in Shared Nothing (non-shared) architectures, system switching technology is often used to provide high availability.

  Non-Patent Document 3 (Aslam Nomani: “Implementing IBM DB2 Universal Database V8.1 Enterprise Server Edition with Microsoft Cluster Server”, International Business Machines Corporation, 2003) describes a database management system in which a plurality of DB storage areas are arranged on each machine. Discloses a system switching method using a DB storage area as a switching unit (load distribution type system switching method). For each DB storage area, an execution system and a standby system are defined. In normal times, the execution system accesses the DB storage area, and when a failure occurs, the standby system accesses the DB storage area to continue the service. In this method, a plurality of DB storage areas belonging to the same machine are distributed to different machines and standby systems are defined. When a failure occurs in one machine, a plurality of DB storage areas belonging to the machine are taken over by different machines. As a result, it is possible to distribute the processing execution load after system switching to a plurality of machines, and to reduce the throughput reduction of the entire system.

Jim Gray and Andreas Reuter: "TRANSACTION PROCESSING: CONCEPTS AND TECHNIQUES", Morgan Kaufmann Publishers, 1993 David Dewitt and Jim Gray: "Parallel Database Systems: The Future of High Performance Database Systems", COMMUNICATIONS OF THE ACM, Vol.35, No.6, 1992 Aslam Nomani: "Implementing IBM DB2 Universal Database V8.1 Enterprise Server Edition with Microsoft Cluster Server", International Business Machines Corporation, 2003

  However, in system switching, a standby machine must be prepared in addition to the executing machine, and the standby machine is idle during normal service execution. In addition, system switching can also be assigned to the standby machine in the form of “mutual standby”, but normal service execution can be assigned to the standby machine. It is often performed in advance ("warm standby", "hot standby"), and standby system resources (such as processes and memory) must be prepared.

  In systems that require standby-only resources that are not in normal operation as described above, resources cannot be used effectively during normal operations. From the perspective of reducing total cost of ownership (TCO) in system construction and operation It is a problem. Furthermore, even if a machine in a certain operating state fails and failover occurs to another operating machine due to system switchover, the processing load is concentrated on the switching destination machine (usually doubles the load when it is not offset) In other words, the processing performance of the entire system deteriorates.

  An object of the present invention is to suppress the standby-only resources that are normally inactive as described above. Another object is to improve processing performance degradation and load imbalance of the entire system after system switching when a failure occurs.

  In order to achieve the above object, in a database management system using a Shared Nothing (non-shared type) architecture, means for switching to an operating unit when a failure occurs is provided. In addition, a means for switching a plurality of servers (logical components providing a DB access function in a DBMS) included in a unit in which a failure has occurred to be distributed to a plurality of operating units is provided. In the means, a function for adding a server to be switched in the configuration to the switching destination unit, a server to be switched, and a DB recovery process for the DB storage area that the server is responsible for accessing and the server It has a function of performing transaction recovery for processing that was being executed when there was a failure. Furthermore, in order to stabilize the processing performance of the entire system after system switchover at the time of failure, the method of determining the server move destination (system switchover destination unit) at the time of failure has Therefore, the function to determine the server move destination, the function that the system statically determines the server move destination internally according to the guideline (policy) specified by the user in advance, or the DBMS fails according to the guideline (policy) specified by the user in advance It has a function of dynamically determining the server migration destination when it occurs.

  According to the present invention, it is possible to reduce standby dedicated resources.

It is a conceptual diagram of this invention. It is a figure which shows the functional block of the database processing system of this embodiment. It is a figure which shows an example of the hardware constitutions of the computer system in this embodiment. It is a figure which shows an example of the server-DB storage area charge information of this embodiment. It is a figure which shows an example of the server movement destination information at the time of the failure in this embodiment. It is a figure which shows an example of the server-unit structure information of this embodiment. 5 is a flowchart illustrating a procedure of system switching processing when a failure occurs in the system monitoring / system switching control mechanism according to the embodiment. 6 is a flowchart illustrating a procedure for system switching processing when a failure occurs in a switching destination unit according to the present exemplary embodiment. It is a flowchart which shows the process sequence of server movement destination unit determination in the system switching process at the time of the failure generation of this embodiment. 10 is a flowchart illustrating a processing procedure for determining a server migration destination unit in a system switching process when a failure occurs according to another embodiment. It is a flowchart which shows the procedure of the user inquiry process after the switching process at the time of failure occurrence in the process request control part of this embodiment.

Hereinafter, the best mode for carrying out the present invention will be specifically described with reference to the drawings.
First, the concept of the present invention will be briefly described with reference to FIG.

  The database management system of the present embodiment includes a processing request reception server (FES: Front End Server) 10 and a DB access server (BES: Back End Server) 20. The DB access server is also called a DB processing server. The server here is a logical functional component in the database management system, and does not indicate a physical machine, that is, an information processing apparatus. The various servers in this example are realized by programs, objects, processes, and threads.

  The processing request reception server (FES) 10 receives and analyzes a query 70 from a user, generates a DB access request, and makes a DB access request to the DB access server. Then, the DB access result is returned to the user as necessary. The DB access server (BES) 20 receives a DB access request from the processing request receiving server 10, operates data in the DB storage area according to the request, and returns a result to the processing request receiving server as necessary. Both FES 10 and BES 20 are implemented by one or more processes or threads.

  The architecture of the database management system of this embodiment is a Shared Nothing (non-shared type) architecture, and the database (for example, table and index) managed by this system is divided into a plurality of divided tables and divided indexes by various methods. , Distributedly stored in a plurality of DB storage areas. A certain DB storage area is associated with a predetermined DB access server, and the DB access server accesses only data (for example, table data and index data) in the DB storage area associated with the DB access server.

  In the example of FIG. 1, the BES 1 normally processes only the access request to the DB storage area 1, and the BES 2 processes only the access request to the DB storage area 2. Similarly, the BES 3 processes only the access request to the DB storage area 3, the BES 4 processes only the access request to the DB storage area 4, and the BES 5 processes only the access request to the DB storage area 5. BES1, BES2, BES3, BES4 and BES5 do not access the same DB storage area. Which BES processes an access request for a certain DB storage area is statically determined in the system by server-DB storage area handling information 50 described later.

  Normally, BES1, BES2, BES3, BES4, and BES5 are all in an operating state, and all resources (process, memory, and the like for realizing the DB access server) are effectively utilized.

  Here, for example, when a failure such as a power supply occurs in the information processing apparatus 3200 and the DBMS 2 (unit 2) having the BES 1, BES 2, and BES 3 goes down, the BES is in the other active DBMS unit (unit 1 and unit 3 in this example). Take over the processing every time and continue the service. That is, the failure target BES1 and BES2 move to the operating unit unit1. Similarly, the fault target BES 3 moves to the operating unit unit 3.

Specifically, disposed in the system monitoring and system switching control mechanism 5 and the unit unit3 same information processing apparatus on the disposed that the unit unit2 failure in the unit unit1 same information processing apparatus and on the The system monitoring / system switching control mechanism 5 is detected and BES1, BES2
, BES3 executes and controls the system switching described below for each BES. BE
An example of system switching in S1 will be described.

  First, the system monitoring / system switching control mechanism 5 that has detected the failure of the BES 1 determines the unit of the server movement destination (system switching destination) of the failure target BES 1 from the server movement destination information 30 at the time of the failure (703). Let's say that unit unit1 is decided. In that case, the shared disk device (60) in which the DB storage area 1 (61) to which the BES1 is responsible for the access is switched so that the information processing device in which the unit unit1 is arranged can be accessed. State (704). Then, the system switching communication is made to DBMS 2 (unit unit 1) by designating BES 1 (705).

  The DBMS 2 (unit unit 1) that has received the communication from the system monitoring / system switching control mechanism 5 first changes the configuration information to add the BES 1 to be switched to the unit unit 1 (712). Next, the fact that BES1 has been added to the unit unit1 is notified to another unit arranged in another information processing apparatus. Then, activation (start) of BES1 to be switched is performed (714). DB recovery processing for the DB storage area 1 is performed as an extension of the startup processing.

  The server-based system switching process described above does not affect the processes of other BESs (BES4) that are operating in the system switching destination unit that is operating. Similarly to BES1, the other failure target servers BES2 and BES3 are independently switched to active units (unit unit1 and unit unit3, respectively).

  Here, the FES (10) that has received the query request 70 from the user (that is, the application program) after switching analyzes the query request and determines a DB storage area in which data to be accessed exists. The DB access request 80 is transmitted to the unit currently operating (after system switchover) in the DB access server in charge of access to the determined DB storage area.

  The requested operating unit executes the received DB access processing at the designated server (BES1), accesses the DB storage area 1 and performs the requested data operation. In this example, “12” of the table data (62) stored in the DB storage area 1 (61) in the database 60 is transmitted to the FES as an execution result. The transmitted table data is returned to the user as an inquiry result by FES (10). That is, it is possible to realize a DB access processing request to the failure target BES 1 by exactly the same processing as the BES 4 operating in the unit unit 1 before the occurrence of the failure.

  As can be seen in this example, the BES1, BES2, and BES3 that were operating in the fault target unit unit2 are not all switched to one specific unit. It is also possible to minimize the degradation of the processing performance of the entire system after switching by specifying the first). Here, considering the transaction arrival rate for the active unit that is the switching destination as a load, unit unit1 is 300% before pair switching and unit unit3 is 200% before pair switching. Compared to 400% when unit unit1 or unit3 is assumed as a specific unit, the effect of load distribution can be seen even when focusing on one unit. For the unit unit1 and unit3, it is desirable in the system design that the BES equivalent to the unit unit2 is placed and operated in the actual system from the normal time. In that case, one BES is switched to the state in which three BES are in operation ( It will be 133% before the pair is switched.

  Here, the server destination information 30 at the time of failure is described as being registered in the DBMS in advance by a user such as the administrator of the DBMS, but the burden of the administrator is automatically generated by the DBMS internally. Can be reduced. In that case, it can be considered that the number of BESs is allotted to be equal to the number of units in the system. Of course, in order to reduce the burden on the administrator, the server migration destination information 30 at the time of failure may be generated by a tool that receives the server-unit configuration information of the system.

Further, the server migration destination unit may be dynamically determined when a failure occurs. In that case, it is conceivable that a DBMS administrator or the like designates a guideline (policy) 90 for determining a server migration destination in advance. Examples of guidelines (policy) for determining the server migration destination include the following.
(1) A server movement destination (switching destination) is selected for an operating unit with a light load such as a CPU usage rate, a transaction arrival rate, and a disk access rate.
(2) A server movement destination (switching destination) is selected for an active unit in which resources such as memory are available.
(3) A plurality of active units are switched, and server movement destinations (switching destinations) are allocated so that BESs are numerically equal.

  Information such as the load and resource availability listed in (1), (2) and (3) above is stored in the DBMS as operation information 100 and is referred to during system switching.

  As described above, when a failure occurs, a plurality of DB access servers (BESs) included in the failure target unit are distributed and switched to a plurality of active units for each server, thereby switching the system after the system switching at the time of the failure. It is possible to minimize the overall processing performance deterioration.

  Here, the FES and the BES are arranged on separate information processing apparatuses, but hardware resources can be effectively used by arranging them on the same information processing apparatus. Further, by implementing the FES function and the BES function shown in this embodiment as a single DB server, the administrator of the database management system does not have to be aware of the FES and BES, and the management cost is reduced. Can be reduced. The switching destination DB server in this example may wait in a state that can be executed by the unit when the server movement destination information 30 is set.

Next, the schematic configuration of the database management system of this embodiment will be described with reference to FIG.
There is an application program 6 created by a user and a database management system 2 that manages the entire database system such as inquiry processing and resource management. The database management system 2 includes a processing request reception server (FES) 10 and a DB access server (BES) 20. In addition, the database management system 2 includes a database buffer 230, a database 3 that permanently or temporarily stores data to be accessed by the database, server movement destination information 30 at the time of failure, server-unit configuration information 40, Server-DB storage area charge information 50, system switching destination determination guideline 90, and operation information 100 are included.

  The processing request reception server (FES) 10 receives and analyzes a query input from the application program 6, generates a DB access request, and makes a DB access request to the DB access server. Then, the DB access result is returned to the application program 6 as necessary. The DB access server (BES) 20 receives a DB access request from the processing request receiving server 10 and accesses the database 3 stored on the external storage device through the database buffer 230 according to the request. In the DB access processing of the DB access server to be switched as described above with reference to FIG. 1, the same database buffer as that used by the BES that was originally operating in the switching destination unit may be used. That is, the database buffer is shared between the original BES and the BES moved by the system switching.

  The database management system 2 is connected to other systems via a network or the like. Further, the processing request receiving server (FES) 10 and the DB access server (BES) 20 are not necessarily arranged on the same information processing apparatus. Each may be arranged on a separate information processing apparatus and function as one database management system via a network or the like. In addition, one database management system can distribute a load of requests from a large number of users by arranging a plurality of FESs. Further, by having a plurality of BESs, the degree of parallelism of data processing is increased, and data processing for a large-scale database can be realized at high speed. Hereinafter, a unit of a database management system having FES or BES arranged in one information processing apparatus will be called a unit or an instance.

  The processing request receiving server 10 performs syntax analysis / semantic analysis of the query, determines an appropriate processing procedure, generates a code corresponding to the processing procedure, and issues a DB access request to the DB access server 20. A control unit 211 is provided.

  The DB access server 20 includes a data processing control unit 221 that performs access control of data on the database 3 in accordance with a DB access request (generated code) received from the processing request receiving server 10.

  The DB access server 20 receives a system switching request in cooperation with the system management and system switching function 5 when a failure occurs in another information processing apparatus or a unit disposed in another information processing apparatus. A request from the system monitoring / system switching mechanism cooperation unit 222, 222 for starting the system switching processing is performed under the control of the server unit system switching processing units 223, 223 for controlling the system switching processing of the designated server subject to failure. Server-unit configuration information management unit 224 that adds the configuration of the server to be switched to the relevant unit as a part, and also starts the switching server, recovers the DB storage area that the server is responsible for accessing, and executes it A server activation control unit 225 for controlling transaction recovery for canceling the update result of the processing that was in progress That. The server-unit configuration information management unit 224 is also provided in the above-described processing reception server 10.

  FIG. 3 is a diagram illustrating an example of a hardware configuration of the computer system according to the present embodiment. The computer system of this example includes information processing apparatuses 3000, 3100, and 3200.

  The information processing device 3000 includes a CPU 3002, a main storage device 3001, a communication control device 3003, an I / O control device 3004, and a terminal 3006. An application program 3008 is placed on the main storage device 3001 and operates using the CPU 3002. When the application program 3008 makes a user inquiry to the processing request reception server 10 of the DBMS 2, the processing request reception server 10 is transmitted via the network 3007 by the communication control device 3003 of the information processing device 3000 and the communication control device 3003 of the information processing device 3100. An inquiry request is sent to

The information processing device 3100 includes a CPU 3002, a main storage device 3001, a communication control device 3003, an I / O control device 3004, an external storage device 3005 such as a magnetic disk device, and a terminal 3006. On the main storage device 3001 of the information processing device 3100, the database management system 2 having the processing request receiving server 10 described above with reference to FIG. 2 is placed and is operating using the CPU 3002. A database 3 managed by the database management system 2 is stored on the external storage device 3005. The processing program 3100 A for implementing a database management system 2 is also stored in the external storage device 300 5 above. The processing request reception server 10 reads / writes data from / from the external storage device 3005 by the I / O control device 3004 and transmits / receives data to / from other information processing devices connected via the network 3007 by the communication control device 3003.

The information processing device 3200 includes a CPU 3002, a main storage device 3001, a communication control device 3003, an I / O control device 3004, an external storage device 3005 such as a magnetic disk device, and a terminal 3006. A database management system 2 having the DB access server 20 described above with reference to FIG. 2 is placed on the main storage device 3001 of the information processing device 3200 and is operating using the CPU 3002. A database 3 managed by the database management system 2 is stored on the external storage device 3005. The processing program 3100 A for implementing a database management system 2 is also stored in the external storage device 300 5 above. The DB access server 20 reads / writes data from / to the external storage device 3005 by the I / O control device 3004 and transmits / receives data to / from other information processing devices connected via the network 3007 by the communication control device 3003.

  Here, the external storage device 3005 storing the databases 3 respectively associated with the two information processing devices 3200 is a shared disk and can be accessed from the other information processing device. Monitoring of the operation status of the database management system 2 (system monitoring) and a system monitoring / system switching control mechanism 5 (sometimes referred to as clusterware) for controlling a system switching operation in response to occurrence of various failures cause the shared disk to be Access control is performed.

  FIG. 4 is a diagram showing an example of server-DB storage area charge information according to this embodiment.

  In the example of FIG. 4, the server-DB storage area management information 50 is in charge of access to the DB storage area identified by the DB storage area name (401) indicating the DB storage area and the DB storage area name (401). It consists of the server name (402) of the DB access server. The information entry 411 indicates that the DB storage area 1 is in charge of access by the BES 1. Similarly, the information entry 412 indicates that the BES2 is responsible for accessing the DB storage area 2, and the information entry 413 indicates that the BES3 is responsible for accessing the DB storage area 3. This information is static system configuration information that is not changed across system switching.

  FIG. 5 is a diagram showing an example of server destination information at the time of failure in the present embodiment.

In the example of FIG. 5, the server migration destination information 30 at the time of failure is the system switch destination when a failure occurs in the server name (501) indicating the DB access server and the server identified by the DB access server. It consists of a unit name (502) of the unit. The information entry 511 indicates that the BES1 is switched to the unit unit1 when the failure occurs, that is, the process is taken over as the BES1 arranged in the unit unit1 and the service is continued. Similarly, the information entry 512 indicates that BES2 is switched to unit unit1 when a failure occurs, and the information entry 513 indicates that BES3 is switched to unit unit3 when a failure occurs. This information may be specified in advance by the DBMS administrator when the system is started, or the system may automatically generate the information inside the system. In that case, a guideline (policy) 90 for determining a server switching destination is set in advance by the user. As an example of the policy (policy),
(1) All DB access servers included in a unit are generated so as to be switched to different units. (2) The DB access servers are generated so that they are evenly switched between a plurality of specific units. .

  In many cases, the server destination information 30 at the time of failure is arranged in a memory on an information processing apparatus on which the DBMS operates and is easily accessed from the DBMS. Further, in this example, only one information entry is described for each of BES1, BES2, and BES3. However, it is possible to have a plurality of entries, that is, to have a plurality of switching destination units. The switching destination may be determined from the unit operation information and the priority as the plurality of switching destination units.

  FIG. 6 is a diagram showing an example of server-unit configuration information according to the present embodiment. Information before the failure of the unit unit2 is assigned to the upper unit 611 to 615, and information to the unit unit2 is assigned to the lower unit 621 to 625. Information after a failure occurs and system switchover is performed is shown.

  In the example of FIG. 6, the server-unit configuration information 40 includes a server name (601) indicating a DB access server and a unit name (602) of a unit in which the DB access server is configured. As shown in the information entry 611 and the information entry 621, the information entry 612 and the information entry 622, and the information entry 613 and the information entry 623, the BES1, BES2, and BES3 that are normally arranged and operated in the unit unit2 Due to the system switching accompanying the generation, the BES1 and BES2 are changed to the unit unit1, and the BES3 is changed to the unit unit3.

  FIGS. 7 and 8 are flowcharts showing the procedure of the system switching process when a failure occurs in the system monitoring / system switching control mechanism and the switching destination unit of this embodiment. FIG. 7 is a system monitoring / system switching control mechanism, and FIG. 8 is a flowchart showing a processing procedure in the switching destination unit.

First, the system monitoring / system switching control mechanism 5 detects the failure of another unit in step 701, and acquires server destination information at the time of failure in step 702. Then, based on the acquired server migration destination information at the time of failure, a server migration destination, that is, a system disconnection destination unit is determined (step 703). When the unit is determined as the switching destination (step 704), the process proceeds to step 705 to switch the shared disk. Further, in step 706, a failure occurrence server (switching target server) is designated for the DBMS, and system switching notification is performed. Further, in step 704, if the unit was not determined as the switching destination, it terminates the process proceeds to step 707 processing.

  The unit (DBMS (2)) that has received the system switching notification from the system monitoring / system switching control mechanism 5 in step 801 changes the server-unit configuration information in step 802. Specifically, the arrangement unit of the designated switching target server is changed to the unit. This is because the designated server moves to the unit by system switching. Subsequently, in step 803, notification is made to another unit arranged in another information processing apparatus that the switching target server has been added to the unit. In step 804, the server to be switched is started (started). In step 805, DB recovery processing is performed on the DB storage area to which the switching target server is responsible for access, and transaction recovery processing for canceling the update result of the processing that was being executed when the failure occurred.

  FIG. 9 is a flowchart illustrating a processing procedure for determining a server migration destination unit in the system switching process when a failure occurs according to this embodiment.

  First, in step 901, a server move destination information entry at the time of failure is searched, and an information entry related to the failure target server is determined. In step 902, the unit name of the server migration destination unit is acquired from the confirmed information entry. Here, in step 903, it is determined whether the acquired unit is currently operating and is valid as a switching destination. If the acquired unit is not in operation, the process returns to step 901 to search for another switching destination unit. If the acquired unit is in operation, the process proceeds to step 904, and the acquired unit is determined as the server migration destination.

  FIG. 10 is a flowchart illustrating a processing procedure for determining a server migration destination unit in a system switching process when a failure occurs according to still another embodiment. Here, a processing procedure for dynamically determining a server migration destination unit when a failure occurs will be described. A guideline (policy) 90 for determining a server migration destination is designated in advance by a DBMS administrator or the like, and the designated information is held in the DBMS.

  First, in step 1001, a guideline (policy) for determining a switching destination is acquired. Then, the guideline (policy) acquired in step 1002 is determined, and the switching destination is determined based on the determination result. In this example, three guidelines (policy) are shown.

  First, when the policy is “lightly loaded unit” in step 1002, the process proceeds to step 1011 and the load information of the own unit is acquired. Examples of load information include a CPU usage rate, a transaction arrival rate, a disk access rate, and the like. Further, in step 1012, load information of other operating units is also collected, and in step 1013, the load statuses of the own unit and other units are compared. If the load of the own unit is lighter than the load of the other units in step 1013, the process proceeds to step 1003 and the own unit is determined as the switching destination. If the load on the own unit is heavier than other units in step 1013, the process proceeds to step 1004, and the own unit is not determined as the switching destination.

  As the second policy, when the policy is “unit with free resources” in step 1002, the process proceeds to step 1021, and the resources of the own unit (control block for managing information related to the memory and server) are free. Get status. Further, in step 1022, the availability of resources in the other operating unit is also acquired, and in step 1023, the availability of resources of the own unit and other units is compared. If the resource of the own unit is larger than that of the other unit in step 1023, the process proceeds to step 1003 and the own unit is determined as the switching destination. When the resource of the own unit is less than that of the other unit in step 1013, the process proceeds to step 1004, and the own unit is not determined as the switching destination.

As a third policy, when the policy is “equally switch to a plurality of active units” in step 1002, the process proceeds to step 1031 and the operation information of other units is acquired. Own unit by the obtained number of-operation unit determines whether the switching destination, particularly if switching suitable as the switching destination, and determines the destination switching the own unit proceeds to step 1003. When own unit is not the destination toggles by number or the like of the server that is the other-operation unit and switching target, the process proceeds to step 1004, not determined as the destination switching the own unit.

  FIG. 11 is a flowchart illustrating the procedure of the user inquiry process after the switching process when a failure occurs in the processing request control unit of the present embodiment.

  First, in step 1101, a user inquiry is received and analyzed, and in step 1102, a DB storage area to be accessed for realizing the user inquiry is specified. Subsequently, in step 1103, the server-DB storage area charge information is searched, and the server (BES) responsible for access to the specified DB storage area is specified (step 1104). In step 1105, the server-unit configuration information is searched, and the unit in which the specified server (BES) is operating is specified (step 1106). A DB access request is transmitted to the server specified before the unit specified here (step 1107). Finally, the DB access result is received from the transmission destination (step 1108) and returned as a user inquiry result (step 1109).

  Note that the processing of the flowcharts shown in FIGS. 7, 8, 9, 10 and 11 is executed as a program in the computer system shown as an example in FIG. However, the program is not limited to the program stored in the external storage device physically connected directly to the computer system as in the example of FIG. It can be stored in a storage medium that can be read and written by a computer, such as a hard disk device or a flexible disk device. Moreover, it can also be stored in an external storage device connected to an information processing apparatus different from the information processing apparatus constituting the computer system of FIG. 3 via a network.

  According to the above, in a database management system using a Shared Nothing (non-shared) architecture, without a standby-only resource (machine, DBMS instance) in case of failure occurrence, after system switchover at the time of failure occurrence However, it is possible to minimize the degradation of the processing performance of the entire system and maintain stable performance.

2 ... Database management system 5 ... System monitoring / system switching control mechanism 20 ... DB access server 30 ... Server move destination information in case of failure 40 ... Server-unit configuration information 50 ... Server-DB storage area charge information 60 ... Database 90 ... System Switching destination determination guideline 100... Operation information.

Claims (7)

Database management in a database management system comprising a plurality of database processing devices having a database processing program, dividing and storing the database in a plurality of storage areas, and performing database processing by executing the database processing program assigned to each of the storage areas In the method
When the detection means of the database processing apparatus in which no failure has occurred detects that a failure has occurred in the database processing apparatus, the database processing apparatus name and the name of the database processing apparatus that is the switching destination at the time of the failure A step of referring to the switching destination information at the time of failure,
Obtaining a database processing device name included in the failure time switching destination information related to the database processing device in which the failure has occurred;
Determining the database processing device indicated by the acquired database processing device name as the database processing device to be switched;
Allocating the storage area allocated to the database processing apparatus in which the failure has occurred in the database processing apparatus serving as the system switching destination to a database processing program operating in the database processing apparatus serving as the system switching destination;
Including starting a database processing program in the database processing apparatus that is the system switching destination,
The database processing program that has performed the start processing uses a database buffer provided in the switching destination database processing device as a DB storage area associated with another database processing program that has been operating in the switching destination database processing device. A database processing method characterized by that. The step of starting the database processing program in the database processing apparatus that is the system switching destination,
Recovering the divided database storage area associated with the failed database processing apparatus;
The database processing method according to claim 1, further comprising a step of recovering a process being executed when the database processing apparatus detects that a failure has occurred.   The database processing method according to claim 1, wherein the failure switching destination information is generated in advance by designation from a user.   The database processing method according to claim 1, wherein the failure switching destination information is generated by the database management system according to a guideline given in advance by a user.   The guideline is generated so that a plurality of database processing programs included in a specific database processing apparatus are all switched to different units, or generated so that an equal number of database processing programs are switched to a specific database processing apparatus. The database processing method according to claim 4, wherein: In a database management system comprising a plurality of database processing devices having a database processing program, dividing and storing a database in a plurality of storage areas, and performing database processing by executing a database processing program assigned to each of the storage areas,
When the detection means of the database processing apparatus in which no failure has occurred detects that a failure has occurred in the database processing apparatus, the database processing apparatus name and the name of the database processing apparatus that is the switching destination at the time of the failure Means for referring to the switching destination information at the time of failure,
Means for acquiring a database processing device name included in the failure time switching destination information related to the database processing device in which the failure has occurred;
Means for determining the database processing device indicated by the acquired database processing device name as the switching-destination database processing device;
Means for allocating the storage area allocated to the database processing apparatus in which the failure has occurred in the database processing apparatus serving as the system switching destination to a database processing program operating in the database processing apparatus serving as the system switching destination;
Means for performing start processing of a database processing program in the database processing apparatus as the system switching destination,
The database processing program that has performed the start processing uses a database buffer provided in the switching destination database processing device as a DB storage area associated with another database processing program that has been operating in the switching destination database processing device. A database processing system characterized by that. Database management in a database management system comprising a plurality of database processing devices having a database processing program, dividing and storing the database in a plurality of storage areas, and performing database processing by executing the database processing program assigned to each of the storage areas In a database management program for realizing the method,
When the detection means of the database processing apparatus in which no failure has occurred detects that a failure has occurred in the database processing apparatus, the database processing apparatus name and the name of the database processing apparatus that is the switching destination at the time of the failure A step of referring to the switching destination information at the time of failure,
Obtaining a database processing device name included in the failure time switching destination information related to the database processing device in which the failure has occurred;
Determining the database processing device indicated by the acquired database processing device name as the database processing device to be switched;
Allocating the storage area allocated to the database processing apparatus in which the failure has occurred in the database processing apparatus serving as the system switching destination to a database processing program operating in the database processing apparatus serving as the system switching destination;
Performing a database processing program start process in the database processing apparatus as the system switching destination;
The database processing program that has performed the start processing uses a database buffer provided in the switching destination database processing device as a DB storage area associated with another database processing program that has been operating in the switching destination database processing device. And a database management program.

Images