JPH06309211A - Distributed information processing system and destroyed information synchronous restoring method - Google Patents

Abstract

PURPOSE:To minimize a part of update information which can not be restored even in case of information destruction and to restore the information to a normal state having no contradiction among all pieces of update information by generating history information by using a history information means each time information is updated and storing it in a history information storage means. CONSTITUTION:At the same time with the update processing of the update information, the history information control means 7 of a processing module 1 requests the update serial number control means 14 of a control module 12 to outputs an update serial number. The update serial number control means 14 outputs the serial number which is unique and sequential in the system and informs all relative processing modules 1 of it. The history control means 7 of the respective processing modules 1 having received it obtain information required to edit history information from their internal update processing means 5 to edit the history information, and store the history information in history information storage means 8. If information is destroyed, an update information restoration processing means 11 restores the update information by using the stored history information.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention restores the destroyed information based on the history information of another processing module even when the update information and the history information stored in a certain processing module are destroyed. The present invention relates to a distributed information processing system and a destruction information synchronization recovery method thereof.

[0002]

2. Description of the Related Art Conventionally, as a countermeasure against a file failure, when a failure occurs during the running of a file update program, a file of history information required at the time of restart is created in the file save.
This is a method prepared by bing. The file saving includes (a) regular dumping ... The contents of the entire file are periodically dumped to another device (for example, a magnetic tape), and history information before resuming is provided. To do. (B) Before-record and after-record ...
A pre-update record (before-record) saved to provide a record of previous history information for restarting,
The updated record (after record), which is saved to provide the record of the current history information, is saved in the journal tape etc. and combined with the regular dump. Recover files. (C) Delayed update
... Used for sequential file processing on magnetic tape, transaction history and old master tape are saved to obtain file history information corresponding to the restart time (for example, May 30, 1972) (See "Information Processing Handbook", pp.3-31 to 3-32, published by Nichi-Ohm Co., Ltd.).

On the other hand, in a distributed processing system consisting of a plurality of processing modules, when information is destroyed by a certain processing module, the file contents saved by the above (a), (b), (c), etc. The damaged update information was restored based on the history information of. In that case, the history information necessary for recovering the destroyed update information is the processing module including at least the processing device and the information storage device (memory).
It consisted only of information about Then, the first method of the update information recovery process is to execute all process modules.
There is a method to trace back to the warranty update information that guarantees the consistency and has no contradiction. That is, a method called "restore" is used to return to the last save point and restore the save file. As a second method, there is a method of restoring the warranty update information only with the processing module in which the information is destroyed. This is the reconstr
This is a method called "ust" to reconstruct the state at the time of failure occurrence.

[0004]

However, the first method of the update information restoration process described above has a problem that the update amount after the guarantee update information has to be discarded even for the normal process module. Occurs. Further, the second method has a problem that a contradiction occurs between the update information of the destruction encounter processing module and other normal processing modules. An object of the present invention is to solve these conventional problems, minimize the unrecoverable amount of update information, and restore to a normal state in which there is no contradiction among all update information, and a destruction information synchronization restoration method thereof. To provide.

[0005]

In order to achieve the above object, the distributed information processing system of the present invention comprises (a) a plurality of processing modules each including at least a processing device (2) and an information storage device (3). (1) and a single management module
In the distributed information processing system in which the module (12) and the module (12) are coupled to each other, each processing module (1) has at least a communication control means (4) for communicating with each other. An update processing means (5) for executing update processing of update information, an update information storage means (6) for storing updated information, and a request for paying out an update serial number, which is necessary for editing history information. A history information control unit (7) that obtains information from the update processing unit (5) and edits the history information, a history information storage unit (8) that stores the edited history information, and periodically, all modules. -Guarantee update information processing means (9) for acquiring the contents of update information of its own module in synchronization with each other, guarantee update information storage means (10) for storing the acquired warranty update information, and information Warranty renewal information rating when destruction occurs Means (10) reads the guaranteed update information stored in the fracture encounters processing module - update information restoration processing means for re-setting the update information Le (11)
In addition, the management module (12) is provided with a communication control means (13) for communicating with the processing module, and a request from the history information control means (7) of the processing module. An update serial number management means (14) for paying out a unique and sequential update serial number in the system, and each processing module.
The guarantee update information management means (15) for controlling the synchronization of the content acquisition processing of the update information of its own module by all the guarantee update information processing means (9) of each module, and each processing module. An update information recovery management means (16) for managing the guarantee update information is provided. Further, the method for recovering destruction information synchronization of a distributed information processing system according to the present invention comprises (b) a processing module (1) including at least a processing device (2) and an information storage device (3), and a single management module. In the distributed information processing system in which the connection device (17) and the (12) are coupled to each other, the update information identifier (22) and the update information identifier (22) each time the update information is updated in each processing module (1). Contents of each before and after update of update information (2
3 and 24), and unique and serial in the system for each update process.
A list of the update serial number (25) sequentially assigned, the module identifier in which the update information of the other processing module that is updated together with the update information and the update information identifier are listed (26, 27, 28) and are generated as an update history, the update history is saved, and when the update information and the history information being saved in a specific processing module are destroyed, they are saved in another processing module. Based on the history information that has been recorded, it is characterized in that not only the destruction encounter processing module in which the information is destroyed, but also the update information of all the other processing modules constituting the distributed information processing system are restored. There is.

[0006]

In the present invention, in the distributed information processing system composed of a plurality of modules, the history information is created by using the history information control means every time the information is updated, and the history information is stored in the history information storage means. deep. Then, when the information destruction is detected, the update information recovery management means and the update information recovery processing means update the update information of all the processing modules based on the history information stored in other normal processing modules. Minimize the unrecoverable amount of information, and restore the normal state without any inconsistency among all update information. In the present invention, since the history information includes information regarding the update serial number and related other processing modules, the unrecoverable portion can be minimized for the update information of all processing modules, and It is possible to restore a consistent state between all update information.

[0007]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 is a configuration diagram of a distributed information processing system showing an embodiment of the present invention. In FIG. 1, 1 is a processing module having a processing function and information storage function, 12 is a management module for managing the distributed information processing system, and 17 is a plurality of processing modules 1 and one management module. -Le 12
Is a coupling device for coupling. The processing module 1 is composed of a processing device 2 and an information storage device 3. A communication control unit 4, an update processing unit 5, a history information control unit 7, a warranty update information processing unit 9, and an update information restoration processing unit 11 are arranged in the processing device 2. In addition, the information storage device 3
An update information storage unit 6, a history information storage unit 8, and a guarantee update information storage unit 10 are arranged in the. Further, the management module 12 is provided with a communication control means 13, an update serial number management means 14, a guarantee update information management means 15, and an update information recovery management means 16. The configurations of the plurality of processing modules 1 are all the same, and thus the description of other components will be omitted.

FIG. 2 is a diagram showing the types and storage states of information stored in the history information storage means shown in FIG. As shown in FIG. 2, the history information storage means 8 in the information storage device 3 of the processing module 1 is a processing module to be updated.
An update identifier 21, an update information identifier 22 to be updated, pre-update contents 23 of the update information, post-update contents 24 of the update information, and an update serial number 25 are stored. 26-28
Is a set of another processing module identifier and update information identifier associated with the update information, and there are as many as necessary.
As described above, in this embodiment, since the history information includes the information about the update serial number and related other processing modules, the unrecoverable portion is minimized for the update information of all the processing modules. In addition, it is possible to restore all update information to a state in which there is no contradiction.

FIG. 3 is a diagram showing an arrangement example of update information distributed in each processing module of FIG. a1 and a2
Are not necessarily the same information. Anything that is related and updated synchronously may be used. Three
Reference numerals 1 to 39 are update information. In the processing module # 1, the update information 31, 35 and 37 are updated, and in the processing module # 2, the update information 32, 33 and 38 are updated and the processing module is updated. -Update information 34, 3 for rule # 3
6 and 39 are updated respectively. Of these, the update information (a1) 31 and the update information (a2) 32 are mutually related information, and the update information (b2) 33 and the update information (b
3) 34 is also information related to each other. Update information (c1) 3
5 and the update information (c3) 36 are the same. Further, the update information (d) 37, the update information (e) 38, and the update information (f) 39 are update information that are not related to other processing modules. These update information (for example, a1 and a
In 2), the target information processing system is divided so as to satisfy the desired conditions at the time of designing, and the deployment destination module is determined.

FIG. 4 is an explanatory diagram of an example of update processing according to the present invention. In FIG. 4, 41 is the content before update of the update information a1, 42 is the content after update of the update information a1, 43 is the content of update information c1, and 44 is the content of update information d. In addition, the left figure (1) is the state j before update, the right figure (2)
Indicates the state k after updating. As is apparent from FIGS. 4 (1) and 4 (2), these update targets are only the update information a1, and the content 43 of the update information c1 and the content 4 of the update information d.
There is no change in 4. In the process in which this distributed information processing system carries out tasks, update information (a1) 31 and update information (a1)
2) The entire set of a plurality of pieces of update information that are mutually related as in 32 needs to be updated synchronously without receiving a conflict from another processing request. When information is destroyed,
For all update information that needs to be restored, it is necessary to create the update history information shown in FIG. 2 every time the update information is updated. In order to create the update history information of FIG. 2, an example of update processing and history information generation processing will be described with reference to FIGS. 3 and 4.

In FIG. 3, when an update processing execution request for the update information (a1) 31 is issued, the update processing means 5 of the processing module # 1 is activated to execute the update processing. The execution result is stored in the update information storage means 6. In this case, the update information (a2) 32 on the processing module # 2 associated with the update information (a1) 31 also performs the update processing in the same manner in cooperation with the processing module # 1. Simultaneously with the execution of this update processing, the history information control means 7 of the processing module # 1
Requests the update serial number management means 14 of the management module 12 to pay out the update serial number. The update sequence number management means 14 issues a unique and sequential update sequence number in the system and notifies all related process modules, in this case, process modules # 1 and # 2. . The history information control means 7 of each of the processing modules # 1 and # 2, which has received this, obtains the information necessary for editing the history information shown in FIG. 2 from the update processing means 5 in the own processing module, Edit history information. When the editing process is completed, the history information storage means 8 stores the history information.

FIG. 5 is a diagram showing an example of the update sequence between the multi-processing modules in FIG. Here, FIG.
Based on the arrangement of the update information shown in (1), the update information a1 indicated by the update information identifier represents the information content at the state s0 as a1 (s0). Other update information c1
The same applies to f. In addition, the state is the state s
The time series changes from 0 to s1 ... S5. The value in the shaded area indicates that the information has been updated. 6 and 7 are diagrams of examples of history information generated in accordance with the update sequence shown in FIG. In FIGS. 6 and 7, 51 to 58 are the contents of the history information, and each information corresponds to the structure of the history information of FIG. In the state s1, history information 51 and history information 5 surrounded by a solid line
5 is processing module # 1 and processing module # 2, respectively
It has been shown to be generated and stored in. State s
In the case of No. 2, since the same is generated and saved, the history information 51 and the history information 52 are saved in the processing module # 1 and only the history information 55 is saved in the processing module # 2 at this point. Will be saved. Since the state s3 is similarly generated and stored, the history information 51, the history information 52, and the history information 53 are stored in the processing module # 1 at this point. In the same way, s4, s
Also in 5, the history information is generated and saved. Thus, the contents 51 to 58 of the history information in FIGS.
Corresponds to the shaded area in FIG. 5, and therefore FIG.
The history information content in FIG. 7 shows the content of the history information at the time when the information was updated.

An example of a recovery method when information destruction occurs will be described below with reference to FIG. Now, in the case where information destruction occurs in the state of s5 in FIG. 5, if the information destruction is only update information, the update information recovery processing means 11 updates using the history information stored in the history information storage means 8. Restore information. In this case, since the history information used for restoration is normal, all the update information can be restored normally, and no conflict occurs with the update information of other processing modules 1. Next, a recovery method when the information destruction extends not only to the update information but also to the history information will be described. If information destruction occurs in the state s5 in FIG. 5, first, periodically, for example, the state s0 in FIG.
At the appropriate time, the guarantee update information managing means 15 of the management module 12 and the guarantee update information processing means 9 of the processing module 1 synchronize with each other to obtain the contents of the update information of all modules. It is acquired and stored in the guarantee update information storage means 10 in the processing module 1 as guarantee update information. At the time of state s5 in FIG. 5, when information destruction occurs in the processing module # 1, the update information restoration processing means 11 reads the guarantee update information stored in the guarantee update information storage means 10, and encounters destruction. By resetting as the update information of the processing module, the processing module # 1 becomes the state s0 of FIG.
Will be restored (first stage).

After that, restoration processing of the update information of all processing modules based on the history information is performed. That is, the oldest update process of the update serial number is searched from the history information 51 to 58 shown in FIGS. Here, as is clear from FIG. 6, the history information 54 and 57 in the state s5 corresponds to the oldest number. Of these, the module # 2 associated with the processing module # 1 which is the destruction encounter processing module is
The pre-update content a2 (s1) is read from the history information 57 and set as the update information in the update information a2 which is the corresponding update information, whereby the module # 2 is restored to the state s4 (second step). In the same manner, the update information e is detected from the history information 56 by searching the update process of the next old number of the update serial number. However, the destruction encounter processing module
Since it is independent of the processing module # 1 which is a
No restoration is necessary (3rd stage). Hereinafter, similarly, the recovery process is repeated until the state is returned to the state s0 (fourth
Stage). As is clear from this example, the information that cannot be recovered because it is related to the information in the failure module, such as the update information a2, is returned to the content before the update so that no contradiction occurs.
Information that is not related to information destruction, such as update information e,
By leaving the update information as it is, it is possible to recover the update information having no contradiction and not to update the update information that does not need to be recovered.

In the above-mentioned method, a normal module is used as a target, and the state is returned to the past in order from the oldest number of history information. However, as another method, all modules are guaranteed update information. After returning to the normal state s0 once, there is a method of recovering the history information to the latest state by targeting only the information unrelated to the failure module # 1 in order from the smallest history information. Even with this method, completely equivalent functions can be exhibited. The above description has been given for the configuration of the distributed information processing system shown in FIG. 1, that is, the case where the processing module 1 and the management module 12 are provided separately. It will be clear that each of the means 13-16 provided will function exactly the same in a configuration that is present in a particular processing module 1. Further, in FIG. 1, the history information storage means 8 and the warranty update information storage means 10 are provided in each processing module 1, but a module other than the processing module 1, for example, a module corresponding to the management module 12 is provided. Of course, even when they are centrally deployed, they function in the same way. Furthermore, it goes without saying that even when the coupling device 17 that couples a plurality of modules is realized by a communication line or a communication network, it functions in the same way. With the above configuration,
When the update information and the history information are destroyed and the warranty update information and the update history information are used for recovery, it is possible to recover the update information that does not have a contradiction and to prevent the update information that does not need to be recovered from being recovered. .

[0016]

As described above, according to the present invention,
A module in which update sequence numbers uniquely and sequentially assigned in the system for each update process and update information of other process modules that are collectively updated with the update information exist.
Since it has a means for generating an update history including a list of pairs of update identifiers and update information identifiers and storing the update history,
It is possible to minimize the unrecoverable portion of the update information and to restore the normal state with no inconsistency among all the update information.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a distributed information processing system showing an embodiment of the present invention.

FIG. 2 is a diagram showing a history information configuration content of a history information storage unit in FIG.

FIG. 3 is a diagram showing an arrangement example of update information in a plurality of processing modules.

FIG. 4 is an explanatory diagram before and after updating in one processing module.

FIG. 5 is a diagram showing an example of a multi-processing module-to-module update sequence in FIG. 3;

6 is a part of a diagram showing an example of history information for each state in FIG.

FIG. 7 is another part of the diagram showing an example of history information for each state in FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Processing module 2 Processing device 3 Information storage device 4 Communication control means 5 Update processing means 6 Update information storage means 7 History information control means 8 History information storage means 9 Warranty update information processing means 10 Warranty update information storage means 11 Update information Recovery processing means 12 Management module 13 Communication control means 14 Update serial number management means 15 Warranty update information management means 16 Update information recovery management means 17 Coupling device 21 Update target processing module identifier 22 Update target update information identifier 23 Update Content before update of information 24 Content after update of update information 25 Update serial number 26 Other processing module identifier and update information identifier pair # 1 27 related to update information Other processing module identifier and update information related to update information Identifier set # 2 28 Other processing module identifier related to update information and update information identifier set # 3 1 update information a1 32 update information a2 33 update information b2 34 update information b3 35 update information c1 36 update information c3 37 update information d 38 update information e 39 update information f 41 update content a1 before update 42 update information a1 update Subsequent contents 43 Contents of update information c1 44 Contents of update information d 51-58 Contents of history information

Claims (2)

[Claims] 1. A distributed information processing system in which a plurality of processing modules each including at least a processing device and an information storage device and a single management module are connected to each other by a connecting device. At least a communication control means for communicating with each other, an update processing means for executing update processing of update information, an update information storage means for storing updated information, and an update History information control means for requesting the payout of the serial number, obtaining the information necessary for editing the history information from the update processing means, and editing the history information;
History information storage means for storing the edited history information, guaranteed update information processing means for periodically obtaining the contents of the update information of the own module in synchronization with all the modules, and Warranty update information storage means for storing warranty update information, and when information destruction occurs, the warranty update information stored in the warranty update information storage means is read and reset as update information for the destruction encounter processing module. An update information recovery processing means is provided, and the management module has a communication control means for communicating with the processing module, and a request from the history information control means of the processing module, whereby the management module in the system is operated. Between the update serial number management means for issuing a unique and sequential update serial number and all modules for the content acquisition processing of the update information of the own module by the guaranteed update information processing means of each processing module. Assurance update information management means for controlling the synchronization of
A distributed information processing system comprising: an update information recovery management unit that manages the guaranteed update information of the file. 2. In a distributed information processing system in which a processing module including at least a processing device and an information storage device and a single management module are connected to each other by a connecting device, each processing module is updated. Each time the information update process is executed, the identifier of the update information, the contents before and after the update of the update information, the update serial number uniquely and sequentially assigned in the system for each update process, and the update Modules that have update information for other processing modules that are updated together with information
A list of a set of the update identifier and the update information identifier is generated as an update history, the update history is saved, and the update information and the history information being saved in a specific processing module are destroyed. In addition to the destruction encounter processing module in which the information is destroyed, based on the history information stored in the other processing module, a plurality of other processing modules constituting the distributed information processing system are included. A method for recovering the destruction information synchronization of a distributed information processing system characterized by recovering all update information.