JPH07244645A - Highly reliable decentralized transaction processing system - Google Patents

Abstract

PURPOSE:To optimize the transmission destination of log data and to efficiently use plural log servers which are connected to a medium and shorten a log processing time by providing a log data transmission destination selection means which selects the transmission destination of the log data by a transaction server. CONSTITUTION:The transaction server 2 accepts a transaction process request made a transaction client 1 by a transaction request transmission and reception part 3, and a transaction processing part 4 processes the transaction request while using data in a transaction data storage part 5. The transaction processing part 4 passes the processing result history of the processing part 4 to a log gathering part 6 and also transmits the transaction processing result from the request transmission and reception part 3 to the transaction client 1. The log data that the log gathering part 6 gathers are passed to a log data transmission destination selection part 7, which determines whether the passed log data are sent to one specific transmission destination or many unspecified log servers; and the log data are sent from a log data transmission and reception part 9 to the determined log servers 8.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a highly reliable distributed transaction processing system which processes a transaction by using a plurality of computers distributed on a network.

[0002]

[Prior art]

(Conventional Example 1) As conventional transaction processing systems, Japanese Patent Laid-Open Nos. 63-34663 and 63-1 are available.
As disclosed in Japanese Patent No. 56258 and Japanese Patent Laid-Open No. 4-199339, one or more business clients that issue transaction requests, one or more business servers that process transaction requests, and one that collects transaction logs. It is composed of the above log server, a transmission medium connecting the business client, the business server, and the log server.

A business client sends a transaction request to a business server, and the business server receives the transaction request and processes the transaction request. The business server transmits the history of transaction processing to the log server and also transmits the transaction request processing result to the business client. The business client receives the processing result of the transaction request from the business server.

The log server receives the transaction processing history from the business server and stores it.

Whenever transaction processing is performed in order to improve reliability, the log data of the transaction is recorded one or more media (log database) dedicated to the log each time.
Have been collected. When a failure occurs in the business server, the log information stored in the log server is used to restore the business server.

(Conventional Example 2) In another conventional transaction processing system, in the conventional example 1, a log is stored 2
If the capacity of the next storage medium is saturated due to the accumulation of logs, no measures have been taken.
The problem of overflowing the log when the secondary storage medium becomes saturated, and even if there are multiple secondary storage media that store the logs, the secondary storage medium that statically stores the logs by each service Since the destination is determined, there is a problem that the capacity of a certain secondary storage medium becomes saturated and even if the capacity of another secondary storage medium has a free space, it cannot be used.

(Conventional Example 3) Another conventional transaction processing system is disclosed in Japanese Patent Laid-Open No. 2-90341. In this method, the configuration as shown in FIG. 14 is adopted in order to obtain the backup of the data storage unit while the transaction process is being executed without affecting the transaction as much as possible. 14
In the figure, 1401 is a transaction processing system for performing transaction processing, 1402 is a backup portion for performing backup by storing data such as magnetic tape, 1403 is a backup portion 140.
Backup acquisition unit management means for managing 2 1404
Is a transaction monitoring means for monitoring whether or not the backup unit to be acquired is subject to transaction processing. 1405 is a partial backup acquisition means for detecting a backup unit that is not accessed for backup acquisition and acquiring a partial backup. 1406. Is a backup correction means for correcting the transaction when the transaction processing is performed on the portion for which the backup is being acquired.

Next, the operation will be described. In order to perform backup processing, backup acquisition unit management means 1
403 instructs the partial backup acquisition unit 1405 to instruct to acquire a part of the database for backup, and the backup part 1402 saves and stores the part of the database on a magnetic tape or the like. At this time, the transaction monitoring unit 140 determines whether the portion being backed up is subject to transaction processing.
4 monitors, and if it is a target, the backup correction means 1406 acquires a backup for performing the correction.

(Conventional Example 4) As another conventional transaction processing system, there is "Nikkei Open System 1".
There is a replica server method described in the articles on pages 134 to 136 of the November 993 issue. FIG. 13 is a diagram showing the configuration of this system. In the conventional method, the master server 1301 has a master table 1302 and a snapshot log of the master table 1302.
The replica server 130 includes a table 1303.
Reference numeral 4 is composed of a replica table 1305.

Next, the operation will be described. The location of the replica server 1304 is registered in the master server 1301. When the master server 1301 is written to the master table 1302 in which the original memory is written, the master server 1301 sends the snapshot log, which is the log of the updated part, to the snapshot log table 1303. The master server 1301 sends a snapshot log from the snapshot log table 1303 to the registered replica server 1304. Upon receiving, the replica server 1304 updates the replica table 1305 in which the duplicate storage is written.

[0011]

Since the conventional examples 1 to 3 are configured as described above, a medium to be backed up, such as a magnetic tape, generally has low access performance.
The problem that the process related to the backup acquisition reduced the efficiency of the transaction process that is executed in the same computer system, and the problem that the backup cannot be performed in the system that cannot mount the backup target medium such as magnetic tape. was there.

[0012] In the conventional example 4, there is a problem that the replica server does not change up to what point in the master data has been updated. If the replica server leaves the network, the log to be updated cannot be obtained. There were problems that the master server had to be copied and that the location of the replica server had to be registered with the master server.

The present invention has been made to solve the above problems, and is an improved method for collecting log data at high speed in a transaction processing system in which a large number of computers are distributedly arranged on a network. Also, in a transaction processing system in which a large number of computers are distributed on a network, when a data storage unit is saved as a backup on a magnetic tape or the like, the backup can be performed without lowering the transaction processing efficiency. To provide an improved method and apparatus for performing and enabling backup to be obtained even in a system that cannot mount a magnetic tape or the like, and to what point the storage medium to be duplicated has been updated And whether the replica server is a network Away also from a when returning eliminating the need to copy from the master server, and an object thereof is to eliminate the need to register the location of the replica server to the master server.

[0014]

According to another aspect of the present invention, there is provided a highly reliable distributed transaction processing system including: a business client that issues a transaction request; a business server that processes the transaction request; and a log server that collects a transaction log. A management server that manages the status of the log server and a transmission medium that connects the business client, the log server, and the management server are provided, and the business server includes a log data destination selection unit that selects a destination of log data. To have.

A highly reliable distributed transaction processing system according to a second aspect is the highly reliable distributed transaction processing system according to the first aspect, wherein the business server comprises a log data discriminating means for discriminating a destination of the log data according to a property of the log data. To do.

A highly reliable distributed transaction processing system according to a third aspect is the highly reliable distributed transaction processing system according to the second aspect, wherein the business server selects a destination of the log data according to the size of the log data. A selection means is provided.

A highly reliable distributed transaction processing system according to a fourth aspect is the highly reliable distributed transaction processing system according to the second aspect, wherein the business server selects a destination of the log data for the transaction processing according to the priority of the transaction. A degree selecting means is provided.

A highly reliable distributed transaction processing system according to a fifth aspect is the highly reliable distributed transaction processing system according to the second aspect, wherein the business server selects a log data transmission destination according to the length of the processing queue of the log server. Data queue selecting means and log server queue selecting means for detecting the length of the processing queue of the log server are provided.

The highly reliable distributed transaction processing system according to claim 6 is the highly reliable distributed transaction processing system according to claim 2, wherein the business server detects the remaining amount of log data storage capacity of the log server. And a log data discriminating unit for each remaining capacity for selecting a destination of the log data according to the remaining amount of the log data storage capacity of the log server, wherein the management server manages the log data storage capacity of the log server. An information management unit; and a log restoration unit that restores the log data distributed to the log server, wherein the business server searches the remaining amount of the log data storage unit of the log server, and a log server remaining capacity retrieval unit. A log server for each remaining capacity for determining the destination of log data according to the remaining capacity of the log data storage means of the log server It includes a separate section.

A highly reliable distributed transaction processing system according to a seventh aspect is the highly reliable distributed transaction processing system according to the first aspect, further comprising a broadcasting communication unit for broadcasting and communicating log data to a log server.

According to another aspect of the present invention, there is provided a highly reliable distributed transaction processing system including a business client for issuing a transaction request, a business server for processing the transaction request, a log server for collecting transaction logs, and a status of the log server. In a transaction processing system including a management server that manages and a transmission medium that connects the business client, the log server, and the management server, the business server includes a backup acquisition unit that acquires the contents of a data storage unit as a backup, and a backup A backup data transmission / reception unit in the business server that transmits data to a remote backup server,
The backup server includes an internal backup server transmission / reception unit for receiving the backup data, a backup data storage unit for storing the backup data in a storage medium, and a backup medium for storing the backup data.

A highly reliable distributed transaction processing system according to claim 9 is a business client that issues a transaction request, a business server that processes the transaction request, a backup data transmission / reception unit in a backup server that receives backup data, and stores the same. A backup server including a backup data storage unit to be stored in a medium and a backup medium to store the backup data, a duplicate storage medium to be duplicated, a duplicate storage medium input / output processing unit having means for inputting and outputting to the duplicate storage medium, From a duplicate storage processing unit having means for performing processing for the duplicate storage medium, a management information input / output unit having means for performing input / output processing of update conditions and the like, and a log data transmitting / receiving unit having means for performing communication Replica management device, Log data storage unit for inputting / outputting a log, a log data storage unit input / output processing unit having means for inputting / outputting to / from the log data storage unit, a log processing unit having means for processing the log, and this log processing unit A log server including a time acquisition unit having means for acquiring time, a log data transmission / reception unit having means for communication, and a log management information input / output processing unit having means for input / output of log management information, Equipped with.

[0023]

In the highly reliable distributed transaction processing system according to claim 1, since the business server is provided with the log data destination selecting means for selecting the destination of the log data, the destination of the log data can be optimized. The log processing time can be reduced by efficiently using a plurality of log servers connected on the medium.

According to another aspect of the highly reliable distributed transaction processing system of the present invention, the business server is provided with log data discriminating means for discriminating the destination of the log data according to the characteristic of the log data, so that the log server most suitable for the characteristic of the log data. Is selected to process the log data, it is possible to efficiently use a plurality of log servers connected to the medium and reduce the log processing time.

In the highly reliable distributed transaction processing system according to the third aspect, the business server is provided with the log data discriminating means for discriminating the size of the log per transaction, so that the log data is simply allocated to the log server. Compared to the case, large log data can be allocated to a log server that processes quickly and small log data can be allocated to a log server that processes slowly. Therefore, you can efficiently use multiple log servers connected to the medium to log. Processing time can be reduced.

In the highly reliable distributed transaction processing system according to claim 4, the business server is provided with a log data priority selecting means for selecting a destination of the log data according to the priority of the transaction, so that the log data is simply transferred to the log server. Compared to the case of allocation, the log data of the transaction that needs to be processed with priority can be processed with priority, so that the processing time of the transaction can be reduced.

A highly reliable distributed transaction processing system according to claim 5 is: a log data queue selecting means for selecting a destination of log data to a business server according to a length of a processing queue of the log server; and a processing queue of the log server. By providing the log server queue selection means for detecting the length of the log server, the log data can be transmitted to the log server having a low processing load after the processing load of the log server is detected. The log processing time can be reduced by efficiently using the above log server.

A highly reliable distributed transaction processing system according to a sixth aspect of the present invention is a log information management means for managing the status of a log server located remotely or locally, and a log server remaining for searching the remaining capacity of the log server existing on the transmission path. In a system in which a plurality of log servers are arranged on a network by providing a capacity search means and a remaining capacity-based log data determination means that determines the destination of log data according to the remaining capacity of the log server existing on the transmission path , Even if the secondary storage medium that stores the log is saturated,
By switching the log storage to a secondary storage medium with sufficient remaining capacity, there is an effect of preventing the log omission,
In addition, the resources can be effectively used by using another secondary storage medium having a sufficient remaining capacity.

A highly reliable distributed transaction processing system according to a seventh aspect comprises a transaction processing means, a log collecting means, a log data discriminating means, a log data broadcasting communication means and a log data transmitting / receiving means, and a certain log data is stored in a plurality of log servers. Even if a disk or computer system that acquires the log fails, the log data with the same contents can be acquired from multiple log servers based on the data of other log servers. Since the failure recovery processing can be performed, the recovery processing can be performed without affecting the transaction processing executed on the business server.

A highly reliable distributed transaction processing system according to claim 8 comprises a backup server, backup data transmitting / receiving means, and backup data acquiring means,
Data for performing backup acquisition is transmitted from the backup data transmission / reception means of the business server to the backup server, and in the backup server, the backup data transmission / reception means receives this and the backup data acquisition means performs backup acquisition on a medium such as a magnetic tape. This makes it possible to perform backup acquisition on a computer system that is different from the business server that is executing transaction processing, and perform backup without reducing the efficiency of transaction processing, or even on systems that cannot mount magnetic tape or the like. A backup can be acquired, and a single backup can perform backup for a plurality of business servers.

In the highly reliable distributed transaction processing system according to claim 9, since the replica management device has an input means for changing the update condition and the replica management device issues an update request, the replica management device leaves the network and returns. When it is done, it is possible to update it from the storage medium without using a computer from the log. Further, the log server does not need to register the position of the replica management device.
Since the log server sends the log and the arrival time information of the log obtained by the time acquisition means to the replica management apparatus together, it is possible to know up to what point the replica apparatus has updated.

[0032]

【Example】

Example 1. Embodiment 1 of the present invention will be described below with reference to the drawings. 1 is a block diagram of the main configuration of a system according to a first embodiment of the present invention. In FIG. 1, 1 is a business client, 2 is a business server, 3 is a transaction request transmission / reception unit having transaction request transmission / reception means, 4 is a transaction processing section having transaction processing means, and 5 is a transaction data storage section having transaction data storage means. , 6 is a log collecting unit having a log collecting unit, 7 is a log data sending destination selecting unit having a log data sending destination selecting unit, 8 is a log server, 9 is a log data sending / receiving unit having a log data sending / receiving unit, 10 is a log A log data discriminating unit having a data discriminating unit, 11 a log data discriminating unit by size having a log data discriminating unit by size, 12 a log data history storage unit having a log data history storing unit,
13 is a log data storage unit having a log data storage means, 1
Reference numeral 4 is a log data priority selection unit having log data priority selection means, 15 is a log data queue selection unit having log data queue selection means, 16 is a log server queue selection unit having log server queue selection means, 17 Is the transmission line, 18
Is a log restoring unit having a log restoring unit, 19 is a log information managing unit having a log information managing unit, 20 is a management server, 21 is a log data broadcasting communication unit having a log data broadcasting communication unit,
Reference numeral 22 is a backup server, 23 is a backup acquisition unit having backup acquisition means, 24 is a backup data transmission / reception unit having backup data transmission / reception means, and 25.
Is a backup data transmission / reception unit having backup data receiving means, 26 is a backup data storage unit having backup data storage means, 27 is a backup medium,
28 is a log data input / output processing unit having a log data input / output processing unit, 29 is a processing unit having a log processing unit, 30 is a time acquisition unit having a time acquisition unit, and 31 is a log management having a log management information input / output unit. An information input / output unit, 32 is a replica management device, 33 is a duplicate storage medium, 34 is a duplicate storage medium input / output processing unit having a duplicate storage medium input / output processing unit, 35 is a duplicate storage processing unit having a duplicate storage processing unit, and 36. Is a management information input / output unit having a management information input / output unit, 37 is a log server remaining capacity searching unit having a log server remaining capacity searching unit, and 38 is a remaining capacity type log server judging unit having a remaining capacity type log server judging unit. is there.

Next, the operation of the first embodiment will be described with reference to FIG. Business client 1, business server 2, log server 8
Communicate via the transmission line 17. Business server 2
Receives the transaction processing request issued by the business client 1 at the transaction request transmission / reception unit 3 (step 301), while using the data in the transaction data storage unit 5 at the transaction processing unit 4,
Process the transaction request (step 302).
The transaction processing unit 4 passes the processing result history of the transaction processing unit 4 to the log collection unit 6 (step 303), and also transmits the transaction processing result from the transaction request transmission / reception unit 3 to the business client 1 (step 314). The log data collected by the log collection unit 6 is passed to the log data destination selection unit 7, and whether the log data is transmitted to one specific destination (log server) or to an unspecified number of log servers. Is determined (step 304) and sent from the log data transmitting / receiving unit 9 to the determined log server 8 (step 313).

According to the first embodiment, by providing the log data delivery destination selection unit for selecting the delivery destination of the log data, the delivery destination of the log data can be optimized. The log server can be efficiently used to reduce the log processing time.

Example 2. A second embodiment of the present invention will be described with reference to FIG. Business client 1 and business server 2,
The log server 8 communicates via the transmission line 17. The business server 2 receives the transaction processing request issued by the business client 1 at the transaction request transmission / reception unit 3 (step 301), and the transaction processing unit 4 processes the transaction request while using the data in the transaction data storage unit 5. (Step 302). The transaction processing unit 4 passes the processing result history of the transaction processing unit 4 to the log collection unit 6 (step 303), and at the same time, receives the transaction processing result from the transaction request transmission / reception unit 3 in the business client 1
(Step 314). The log data collected by the log collection unit 6 is passed to the log data destination selection unit 7 (step 304), and the log data determination unit 10 uses the attributes of the log data as criteria for determining the destination of the log data (log server). Is determined (step 306) and sent from the log data transmitting / receiving unit 9 to the determined log server 8 (step 313).

According to the second embodiment, the business server is provided with the log data discriminating section for discriminating the destination of the log data depending on the characteristic of the log data, so that the log server optimal for the characteristic of the log data is selected and the log is recorded. Since the data is processed, the log processing time can be reduced by efficiently using the plurality of log servers connected to the medium.

Example 3. A third embodiment of the present invention will be described with reference to FIG. The business client 1, the business server 2, and the log server 8 communicate with each other via a transmission line 17. The business server 2 receives the transaction processing request issued by the business client 1 at the transaction request transmission / reception unit 3 (step 401), and the transaction processing unit 4 processes the transaction request while using the data in the transaction data storage unit 5. (Step 4
02). The transaction processing unit 4 passes the processing result history of the transaction processing unit 4 to the log collection unit 6 (step 403). The log data collected by the log collection unit 6 is
It is passed to the size-based log data determination unit 11 in the log data determination unit 10 of the log data destination selection unit 7, and the size-based log data determination unit 11 detects the size of the log per transaction (step 404). If the size of the log data is large, the destination of the log data is determined to be a log server having a high processing speed (step 405). If the size of the log data is small, the destination of the log data is determined to be a log server having a low processing speed (step 405). Step 406). The log data is sent from the log data transmitting / receiving unit 9 to the determined log server 8 (step 407). Transaction request transmitter / receiver 3
Sends the transaction processing result to the business client 1 (step 408). The log server 8 stores the log data collected by the business server 2, the log data transceiver 9 receives the log data sent from the business server 2, and the log data history storage 12 stores the log data. The log data is received from the transmission / reception unit 9, the log data is assigned an index number, and the log data generation time and the index number are stored. Log data storage unit 13
Receives and stores the log data from the log data history storage unit 12. As described above, since the log server for processing log data is selected according to the size of the log data by providing the log data discriminating unit 11 for each size in the business server 2, it is possible to select a plurality of log servers having different log data processing speeds. It is possible to reduce the delay of the business transaction processing due to the log data processing and to improve the processing speed of the business server as compared with the case of processing the log data evenly.

Example 4. A fourth embodiment of the present invention will be described with reference to FIG. The business client 1, the business server 2, and the log server 8 communicate with each other via a transmission line 17. The business server 2 receives the transaction processing request issued by the business client 1 at the transaction request transmission / reception unit 3 (step 501), and the transaction processing unit 4 processes the transaction request while using the data in the transaction data storage unit 5. (Step 5
02). The transaction processing unit 4 passes the processing result history of the transaction processing unit 4 to the log collection unit 6 (step 503). The log data collected by the log collection unit 6 is
It is passed to the log data destination selecting unit 7, and the priority level of the log data discriminating unit 14 of the log data discriminating unit 10 detects the processing priority of the transaction (step 504). The log data is also processed with a high priority (step 505), and if the transaction has a low priority, the log data is also processed with a low priority (step 506). It is sent from the log data transmitting / receiving unit 9 to the determined log server 8 (step 50).
7). The transaction request transmission / reception unit 3 transmits the transaction processing result to the business client 1 (step 508). As described above, by providing the log data discriminating unit 14 by priority in the business server 2, the processing priority of the log data is selected according to the priority of the transaction, so that when processing a transaction with a high priority, The processing of log data does not become a bottleneck, and transaction processing speed can be improved.

Example 5. A fifth embodiment of the present invention will be described with reference to FIG. The business client, the business server, and the log server communicate with each other via a transmission line. Business server 2
Receives the transaction processing request issued by the business client 1 at the transaction request transmission / reception unit 3 (step 601), while using the data in the transaction data storage unit 5 at the transaction processing unit 4,
Process the transaction request (step 602).
The transaction processing unit 4 passes the processing result history of the transaction processing unit 4 to the log collection unit 6 (step 60).
3). The log server queue search unit 16 monitors the processing queue of the log server 8 (step 604), and when the log data is generated in the business server 2, the log server having the shortest queue is the log data queue selection unit. 15 is notified (step 605). The log data collected by the log collection unit 6 is passed to the log data destination selection unit 7, and the log data queue selection unit 15 of the log data determination unit 10 logs using the queue length of the log server as a determination factor. The data transmission destination (log server) is determined, and the data is transmitted from the log data transmission / reception unit 9 to the determined log server 8 (step 606). The transaction request transmission / reception unit 3 transmits the transaction processing result to the business client 1 (step 607). In this way, by providing the log data queue selecting unit 15 and the log server queue searching unit 16 in the business server 2, the queue of the log server does not become a bottleneck when processing log data.
The processing speed of log data can be improved.

Example 6. Embodiment 6 of this invention is shown in FIGS.
Will be described. The log server 8 includes a log data transmission / reception unit 9, a log data history storage unit 12, and a log data storage unit 13. The management server 20 includes a log restoration unit 18, a log data transmission / reception unit 9, and a log information management unit 19. The log restoration unit 18 and the log information management unit 19 may be installed in the business server 2. The business server 2, 3 is a transaction request transmission / reception unit, 4 is a transaction processing unit, 5 is a transaction data storage unit, 6 is a log collection unit, 9 is a log data transmission / reception unit, 37 is a log server remaining capacity search unit, and 38 is the front. It is composed of a log server discriminating unit by capacity. The log information management unit 19 in the management server 20 determines the availability of the log data storage units 13 and 13A connected to the transmission line 17, the log server 8,
The position of 8a is stored. Each log data storage unit 13, 13
The status of the free space of A may be notified from the log servers 8 and 8A to the log information management unit 19 on a regular basis, or may be confirmed by the log information management unit 19 to the log servers 8 and 8a. Good. The log data restoration unit 18 is used to restore the log data having the same attribute distributed in the plurality of log data storage units 13 and 13a by switching the log server.

Next, the operation will be described. The business client 1 requests the business server 2 for transaction processing. The business server 2 transmits the progress of the processing as log data to the log server 8 by the log data transmitting / receiving unit 9. The log server 8 receives the log data by the log data transmission / reception unit 9, writes the log data generation time and the log number in the log data history storage unit 12, and records the contents of the log data in the log data storage unit 13. Now, it is assumed that the free space in the log data storage unit 13 is running low (step 701). The log server 8 uses the fact that the free space is low as the business server 2 using it.
(Step 702). The business server 2 makes an inquiry from the log server remaining capacity searching unit 37 to the management server 20 in order to search for another log server having a sufficient capacity (step 703). It is assumed that the log information management unit 19 selects the log server A8a having the largest free space in the secondary storage from among the log servers in its own network management. The log information management unit 19 notifies the business server 2 that the log server A8a has been selected (step 704). The business server 2 that has received the notification transmits the log data to the log server 8 and the log server A8a at the same time until the log data discriminating unit 38 for remaining capacity is ready to accept the log data of the log server A8a (step 705). . When the log server A8a is ready to accept log data,
The business server 2 is notified that the preparation for acceptance has been completed, and the log information management unit 19 of the management server 20 is notified (step 706). The log information management unit 19 records it (step 707). The business server 2 that has received the notification receives the remaining capacity-based log data determination unit 3
At step 8, the transmission of the log data to the log server 8 is stopped, and the log data is transmitted only to the log server A8a (step 7).
08). Next, the restoration of the distributed log data will be described. The operation of restoring the distributed log data fragments in series is performed by the log data restoration unit 18. The log data restoration unit 18 can obtain the information of which log server log should be combined at that time point by referring to the switching preparation completion time point of the log information management unit 19. Based on this information, the log data restoration unit 18 can restore the distributed log data fragments in series.
According to the sixth embodiment, by providing the log information management unit remotely or locally, it is possible to prevent the log disk from falling into a saturated state, so that there is an effect of improving the reliability of the system, and a sufficient capacity is available. By switching to another log disk, the resource can be effectively used.

Example 7. Embodiment 7 of the present invention will be described with reference to FIG. FIG. 1 shows a case in which a certain log data is broadcast to and communicated with a plurality of log servers, and the logs having the same contents are acquired by the plurality of log servers, so that a failure occurs in a disk or a computer system acquiring the logs. Also shows that the failure recovery process can be performed based on the data of another log server. In the figure, the business server 2 is the transaction processing unit 4, Log collection unit 6, log data determination unit 1
0, the log data transmitting / receiving unit 9. The log data discriminating unit 10 includes a log data broadcast communication unit 21. The log server 8 includes a log data transmission / reception unit 9 and a log processing unit 29.
And a log data input / output processing unit 28 and a log data storage unit 13. The log server A8a includes a log data transmission / reception unit 9a, a log processing unit 29a, a log data input / output processing unit 28a, and a log data storage unit 13a. FIG. 8 shows the operation of the log data broadcast communication unit 21 in the log data discrimination unit 10. The transaction processing unit 4
While performing the transaction processing requested by the business client, the processing result history is passed to the log collection unit 6 as log data and requested to collect it. Log collection unit 6
Requests the log data determination unit 10 to make a determination according to the contents of the log data and transmit the log data to an appropriate log server (step 801). The log data broadcast communication unit 21 in the log data determination unit 10 analyzes the contents of the log data given from the log collection unit 6, and determines the group of log servers to which the log data should be transmitted from the transaction identifier or the like. (Step 80
2). Here, it is assumed that the group including the log server 8 and the log server A8a is selected. Next, the log server 8
Then, the log data transmission / reception unit 9 is requested to perform broadcast communication of the log data on the transmission path with the group address including the address of the log server A8a on the network as the destination (step 803). The log data transmitting / receiving unit 9 broadcasts the log data on the transmission line 17 by using the given group address of the log server group as a destination. The log server 8 and the log server A 8a receive this log data by the respective log data transmitting / receiving units 9 and 9a, and the log processing unit 2
9 and 29a are log data input / output processing units 28 and 2
It is stored as a log in the log data storage units 13 and 13a via 9a. Next, restoration at the time of failure will be described. If some failure occurs during the transaction processing of the business server 2, the transaction processing unit 4
Asks the log data broadcast communication unit 21 to tell that the log servers collecting the logs related to the transaction are the log server 8 and the log server A8a, and to send the log data for restoration to either of them. Request. Transactions can be restored even if the system or disk of any log server fails. According to the seventh embodiment, the contents to be acquired in the log are broadcast and communicated as data on the communication path, and are received by a plurality of systems at the same time and acquired in the log, thereby improving reliability and possibility of the entire system. It becomes possible.

Example 8. Although the log server belonging to a certain group address is determined for each transaction in the above embodiment, it may be fixed for each business service in advance. Also, it may be transmitted to all log servers by using broadcast communication, and the log server may select and store only necessary items in the log data storage unit.

Example 9. Embodiment 9 of this invention is shown in FIGS.
0 will be described. FIG. 1 is a block diagram of an embodiment showing that backup acquisition can be performed by a computer system different from a business server executing transaction processing. In FIG. 1, 17 is a transmission line,
Reference numeral 2 is a business server that performs transaction processing requested by a business client, 22 is a backup server as a computer that performs centralized backup, and 4 is a transaction processing unit that executes transaction processing.
5 is a transaction data storage unit that is accessed for execution of transaction processing, 23 is a backup acquisition unit that acquires the contents of the transaction data storage unit as a backup, and 24 is a backup that transmits and receives backup data on the transmission path 17 within the business server 2. A data transmission / reception unit, 25 is a backup data transmission / reception unit that transmits / receives backup data to / from the transmission path 17 in the backup server 22, 26 is a backup data storage unit,
A backup medium 27 stores the contents of backup data.

Next, the operation will be described with reference to FIGS. 9 and 10. FIG. 9 shows the operation of the backup acquisition unit 23 in the business server 2, and FIG. 10 shows the operation of the backup storage unit 26 in the backup server 22. In FIG. 9, the backup acquisition unit 23 sequentially sequentially cuts out the contents of the transaction data storage unit 5 (step 901) and configures the contents as backup data (step 902). A request for transmission to the backup server 22 is made (step 903). When the backup completion data is received from the backup server 22 (step 904), it is determined whether or not all the contents of the transaction data storage unit 5 have been cut out (step 905), and if not, the transaction data storage unit is further extracted. The contents of the next part of 5 are cut out and configured as backup data, and the backup data transmitting / receiving unit 24 is requested to transmit to the backup server 22. This operation is repeated until the entire contents of the transaction data storage unit 5 are cut out. In FIG. 10, the backup data storage unit 26 receives the backup data received from the backup data transmission / reception unit 25 (step 1001), stores the contents in the backup medium 27 (step 1002), and performs the operation on the backup data transmission / reception unit 25. Request transmission of backup completion data to the server 2 (step 100
3). If this data is the last backup data, the backup is completed. If not, waits for the next backup data (step 100).
4). This operation is repeated until the final backup data is received.

According to the ninth embodiment, the backup server, the backup data transmission / reception unit, and the backup data acquisition unit are provided, and the data for performing the backup acquisition is transmitted from the backup data transmission / reception unit of the business server to the backup server for backup. In the server, the backup data transmission / reception unit receives this and the backup data acquisition unit performs backup acquisition on a medium such as a magnetic tape. Therefore, backup acquisition can be performed on a computer system different from the business server executing the transaction processing. This makes it possible to perform backups on a backup server without reducing the efficiency of transaction processing on the business server, and to make backups even on systems that cannot mount magnetic tapes. A backup regarding business server, there is an effect that the load of the backup management is reduced because it can be performed centrally by a single backup server.

Example 10. In the above-described embodiment, the backup completion data is transmitted from the backup server 22 to the business server 2 every time one backup data is stored. However, in order to reduce the communication load, each backup data storage is performed. It is also possible to send the backup completion data to.

Example 11. Further, in the above embodiment, the backup acquisition unit 23 independently cuts out the backup data from the data storage unit, and the backup data storage unit 2
6 was requested to store the data, but a method in which the backup data storage unit 26 voluntarily requests the backup acquisition unit 23 to transmit the backup data is also conceivable.

Example 12 It is an invention aimed at updating duplicate storage using a log. In FIG. 1, 2
Is a business server. 22 is a backup server. Reference numeral 8 is a log server. Inside thereof, 13 is a log data storage section for inputting / outputting logs, 28 is a log data storage section input / output processing section having means for inputting / outputting to / from the log data storage section 13, and 29 is means for performing processing on logs. A log processing unit, 30 is a time acquisition unit having means for acquiring time for the log processing unit 29, 9 is a log data transmitting / receiving unit having means for performing communication, and 31 is means for inputting / outputting information of log management. This is a log management information input / output processing unit. Also, 3
Reference numeral 2 is a replica management device. Inside thereof, 33 is a duplicate storage medium, 34 is a duplicate storage medium input / output processing section having means for inputting / outputting to / from the duplicate storage medium 33, 35 is a duplicate storage processing section having means for performing processing for duplicate storage, and 36. Is a management information input / output unit having means for inputting / outputting update conditions and the like, and 9 is a log data transmitting / receiving unit having means for performing communication. Next, the operation of the twelfth embodiment will be described with reference to FIGS.
Will be described with reference to. 11 is a flowchart showing the operation of the log server 8, and FIG. 12 is a flowchart showing the operation of the replica management device 32. The operation of the log server 8 and the operation of the replica management device 32 will be described in this order. The operation of the log server 8 will be described. Initialization, operation when received from the business server 2, replica management apparatus 3
2 will be described in order of the operation when received. First, the initialization of the log server 8 will be described (step 1102).
In the log server 8, the log processing unit 29 uses the log data input / output processing unit 28 to obtain the log start time from the log data storage unit 13. If there is no log start time, the log processing unit 29 acquires the current time from the time acquisition unit 30,
The log data input / output processing unit 28 is used to write to the log data storage unit 13. The operation when the log server 8 receives a log from the business server 2 will be described (step 11).
03, 1107). Logs shall be sent in the order in which they occur. When the log data transmitting / receiving unit 9 receives the log, the log server 8 passes it to the log processing unit 29. Log processing unit 2
9 acquires the time from the time acquisition unit 30 and adds the time information to the log information. Then, the log processing unit 29 passes the information to the log data input / output processing unit 28, and if writing is possible, the log data input / output processing unit 28 writes the information in the log data storage unit 13. If the log cannot be written, the log processing unit 29
Receives from the log management information input / output unit 31 how much the log should be erased, and erases the log in the order of oldness. At this time, the time added to the newest log erased is set as the log start time, and the log processing unit 29 causes the log data input / output processing unit 2 to operate.
8 is used to write in the log data storage unit 13. The operation when the log server 8 receives from the replica management device 32 will be described. When the log data transmission / reception unit 9 of the log server 8 receives the replica data from the replica management apparatus 32, the log data transmission / reception unit 9 transfers the log data to the log processing unit 29 (step 1103). If the last update time of the duplicate storage medium 33, which is the content of the received data, is before the log start time (step 1104), the content to restore the backup is transmitted to the replica body management device 32 (step 1105). If the last update time of the duplicate storage medium 33 is after the log start time (step 11
04), the log processing unit 29 is the log data input / output processing unit 28.
Is used to acquire the log data after the last update time of the duplicate storage medium 33. In addition, the log processing unit 29 is the time acquisition unit 3
Use 0 to get the current time. Then, the log data obtained by using the log data transmitting / receiving unit 9 and the obtained current time are transmitted to the replica management device 32 (step 110).
6).

Next, the operation of the replica management device 32 will be described. The initialization and update operations will be described in this order. First, the initialization of the replica management device will be described (step 1).
202). In the replica management apparatus 32, the duplicate storage processing unit 35 uses the duplicate storage medium input / output processing unit 34 to retrieve the position information of the log server 8, the position information of the backup server 22, and the final update of the duplicate storage medium 33 from the duplicate storage medium 33. Get time and update condition. If there is no such information, the duplicate storage processing unit 35 uses the management information input / output unit 36 to obtain the information, and uses the duplicate storage medium input / output processing unit 34 to write to the duplicate storage medium 33 to send / receive log data. Restore is performed from the backup server 22 using the unit 9 and the duplicate storage medium input / output processing unit 34. Then, at this time, the time when the backup is started is obtained and set as the last update time and written in the duplicate storage medium 33. Next, the operation when the replica management device 32 updates will be described. When the copy storage processing unit 35 satisfies the update condition (step 1203), it uses the log data transmitting / receiving unit 9 to send an update request to the log server 8 with the data of the last update time (step 1204). . The log data transmission / reception unit 9 of the replica management device 32 receives the reply from the log server (step 1205). If the content is to list the backup (step 1206), the duplicate storage processing unit 35 uses the log data transmitting / receiving unit 9 and the duplicate storage medium input / output processing unit 34 to restore from the backup server 22. At this time, the time at which the backup operation is started is obtained and set as the final update time and written in the duplicate storage medium 33 (step 1107). If the content is not to restore the backup (step 1206), the replica management apparatus updates the plurality of storage media 33 by using the duplicate storage medium input / output processing unit 34 with the sent log data and the data of the final time. (Step 1108).

By the way, the management information input / output unit 36 changes the update condition when it receives the instruction to change the update condition.
Pass the information to 5. The information is written in the duplicate storage medium 33 by using the duplicate storage medium input / output processing unit 34. Also,
Since the last update time is written in the duplicate storage medium 33, the final update time can be obtained from the duplicate storage medium 33 by using the management information input / output unit 36, the duplicate storage processing unit 35, and the duplicate storage medium input / output processing unit 34. it can.

As described above, in the twelfth embodiment, by using the management information input / output unit 34, it can be seen that up to what point the duplicate storage medium 33 has been updated. Further, since the replica management device 32 requests the log server 8 to update and updates it in the log, when the replica management device 32 leaves the network and returns, it is not necessary to copy from the transaction data storage unit 6 from the beginning. Further, it is no longer necessary to register the position of the duplicate storage medium 33 in the transaction data storage unit 5 or the log server 8.

Example 13 As an example, the disk of a portable computer can be updated. That is, after connecting to a computer network, writing data to a disc, and disconnecting from the computer network and used independently, it is possible to connect to the computer network and update only the changed data to the disc.

Example 14. Further, in the above-mentioned embodiment, it is assumed that the logs are transmitted in order, but if the log data itself has a description of the data indicating the time when the log itself occurs, the logs may not be transmitted in order. In the above-mentioned embodiment, the log server 8 can also log a plurality of disks. At this time, the log server 8 and the replica management device 32 need to know the information about which storage medium.

[0055]

According to the highly reliable distributed transaction processing system of claim 1, a business client that issues a transaction request, a business server that processes the transaction request, a log server that collects transaction logs, and a log server A configuration is provided that includes a management server that manages a state, and a transmission medium that connects the business client, the log server, and the management server, and the business server includes a log data destination selection unit that selects a destination of log data. Therefore, the destination of the log data can be optimized, and the log processing time can be reduced by efficiently using the plurality of log servers connected to the medium.

A highly reliable distributed transaction processing system according to a second aspect is the highly reliable distributed transaction processing system according to the first aspect, wherein the business server comprises a log data discriminating means for discriminating a destination of the log data according to a property of the log data. Since it is configured to process, log data that is most suitable for the nature of the log data can be selected to process the log data, and multiple log servers connected to the medium can be used efficiently to reduce the log processing time. it can.

A highly reliable distributed transaction processing system according to claim 3 is the highly reliable distributed transaction processing system according to claim 2, wherein the business server selects a destination of the log data according to the size of the log data. Since it is configured to have a selection means,
This has the effect of reducing the delay in business transaction processing due to log processing and improving the processing speed of the business server.

A highly reliable distributed transaction processing system according to a fourth aspect is the highly reliable distributed transaction processing system according to the second aspect, wherein the business server selects a destination of the log data for the transaction processing according to the priority of the transaction. Since the configuration is provided with the degree selecting means, the transaction log data can be preferentially processed as compared with the case where the log data is simply allocated to the log server. Therefore, the transaction processing can be performed. You can save time.

A highly reliable distributed transaction processing system according to a fifth aspect is the highly reliable distributed transaction processing system according to the second aspect, wherein the business server selects a log data destination according to the length of the processing queue of the log server. Since the data queue selecting means and the log server queue selecting means for detecting the length of the processing queue of the log server are provided, the log having a low processing load after the processing load of the log server is detected. Log data can be sent to the server, and multiple log servers connected on the medium can be efficiently used to reduce log processing time.

A high-reliability distributed transaction processing system according to a sixth aspect is the high-reliability distributed transaction processing system according to the second aspect, wherein the business server detects the remaining amount of log data storage capacity of the log server, and a log data remaining capacity searching means. And a log data discriminating unit for each remaining capacity for selecting a destination of the log data according to the remaining amount of the log data storage capacity of the log server, wherein the management server manages the log data storage capacity of the log server. An information management unit; and a log restoration unit that restores the log data distributed to the log server, wherein the business server searches the remaining amount of the log data storage unit of the log server, and a log server remaining capacity retrieval unit. A log server for each remaining capacity for determining the destination of log data according to the remaining capacity of the log data storage means of the log server Since it is configured to include another part, it is possible to switch to another log disk that has the effect of improving the system reliability and prevents the log disk from falling into a saturated state and has sufficient capacity. Due to
There is an effect that resources can be used effectively.

The high-reliability distributed transaction processing system according to claim 7 is the high-reliability distributed transaction processing system according to claim 1, further comprising a broadcasting communication unit for broadcasting communication of log data to a log server. Is broadcasted as data on the communication path, and simultaneously received by a plurality of systems and acquired in a log, the reliability and availability of the entire system can be enhanced.

According to another aspect of the present invention, there is provided a highly reliable distributed transaction processing system including a business client for issuing a transaction request, a business server for processing the transaction request, a log server for collecting transaction logs, and a status of the log server. In a transaction processing system including a management server that manages and a transmission medium that connects the business client, the log server, and the management server, the business server includes a backup acquisition unit that acquires the contents of a data storage unit as a backup, and a backup A backup data transmission / reception unit in the business server that transmits data to a remote backup server,
And a backup data storage unit for storing the backup data in the backup server, a backup data storage unit for storing the backup data in a storage medium, and a backup medium for storing the backup data. Therefore, the backup data transmission / reception unit of the business server transmits the data for backup acquisition to the backup server. In the backup server, the backup data transmission / reception unit receives the data and the backup data acquisition unit receives a medium such as a magnetic tape. Since backups are acquired on the computer, backups can be acquired on a computer system that is different from the business server that is executing transaction processing, and the efficiency of transaction processing on the business server can be reduced. Backup server, backup can be acquired even in a system that cannot be equipped with magnetic tape, etc. Also, backup of multiple business servers can be centrally performed by one backup server. This has the effect of reducing the load on the.

A highly reliable distributed transaction processing system according to claim 9 is a business client that issues a transaction request, a business server that processes the transaction request, a backup server internal backup data transmission / reception unit that receives backup data, and stores it. A backup server including a backup data storage unit to be stored in a medium and a backup medium to store the backup data, a duplicate storage medium to be duplicated, a duplicate storage medium input / output processing unit having means for inputting and outputting to the duplicate storage medium, From a duplicate storage processing unit having means for performing processing for the duplicate storage medium, a management information input / output unit having means for performing input / output processing of update conditions and the like, and a log data transmitting / receiving unit having means for performing communication Replica management device, Log data storage unit for inputting / outputting a log, a log data storage unit input / output processing unit having means for inputting / outputting to / from the log data storage unit, a log processing unit having means for processing the log, and this log processing unit A log server including a time acquisition unit having means for acquiring time, a log data transmission / reception unit having means for communication, and a log management information input / output processing unit having means for input / output of log management information, Since it is equipped with a log server and a replica management apparatus, it is possible to update the duplicate storage medium and to know when the state of the duplicate storage medium is the current state of the duplicate storage medium.

[Brief description of drawings]

FIG. 1 is a block diagram of a first embodiment of a highly reliable distributed transaction processing system according to the present invention.

FIG. 2 is an explanatory diagram of Embodiment 6 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 3 is a flowchart of Embodiments 1 and 2 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 4 is a flowchart diagram of Embodiment 3 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 5 is a flow chart of Embodiment 4 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 6 is a flow chart of Embodiment 5 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 7 is a flowchart diagram of Embodiment 6 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 8 is a flowchart diagram of Embodiment 7 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 9 is a flowchart diagram of Embodiment 9 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 10 is a flowchart diagram of Embodiment 9 of the highly reliable distributed transaction processing system according to the present invention.

FIG. 11 is a flowchart diagram of a twelfth embodiment of the highly reliable distributed transaction processing system according to the present invention.

FIG. 12 is a flow chart diagram of a twelfth embodiment of the highly reliable distributed transaction processing system according to the present invention.

FIG. 13 is a block diagram of a transaction processing system of Conventional Example 4.

FIG. 14 is a block diagram of a transaction processing system of Conventional Example 3.

[Explanation of symbols]

 1 Business Client 2 Business Server 3 Transaction Request Transmitter / Receiver 4 Transaction Processor 5 Transaction Data Storage 6 Log Collection 7 Log Data Destination Selection 8 Log Server 8a Log Server A 9 Log Data Transmitter 9a Log Transmitter A 10 Log Data discriminating unit 11 Size-dependent log data discriminating unit 12 Log data history storing unit 12a Log data history storing unit 13 Log data storing unit 13a Log data storing unit A 14 Log data priority sorting unit 15 Log data queue sorting unit 16 Log server Queue selection unit 17 Transmission unit 18 Log restoration unit 19 Log information management unit 20 Management server 21 Log data broadcasting communication unit 22 Backup server 23 Backup acquisition unit 24 Backup data transmission unit 25 Backup data reception unit 26 Backup Updater storage unit 27 Backup medium 28 Log data input / output processing unit 28a Log data input / output processing unit A 29 Log processing unit 29a Log processing unit A 30 Time acquisition unit 30a Time acquisition unit A 31 Log management information input / output processing unit 31a Log management Information input / output processing unit A 32 Copy storage management device 33 Storage medium 34 Storage medium input / output processing unit 35 Copy storage processing unit 36 Management information input / output unit 37 Log server remaining capacity search unit 38 Remaining capacity-based log server determination unit 1301 Master server 1302 Master table 1303 Snapshot log table 1304 Replica server 1305 Replica table 1401 Transaction processing system 1402 Backup part 1403 Backup acquisition unit management part 1404 Transaction monitoring part 1405 Minute backup acquisition unit 1406 backup correction unit

Front Page Continuation (72) Inventor Tetsuo Nakagawaji 1-1-1 Ofuna, Kamakura-shi Information Systems Research Center, Mitsubishi Electric Corporation

Claims (9)

[Claims] 1. A business client that issues a transaction request, a business server that processes the transaction request, a log server that collects transaction logs, a management server that manages the status of this log server, and the business client, A transmission medium that connects a log server and a management server, and the business server includes log data destination selection means for selecting a destination of log data. 2. The highly reliable distributed transaction processing system according to claim 1, wherein the business server comprises log data discriminating means for discriminating the destination of the log data according to the nature of the log data. 3. The highly reliable distributed transaction processing system according to claim 2, wherein the business server comprises a log data size selection means for selecting the destination of the log data depending on the size of the log data. 4. The highly reliable distributed transaction processing system according to claim 2, wherein the business server comprises log data priority selection means for selecting a destination of log data for transaction processing according to the priority of the transaction. 5. The business server comprises a log data queue selecting means for selecting a destination of log data according to the length of a processing queue of the log server, and a log server queue for detecting the length of the processing queue of the log server. 3. The highly reliable distributed transaction processing system according to claim 2, further comprising: selecting means. 6. The business server comprises a log data remaining capacity searching means for detecting the remaining capacity of the log data storage capacity of the log server,
Log data management means for managing the log data storage capacity of the log server, the log data determining means for selecting a destination of the log data according to the remaining capacity of the log data storage capacity of the log server. A log server remaining capacity searching means for searching the remaining capacity of the log data storage means of the log server, and a log restoring means for restoring the log data distributed to the log server, The highly reliable distributed transaction processing system according to claim 2, further comprising: a remaining capacity-based log server determination unit that determines a destination of the log data according to the remaining capacity of the log data storage unit of the log server. 7. The highly reliable distributed transaction processing system according to claim 1, further comprising a broadcasting communication unit for broadcasting communication of log data to a log server. 8. A business client that issues a transaction request, a business server that processes the transaction request, a log server that collects transaction logs, a management server that manages the status of this log server, and the business client, In a transaction processing system including a transmission medium connecting a log server and a management server, the business server is a backup acquisition unit that acquires the contents of a data storage unit as a backup, and a business that transmits backup data to a remote backup server. Backup data transmission / reception unit in the server,
The backup server includes a backup data transmission / reception unit in the backup server for receiving the backup data, a backup data storage unit for storing the backup data in a storage medium, and a backup medium for storing the backup data. A highly reliable distributed transaction processing system. 9. A business client that issues a transaction request, a business server that processes the transaction request, a backup server internal backup data transmission / reception unit that receives backup data, a backup data storage unit that stores the backup data in a storage medium, and the aforementioned A backup server including a backup medium for storing backup data, a duplicate storage medium to be duplicated, a duplicate storage medium input / output processing unit having means for inputting and outputting to the duplicate storage medium, and processing for the duplicate storage medium A replica management device including a duplicate storage processing unit having means, a management information input / output unit having means for performing input / output processing such as update conditions, and a log data transmitting / receiving unit having means for communication, and log input / output. Log data storage area A log data storage unit input / output processing unit having means for inputting / outputting to / from the storage unit, a log processing unit having means for performing processing on the log, a time acquisition unit having means for acquiring time of the log processing unit, and communication. A highly reliable distributed transaction processing system, comprising: a log server including a log data transmission / reception unit having a means for performing and a log management information input / output processing unit having a means for inputting / outputting information of log management.