JP2001282762A - Device and method for backing up data in distributed processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a device and a method for backing up data in a distributed processing system, by which recovery processing of a database is speedily performed and a system is never stopped for a long time. SOLUTION: Databases stored in database storage parts 122 and 142 are backed up at a prescribed time interval in backup data storage parts 124 and 144 and the update information of the databases after the start of backup is held in archive data storage parts 126 and 146 as archive data. When a fault occurs, the backup data are copied, and then all the data bases are recovered while using the archive data. In this case, the stop time of the system at the time of recovery can be made only as long as the time for recovering the databases from the archive data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a plurality of servers,
The present invention relates to a data backup apparatus and method in a distributed processing system including a plurality of clients connected to the respective servers and performing distributed processing of a control target.

[0002]

2. Description of the Related Art Computer processing has been performed in various fields due to the recent development of information processing apparatuses and information processing techniques. For example, an online processing system in a financial institution or the like is well known. In this system, as a recovery measure in the event of a system failure, the system is completely duplicated, and one is a normal operation (working) system and the other is a standby standby system. If a failure occurs during operation, switch to the standby system,
It is considered to continue the service.

[0003] In this case, the data to be processed by the system and the database storing information required by the system for the processing so that the system can be switched from the active system to the standby system at any time are stored in the active system. If the contents of the active database are updated as a result of the processing set by each of the standby systems and executed by the active system,
Update the corresponding part of the standby database to the same content,
We operate to keep the contents of the database the same.

In order to maintain the database identity, the standby system simply stands by and does not execute any other processing. Therefore, the standby system is updated from the active system to the standby system. It is only necessary to transmit the data write processing to the database, and no processing such as contention control for database access is required.

In such an online processing system, although the system scale is large, the social impact of stopping the system due to a failure is enormous.
It is unavoidable that the burden of equipment costs due to the weight increase is unavoidable.

On the other hand, in a small-scale system such as an information processing system in a company, it is difficult to employ the above-described complete duplex system from the viewpoint of effective use of resources. In many cases, the standby system is also configured to execute another process different from the task processed by the active system.

In particular, in recent years, computers have been reduced in size and performance, and high-speed processors, large-capacity main memories, large-capacity hard disk drives, and the like have been mounted at low cost. As an information processing system, many clients are connected to multiple servers,
A distributed processing system in which processing in each client is distributedly controlled by each server is employed.

[0008] In such a small-scale system or a distributed processing system in a company, the contents of a database constructed in each server are kept the same, and when a failure occurs in one of the servers, processing is performed in another server. Database that can be migrated, and in the event of a failure such as the database built on one of the servers being destroyed, the database on the failed server can be restored using the database on the normal server Various backup functions have been proposed.

For example, the identity of the contents of the database constructed in each server is always maintained, and when a failure occurs,
There is one in which file recovery processing is performed using data of another server (Japanese Patent Laid-Open No. Hei 7-114495,
See JP-A-5-265829. As a method of backing up a database, there is a method of shortening a backup processing time by collecting target data files into an archive file before backup (see Japanese Patent Application Laid-Open No. H11-53240).

Further, as a method for recovering a failed database by using a database of a normal server when a failure such as the database of any server is destroyed, for example, a method shown in FIG. 6 is known. ing.

In the distributed processing system shown in FIG. 6, a plurality of servers 1 and 2 and clients 3 and 4 are connected by a communication cable 5 such as a LAN. Storage units 6 and 7
Is constructed. The servers 1 and 2 are configured to process the processing requests of the clients 3 and 4 in a distributed manner.

Servers 1 and 2 and clients 3 and 4
Are server applications 1b stored according to the tasks to be processed by the distributed processing system,
2b and user applications 3a, 4a,
A predetermined process is performed by these applications 3a and 4a.

When the contents of the databases of the servers 1 and 2 are updated by the processing requests from the clients 3 and 4, each of the servers 1 and 2
Communication between servers is performed by a and 2a, and the updated content of each database is reflected on another database, thereby maintaining the same content of the databases.

In such a distributed processing system, if a failure that destroys the contents of the database of one server, for example, the server 2 occurs for some reason, the failure is performed using the contents of the database of the normal server 1. Repair the contents of the database where the error occurred.

In this case, conventionally, the database was restored in the following procedure using the database export / import functions.

That is, first, the processing operations of the clients 3 and 4 are stopped, and the contents of the database storage unit 6 are stored in the backup data storage unit 6a by using the export function of the backup processing unit 8a of the normal server 1. Next, using the import function of the backup processing unit 9a of the server 2, the contents of the backup data storage unit 6a are imported into the database storage unit 7, and the database updated after the occurrence of the database failure until the system is stopped. Reflect the updated content, and then restart the operation of the distributed processing system.

In the database export / import function, the following processing is performed.
That is, the export function decomposes the database into information on table definitions, information on data, information on index definitions, information on trigger definitions, and information on integrity constraints, and backs up these information as export data. The data is stored in the data storage units 6a and 7a. Such data processing is performed because the data amount is reduced as much as possible and the backup data storage unit 6
This is to secure the capacities a and 7a. In the import function, a process of reconstructing a database from the data decomposed as described above is performed.

[0018]

However, when the database is backed up using the above-mentioned export and import functions, hardware resources for saving can be saved, but a great deal of processing is required for the database recovery processing. The disadvantage that time is required arises.

For example, in the processing using the import function in a general database, the processing time largely depends on the index definition of the database. That is, the rebuilding time of the index depends on the data amount, the number of indexes, and the number of fields (the number of columns) of the table used for the index, and the building time = data amount × index number × used field number × per field Is calculated as the processing time. In a database in a production system, especially, the amount of data such as the result table is large, and the user application needs to be able to quickly extract data from this viewpoint from many viewpoints. Have been. Therefore, a lot of import processing time is required.
In systems other than the production system that need to process the database from various viewpoints, the import process similarly takes a long time.

Further, during execution of the database restoration work, it is necessary to stop the processing of the normally operating server, that is, to suspend the entire distributed processing system during this time. This is because if a normally operating server accepts processing from a client during database restoration work and the database is updated, the contents of the database of the normal server and the database being restored will not match That's why.

In particular, when applied to control of machine equipment on a production line (hereinafter referred to as "process control") and production management work in a manufacturing factory where continuous operation is required for 24 hours, processing for restoring a database failure is performed. If the system downtime required for a long time is long, it will have a great effect on the maintenance of production efficiency.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and in a distributed processing system including a plurality of servers and clients, a database recovery process is quickly performed, and the system is constructed for a long time. An object of the present invention is to provide a data backup device and method in a distributed processing system that does not stop.

[0023]

In order to achieve the above object, the present invention provides a distributed processing system comprising a plurality of servers and a plurality of clients connected to the respective servers and performing distributed processing of a control target. The server includes a database storage unit that stores a database that is updated by a process performed on the control target by the plurality of clients, and a data replication process that is performed based on the database update process performed by the client that processes the control target. A replication trigger generation unit that generates a trigger, an update information transfer unit that transfers update information of the database to another server based on the data replication trigger, and the database of the other server based on the transferred update information. A database update processing unit to update, and operation of the database at predetermined time intervals A backup processor for performing backup process, the backup data storage unit for storing the backup data according to the operation during the backup process,
An archive data creation unit that creates archive data based on the update information of the database that occurs after the start of the backup process during operation; and an archive data storage unit that stores the archive data. The backup data and the archive The database is restored based on data.

Further, according to the present invention, according to a process for a control target of a plurality of clients connected to a plurality of servers,
Updating the database of each server connected to each client, generating a data replication trigger based on the database update processing, and updating the database update information based on the data replication trigger to another Transferring to a server, and updating the database of the other server based on the transferred update information, in a distributed processing system, performing backup processing during operation of the database at predetermined time intervals, Saving backup data, creating and saving archive data based on update information of the database that occurs after the start of the in-operation backup process, and when a failure occurs in one of the plurality of servers, The back of the other normal server With the copy process the Updater, characterized by comprising a combination of a step of restoring the database from said archive data of the other normal the server.

[0025]

FIG. 1 shows the configuration of a distributed processing system 10 to which the present invention is applied. In this distributed processing system 10, a plurality of servers 12, 14 and a plurality of clients 20, 22, 24, 26, 30, and 32 are connected to a LAN.
And the like.

When the distributed processing system 10 is operating normally, for example, the server 12 and the client 2
0, 22, 24, and 26 share processing of a process control system of a production line, for example, processing of controlling a manufacturing machine for producing a photographic film described later, and a server 14 and clients 30 and 32 execute production planning. , And is responsible for processing related to production management systems such as raw material management and inventory management.

The server 12 is a server for sharing the processing of the process control system, and the clients 20, 22, 24, 26
Database storage unit 122 for accumulating data such as production instruction information and production history information to be described later transmitted and received between
A backup processing unit 123 for backing up the contents of the database storage unit 122 online at regular time intervals and storing the backup data in the backup data storage unit 124;
An archive data creation unit 125 that creates archive data based on the update information of the database updated by the processes 4 and 26 and stores the archive data in the archive data storage unit 126, and a database update process by the clients 20, 22, 24, and 26 A replication trigger generation unit 132 that generates a data replication trigger based on the data replication trigger, and updates the database update information based on the data replication trigger to another server 14.
, And a database update processing unit 128 that updates data in the database storage unit 122 based on a database update processing request.

The archive data creation unit 125 creates archive data in accordance with new update information after the backup of the database in the database storage unit 122 to the backup data storage unit 124. Before the backup starts, the archive data storage unit 125 creates the archive data. 1
The archive data stored in 26 is discarded.
The time interval of the backup process to the backup data storage unit 124 can be arbitrarily set by the operator. For example, the period can be set to one day (24 hours). All the archive data from the start of the backup process to the start of the next backup process is accumulated in the archive data storage unit 126.
In addition, the update information transfer unit 130 performs update information transfer processing at fixed time intervals set by a system administrator.

The server 14 is a server for sharing the processing of the production management system, and has a database storage unit 142 for storing data such as a production plan, raw material management, and inventory management created between the clients 30 and 32. In addition, a backup processing unit 143, a backup data storage unit 144, an archive data creation unit 145, an archive data storage unit 146, a database update processing unit 148, an update information transfer unit 150, and a replication trigger generation unit 152 configured similarly to the server 12. Having.

The clients 20, 22, 24, 26
Based on the production instruction information from the process management system server 12, the cutting machine 40, the winding machine 42, and the exterior machine 4 shown in FIG.
6. User applications 220, 222, 224, and 226 for controlling the packing machine 50 and transmitting manufacturing history information and the like collected from each device to the server 12.
It has. Similarly, the clients 30 and 32 include user applications 230 and 232 for processing operations such as production planning, raw material purchase management, and product inventory management with the production management server 14.

The cutting machine 40 controlled by the client 20 cuts a master roll, which is a photographic film material wound into a roll, into a predetermined width to produce a slit roll. The winding machine 42 controlled by the client 22 performs a process of winding the slit roll supplied from the cutting machine 40 into the cartridge case formed and assembled by the forming assembly machine 44. The exterior machine 46 controlled by the client 24 includes the small box supply device 4
A process for packaging the film cartridge with the small box supplied from 8 is performed. The packing machine 50 controlled by the client 26 performs a process of packing the film cartridge stored in the small box on the cardboard supplied from the cardboard supply device 52.

Servers 12, 14 and Client 2
Although 0, 22, 24, 26, 30, and 32 are not shown, in terms of hardware, similar to a general PC (personal computer), a processor, a main memory,
It consists of a hard disk, network interface, etc.
It operates under the control of the operating system installed on each server and client.

Next, the operation of the distributed processing system 10 during normal operation will be described with reference to FIG.

First, the client 20 receives the individual production instruction information from the server 12 of the process control system and displays it on the monitor screen. The same processing is performed in the other clients 22, 24, and 26 of the process control system.

The individual production instruction information is created by the clients 30 and 32 connected to the production management system server 14 and transmitted to the process control system server 12 via the server 14. Information from the process control clients 20, 22, 24, and 26 gathered in the server 12 is transmitted to the production management server 14.

The individual production instruction information transmitted from the server 12 to the client 20 includes a production lot number of a master roll serving as a photographic film material suitable for the type of product specified by the production plan data. From among the master rolls stored in the storage, a master roll with a corresponding production lot number is supplied to the cutting machine 40 by a self-propelled carrier or the like.

The client 20 sends the slitting conditions to the cutting machine 40, such as the transport speed of the master roll and the cutting machine 4
The setting data of the inspection condition by the surface inspection device provided in the area 0 is transmitted. The cutting machine 40 cuts the master roll into a certain width based on the operating conditions transmitted from the client 20, thereby producing a slit roll.

At this time, a plurality of slit rolls are manufactured from one master roll. To make it possible to identify the slit rolls in the subsequent steps, the emulsion number, master roll manufacturing lot number, A label in which the slit roll number is printed with a bar code is created and affixed to each slit roll. This barcode is read by a barcode reader and input to the client 20.

Next, the slit roll is moved to the client 2
2 and is stored in a cartridge case.

Here, the individual production instruction information transmitted from the database storage unit 122 of the server 12 includes the emulsion number of the slit roll used for production, the production lot number of the master roll, and the slit roll number.

The winding machine 42 takes out the slit roll specified by the individual production instruction information from the storage and sets it. The bar code affixed to the set slit roll is read, and it is checked whether or not the slit roll corresponds to the designated slit roll.
Is transmitted to the server 12 via the. Then, the client 22 operates the winding machine 42 according to the conditions specified by the individual production instruction information from the server 12.

In the punching section of the winding machine 42, perforation is punched in accordance with the specified format. In the side print section, a latent image of product information is recorded. As the product information, an ID number (film ID number) using a barcode, a frame number, a product abbreviation, a manufacturer name, and the like are recorded. The film ID number is
It is stored in the server 14 together with the order number of the production plan data, and is also written in the production instruction information transmitted to the server 12 and transmitted to the client 22 that controls the winding machine 42.

When the side print section prints the film ID number, this film ID number is fed back to the client 22 together with other side print information, and the film I number specified by the individual production instruction information is sent.
It is confirmed whether or not it matches the D number. The client 22 further transmits other information obtained so far, that is, the emulsion number, the production lot number of the master roll, the slit roll number, and the like to the server 12 in association with the film ID number. This information is stored in the server 12 and also transmitted to the production management server 14.

After the side print processing, the cutting section is operated to cut the long photographic film into a length corresponding to the number of frames that can be photographed, thereby producing a photographic film strip. The manufacturing conditions for operating the cutting unit are included in the individual production instruction information transmitted to the client 22 via the server 12 as before.

The long photographic film cut into a predetermined length is
It is transported to the winding section. A cartridge case formed and assembled by the forming and assembling machine 44 is supplied to the winding unit from a predetermined stacking unit in tray units. At this time, the tray ID number assigned to the tray is read and the client 2
2 to the server 12.

Since the server 12 stores the cartridge ID number of the cartridge case accommodated in the tray and the manufacturing history information in association with the tray ID number,
What kind of range of cartridge ID number the cartridge case supplied to the winder 42 has, as well as what order number the cartridge case is manufactured, and the parts of any manufacturing lot number Can be used to track and confirm information such as whether or not it has been assembled.

When sequentially supplying the cartridge cases to the winding section, the cartridge ID number is read by the bar code of the label, and the read cartridge ID number is read.
The number is immediately entered into the client 22.

The cartridge case at the reading position corresponds to a cartridge case in which the side print unit is combined with a photographic film strip to be subjected to the next side print. Therefore, the client 22 checks the cartridge ID number immediately before the side print unit performs the side print.

The cartridge ID number read in this manner is transmitted to the server 12, where the cartridge ID number is compared with the film ID number assigned to the photo film strip by side printing.
As described above, the server 12 stores the film ID number and the cartridge ID number assigned to the product to be produced at the time of production plan data creation. It is possible to identify whether or not the cartridge ID number is appropriate.

The rear end of the photographic film strip is locked to the spool in the cartridge case by the operation of the winding section, and the spool is driven to wind the photographic film strip into the cartridge case. The product photo film cartridge is completed.

As described above, since the manufacturing history information of the photographic film strip such as the emulsion number, the master roll manufacturing lot number, and the slit number is already known, the manufacturing history information of the cartridge case and the photographic film strip is The information is also stored as production management information in the production control system server 14 via the process control system server 12 in association with the cartridge ID number or the film ID number.

Further, the photographic film cartridge which has become a product through the winding section is stored in the stacking section in a state where the photographic film cartridge is stored in a tray capable of storing a predetermined number of the photographic film cartridges. At this time, tray I
The D number is read and transmitted to the client 22. The client 22 checks the ID number (cartridge I) of the photo film cartridge stored in the tray.
D number or film ID number usage record and missing number information,
Further, the production history information is correlated and fed back to the server 12. This information is also transmitted from the server 12 to the production management server 14.

Thus, information such as the ID number of the photographic film cartridge stored in the stacking unit, the product type, the manufacturing history of the photographic film strip and the cartridge case can be grasped in correspondence with the tray ID number.

The client 24 controls the exterior machine 46 and the small box supply device 48. Photo film cartridge
After being conveyed from the winding machine 42 to the exterior machine 46 and packed in a film case, it is stored in a small box.

For the film case, which production lot number of the P case is to be used is specified by the production instruction information instructed from the server 12, and the production lot number of the used film case is determined via the client 24 through the process. It is sent to the server 12 of the control system.

The small box is provided with a display indicating the type of product and a bar code. The bar code indicating the type of product is printed in advance at the time of manufacturing the small box, but the bar code indicating the packaging material ID number is used when the small box is supplied from the small box supply device 48 to the exterior machine 46. Printed.

The packaging material ID number to be printed is determined by the production management system server 14 when the production plan is created. Then, as in the case of the side print of the film ID number, after confirming that the cartridge ID number of the photographic film cartridge packaged therein matches the packaging material ID number transmitted as production instruction information, the packaging material ID is checked. The number is printed.

The photographic film cartridge packed in the small box is then transferred to the packing machine 50 controlled by the client 26. The packing machine 50 receives the supply of cardboard boxes for packing from the cardboard supply device 52, and packs, for example, 1000 photo film cartridges into cardboard boxes.

When packing a predetermined number of products in a cardboard box, the packing section prints a packing ID number as a bar code on the surface of the cardboard box. This packing ID number is determined by the server 14 of the production management system in the same manner as the above-mentioned packaging material ID number.
For each product, the packaging material I printed by the exterior machine 46
One package ID number is assigned to the usage range of the D number. The correspondence is stored in the server 14 of the production management system. Therefore, if the packing ID is read from the cardboard box, the use range of the packaging material ID number of the product packed therein can be grasped.

On the other hand, the clients 30 and 32 perform operations such as creation of a production plan, purchase management of raw materials, and inventory management of products with the server 14 of the production management system. These clients 30 and 32 are operated by the workers in the production management department, and the above-described production instruction information is transmitted to the process management server 12 and collected and transmitted by the process management server 12. Product management data is created from the manufacturing history data of the product.

Next, the distributed processing operation will be described based on the system configuration of the clients 20 and 30 shown in FIG. Note that the same applies to the clients 22, 24, 26, and 28, and a description thereof will be omitted.

The client 20 includes a process control module 20a for controlling the cutting machine 40, a connection destination server information parameter 20b, and a connection destination server check module 20c. Similarly, the client 30 includes a production management module 30a for performing a production management process, a connection destination server information parameter 30b, and a connection destination server check module 30c. The servers 12 and 14 include operation state information tables 122a and 142a, respectively.

When the distributed processing system 10 is in a normal operation state, the client 20
The client 30 is connected to the production management system server 14 and performs production management work. For this reason, the connection destination server information parameter 20b of the client 20 specifies the connection destination server 12 by default, and specifies the production management server 14 as the backup server. The connection destination server information parameter 30b of the client 30 is
Conversely, the connection destination server 14 is specified by default, and the production management server 12 is specified as the backup server.

When the distributed processing system 10 is started, the connection destination server check module 20 of the client 20
The server whose c is specified in the connection destination server information parameter 20b (the server 12 initially specified by default)
Check. That is, the connection destination server check module 20c stores the server 1 specified by default.
Check the operation status information table 122a of the second server,
If No. 2 is operating normally, a connection is established. The operation of the client 30 is the same.

If the server 12 is not in the normal operation state, the operation state information table 14 of the server (the server 14 in the case of the client 20) designated on the backup side is used.
2a is confirmed, and if the server 14 is operating normally, a connection is established. If both servers 12 and 14 are not in the normal operation state, a connection error is displayed and the operation is stopped.

When the connection is established, the connection destination server check module 20c notifies the process control module 20a of the connection port information, and thereafter periodically polls the operation status information table 122a of the connection destination server. If the operation state of the client 20 is stopped, the application (process control module 20
a) is automatically stopped. The operation of the client 30 is the same.

When one of the servers fails and stops for some reason while the distributed processing system 10 is operating in a normal state, the client 20 or 30 connected to the stopped server is connected to the above-mentioned server. To stop automatically.
For example, when the server 12 of the process control system is stopped due to a failure, the worker operates the stopped client 20 to change the server 12 set as the default in the connection destination server information parameter 20b to the production management set as the backup side. The connection destination management software (not shown) is started to change the connection destination to the system server 14.

In the connection destination management software of the client 20, it is specified that the server connected at normal time is the server 12 and the server connected at the time of failure is the server 14. Reference numeral 20c denotes an operation status information table 142 of the server 14.
With reference to a, a connection with the server 14 is established.

At this time, since the database storage unit 122 of the server 12 and the database storage unit 142 of the server 14 have the same contents, as described later, even if the client 20 is connected to the server 14, The control process can be continuously executed. The same applies when the server 14 is stopped due to a failure. Therefore, the film manufacturing process can be continued without any trouble even if one of the servers 12 and 14 has a failure.

Next, the backup operation during the normal operation of the distributed processing system 10 will be described with reference to the processing flowcharts of FIGS. 1, 4 and 5.

In the flowchart shown in FIG. 4, the result of processing by the clients 20, 22, 24, and 26 on the control target of the cutting machine 40 or the like, or the result of processing by the operator, insertion, change, and deletion in the database of the database storage unit 122 When an update processing request such as the above is generated, the database update processing unit 128 of the server 12 sends an SQ indicating update information generated in accordance with the update processing request.
An L (Structured Query Language) is accepted (step S1). Then, whether this SQL is an update SQL for updating the database of its own database storage unit 122, or the database storage unit 142 of another server 14
Is determined to be a propagation SQL for updating the database (step S2).

If the update SQL is determined, the database update processing unit 128 performs update processing such as insertion, change, and deletion of the database into the database storage unit 122 in accordance with the content of the instruction of the SQL (step S).
3) The copy trigger generation unit 132 is instructed to generate a copy trigger. Duplicate trigger generation unit 13 receiving the command
2 generates a replication trigger for replicating the database (step S4), and based on the replication trigger, makes the update SQL a propagation SQL and supplies it to the update information transfer unit 130 (step S5).

In the flowchart of FIG. 5, the update information transfer unit 130 performs polling at a fixed time interval (step S21), and when the propagation SQL is supplied from the duplication trigger generation unit 132 (step S22), the The SQL is updated to the database update processing unit 148 of the server 14.
(Step S23). Then, when the propagation SQL reception signal is received from the database update processing unit 148 (step S24), the propagation SQL of the update information transfer unit 130 is deleted (step S25).

On the other hand, in the flowchart of FIG. 4, the database update processing unit 148 of the server 14 determines whether the SQL received in step S1 is the update SQL generated in the clients 30 and 32, SQL is determined (Step S
2).

If it is determined that the update is the propagation SQL, the database update processing unit 148 determines whether the update information of the propagation SQL is an insertion (step S6), a change (step S7), or a deletion (step S8). ) Is determined. Further, it is determined whether or not each change information has data (steps S9, S10, S11). In addition,
If there is no data, error processing is performed on the assumption that a transfer failure has occurred (step S12).

All the propagation SQLs transferred through the above processing
Is performed (step S13), the database of the database storage unit 142 is updated (step S13).
3). In this case, the database update processing unit 148 issues a replication trigger generation prohibition instruction to the replication trigger generation unit 152. Therefore, by the occurrence prohibition command, it is possible to prohibit the update information transfer unit 150 from transmitting update information based on the propagation SQL to the server 12.

As described above, the servers 12 and 14
Of the database storage units 122 and 142 are maintained the same, though asynchronously.

On the other hand, the backup processing sections 123 and 143 of the servers 12 and 14 store the database storage sections 122 and 1 at regular time intervals such as in units of one day (24 hours).
42 is stored in the backup data storage unit 12 during operation.
4, 144, and the database storage units 122, 1
42 backup data is created. This backup operation includes a full backup method in which all data contained in the database storage units 122 and 142 are backed up at one time, and only files whose data has been updated after the last full backup or newly created data. Differential backup methods for backing up data, incremental backup methods for backing up only files whose data has been updated since the last full backup or differential backup, and newly created data are known, and these backup methods are known. The backup data storage units 124 and 144 are created using any one of the above.

In this case, the archive data creation unit 12
5 and 145, at the timing when the creation of the backup data is started, discard the contents of the archive data storage units 126 and 146 that have accumulated the update information of the database up to that time, and generate the data in the database storage units 122 and 142 after this timing. The content of the new update process is stored as archive data in the archive data storage units 126, 1
The work to record in 46 is started.

Here, the backup data storage unit 12
The operation of backing up the database to the servers 4 and 144 is performed using the file copy function of the servers 12 and 14 at the OS level. The backup by the file copy function is performed by the database storage unit 122, 1
Unlike the above-described operation using the export function in which the structure information and the data themselves are converted into different formats and output to a file, the database is copied in its original format, and although a large hard disk resource is required, There is an advantage that the time required for backup is short.

For example, in the backup of a database having a data amount of 3 gigabytes, the backup by the export function takes about 600 minutes, whereas the backup by the file copy function takes about one-tenth of the time, ie, about 60 minutes. can do. However, when the backup data is created by the file copy function, it is necessary that the files in the database storage units 122 and 142 and the data in the backup data storage units 124 and 144 have completely the same settings and file names. is there.

With the above configuration, backup data is created while the servers 12 and 14 are running, and new update information for the database storage units 122 and 142 after the start of backup data creation is stored in the archive data storage unit 126. , 146, and efficiently and quickly in the database storage units 122, 1
42 backups can be performed. Further, as described later, when a failure such as the content of one database is destroyed, the content of the database can be quickly restored using the backup data and the archive data.

Next, a data recovery method when a failure occurs in which the contents of the database storage unit 142 of the server 14 is destroyed will be described.

In this case, the clients 30 and 32 connected to the server 14 switch the connection destination to the server 12 and operate the distributed processing system 10 by the server 12 alone. This allows you to stop the system
Production can be continued. Next, a recovery operation on the hardware of the failed server 14 is performed. For example, replacement of a hard disk as the database storage unit 142 is performed.

After performing the above processing, the contents of the backup data storage unit 124 of the server 12 are restored (file copy) to the database storage unit 142 of the server 14. By this operation, the database storage unit 1
A copy of the database identical to the contents at the time of backup of No. 22 is constructed in the database storage unit 142. During this process, the archive data storage unit 126 of the server 12 records the archive data since the start of the backup during operation.

The operation for restoring the contents of the backup data storage unit 124 to the database storage unit 142 of the server 14 is performed by using the file copy function as described above. Therefore, each server 1
The databases of the database storage units 122 and 142 are completely identical, including the settings, are placed on the hard disk drive of the same name, and are connected to the server when one server fails. It is necessary to set a different global name for external access in order to take over the processing of the client that has been performed.

Next, after the backup data copy processing is completed, the operation of all the clients 20, 22, 24, 26, 30, 32 connected to the server 12 is performed so that the database storage unit 122 is not updated. Is temporarily stopped. In this state, the contents of the archive data stored in the archive data storage unit 126 of the server 12 are applied (written) to the database storage unit 142 of the server 14.
I do. This write operation is automatically performed by a so-called roll-forward process. This roll-forward function is provided by being incorporated in general database processing software. Storage units 122 and 142
Is restored to the same content.

After the above recovery work is completed, the clients 30 and 32 are connected to the server 14 again,
Restarting the distributed processing systems 1 and 2
0 operation is resumed.

Here, the time during which the distributed processing system 10 is completely stopped is only when the archive data excluding the backup data is applied to the database.
For example, if the time interval of the backup during operation is set to one day, the amount of the archive data is 1 at a maximum.
This is the amount of database updates for the day. The time required to restore the database from the archived data is several minutes to several tens of minutes, so that the period during which the distributed processing system 10 is stopped can be suppressed to this level, and the database recovery by the conventional export and import functions can be performed. System stoppage time can be greatly reduced as compared with the time required for (8 hours).

[0090]

As described above, according to the present invention, in a distributed processing system constituted by a plurality of servers and clients, if it becomes necessary to restore a database constructed in a server, the system is constructed for a long time. The database can be restored in a very short time without stopping.

Further, cost performance can be improved without increasing the scale of the system.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a distributed processing system according to the present invention.

FIG. 2 is a block diagram showing a configuration when a distributed processing system according to the present invention is applied to a production line.

FIG. 3 is a connection control block diagram of a client in the distributed processing system according to the present invention.

FIG. 4 is a flowchart showing a processing procedure in a database update processing unit in the distributed processing system according to the present invention.

FIG. 5 is a flowchart showing a processing procedure in an update information transfer unit in the distributed processing system according to the present invention.

FIG. 6 is a block diagram showing a configuration of a conventional distributed processing system.

[Explanation of symbols]

10 Distributed processing system 12, 14 Server 20, 22, 24, 26, 30, 32 Client 122, 142 Database storage unit 123, 143
.. Backup processing units 124 and 144 Backup data storage units 125 and 145 Archive data creation units 126 and 146 Archive data storage units 128 and 148 Database update processing units 130 and 150 Update information transfer units 132 and 152
… Replication trigger generator

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G06F 12/00 545 G06F 12/00 545A 12/16 310 12/16 310M 15/16 620 15/16 620B F Terms (reference) 5B018 GA06 HA05 KA13 KA22 MA12 RA11 5B034 BB02 DD06 5B045 BB12 BB28 BB47 DD18 JJ13 JJ22 JJ26 JJ32 JJ42 5B082 DA02 GA03 HA00 9A001 BB03 DD09 EE01 EE06 JJ01 LL02 LL07 LL09

Claims (3)

[Claims] 1. A distributed processing system comprising: a plurality of servers; and a plurality of clients connected to the respective servers and performing distributed processing of a control target, wherein each of the servers is configured to control a plurality of clients for the control target. A database storage unit that stores a database that is updated by processing; a replication trigger generation unit that generates a data replication trigger based on an update process of the database by a process on the control target of the client; An update information transfer unit that transfers update information of a database to another server; a database update processing unit that updates the database of the other server based on the transferred update information; operation of the database at predetermined time intervals A backup processing unit for performing medium backup processing; A backup data storage unit that stores backup data obtained during the backup process in operation; an archive data creation unit that creates archive data based on update information of the database generated after the start of the backup process in operation; and stores the archive data. An archive data storage unit for recovering the database based on the backup data and the archive data. 2. A process for updating a database of each server connected to each client according to a process for a control target of a plurality of clients connected to a plurality of servers; and copying data based on the updating process of the database. Generating a trigger; transferring the update information of the database to another server based on the data replication trigger; updating the database of the other server based on the transferred update information; Performing a backup process during operation of the database at predetermined time intervals and storing backup data; and archive data based on update information of the database generated after the start of the backup process during operation. Create and save And when a failure occurs in one of the plurality of servers, copying the backup data of the other normal server and restoring a database from the archive data of the other normal server. A data backup method characterized by combining and. 3. The data backup method according to claim 2, wherein said copy processing of said backup data is performed while continuing the operation state of said plurality of clients by another normal server.