JP2013037384A - Data distribution system, management server, client device, management server program, and client device program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To decrease a frequency of monitoring of data servers by a management server for reducing a load of the management server, but properly perform data writing of a client without increasing failure.SOLUTION: A data distribution system comprises: a plurality of database units 11, 12 and 13; a management server 2 which monitors operation states of respective database units, creates information about the operation states of the respective database units, and discloses the information in a browsable manner; and a client 3 which selects the writable database unit 13 on the basis of the information about the operation states of the respective database units, and when data cannot be written in the selected database unit 13 in spite of the information about the operation states of the respective database units in a step S13, causes the management server 2 to update the operation state of the selected database unit 13 from a writable state to a non-writable state in a step S14.

Description

  The present invention relates to data writing and operation status monitoring in a data distribution system.

  With the development of ICT (Information and Communication Technology) technology, the amount of data distributed has increased explosively. Distributed storage and distributed processing systems are used to process these.

  In the distributed processing system, each of a plurality of servers having the same function is assigned to execute processing, so that the scale of the system can be expanded by increasing the number of servers. In other words, the distributed processing system becomes a scale-out system.

  When data writing from clients is distributed, a plurality of data servers share and accept data writing. There are various methods of sharing, but there are a method (Non-Patent Document 1) that allocates data based on a hash value of data ID (IDentification), a method that randomly selects (Non-Patent Document 2), and the like.

  In any of the methods, a mechanism based on the assumption that a data server failure occurs is often adopted, and for this purpose, the management server performs alive monitoring of the data server. In addition, the data server to be written to is selected from the currently operating ones according to the rules exemplified in Non-Patent Document 1 and Non-Patent Document 2.

  If the client passes data to the management server and substitutes for writing, the management server will be responsible for data input and output in addition to access from the client, resulting in an extremely high load on the management server. The system will not scale out.

  Therefore, when the client acquires data server operating status information from the management server and the client writes directly to the data server, the management server is not responsible for data input / output, reducing the load on the management server and scaling out the system. (Non-Patent Document 2).

Kazuyuki Shudo, “Scale-Out Technology”, Information Processing, Vol. 50, no. 11, November 2009 Hiroshi Sato, Takeshi Inoue, Takahiro Yamazaki, Yohei Katayama, Keiji Shimizu, Keiichi Koyanagi, “Distributed sampling storage for sensor data and its sample size control method”, FIT2010, D-022, pp. 228 141-142, 2010

  Here, the operation status information (internal list) of the data server managed by the management server within itself is updated every time the operation status changes. However, the information on the operational status of the data server (public list) that the management server discloses to the client is not updated every time the operational status changes, but based on the internal list every time the expiration date has passed. Updated. Therefore, the internal list and the public list do not always match.

  The operating state of the data server may change before the expiration date of the public list expires. Therefore, there is a possibility that the data server that is supposed to be alive is dead at the time of writing, in which case the client fails to write. If the frequency of alive monitoring is increased in order to keep this possibility low, the load on the management server will also increase.

  Depending on the data server selection method, it may be possible to reselect a separate data server when writing fails. However, for example, when random selection of data servers is required, and it is important to select all live data servers with equal probability, this re-selection causes a bias. Re-selection is not desirable because uniformity is lost. In addition, when the selection of the data server is uniquely determined, for example, when the data server is allocated based on the hash value of the data ID in an environment where the data server is not made redundant, it is impossible in principle to select again.

  Although it is possible to distribute the management servers and scale them out, it is preferable to reduce the number of servers even if they are concentrated or distributed due to cost problems. In any case, a technique for reducing the load on the management server is important.

  Therefore, in order to solve the above-described problem, the present invention reduces the load on the management server by reducing the frequency of monitoring the data server by the management server without increasing the number of client write failures. The purpose is to eliminate re-selection.

  In order to achieve the above object, the client device can write the operating status of the selected data server to the management server when data cannot be written to the selected data server, contrary to the operating status information of each data server. It was decided to update it from being able to write.

  Specifically, the present invention provides a plurality of data servers, a management server that monitors the operating status of each data server, creates information on the operating status of each data server, and makes it available for viewing, and each data server The information on the operating state of the data server is acquired from the management server, based on the information on the operating state of each data server, a writable data server is selected, and contrary to the information on the operating state of each data server, A data distribution system comprising: a client device that causes the management server to update the operation state of the selected data server from writable to non-writable when data cannot be written to the selected data server.

  Further, the present invention provides an operation state monitoring unit that monitors the operation states of a plurality of data servers, an operation state creation unit that creates and displays information on the operation states of each data server, and a client device. Obtain data server operating state information, select a data server that is writable based on the operating state information of each data server, and select the selected data server against the operating state information of each data server. An operation state update unit that updates the operation state of the data server selected by the client device from writable to non-writable based on a notification to that effect when data cannot be written to the data server. Management server.

  Further, the present invention is based on an operating state acquisition unit that acquires information on the operating state of each data server from a management server that monitors the operating states of a plurality of data servers, and information on the operating state of each data server, A data server selection unit for selecting a writable data server, and the data server selection unit when data cannot be written to the data server selected by the data server selection unit, contrary to information on the operating state of each data server. And a non-writable notification unit that causes the management server to update the operating state of the data server selected by the management server from writable to non-writable.

  Further, the present invention provides an operation state monitoring procedure for monitoring the operation states of a plurality of data servers, an operation state creation procedure for creating information on the operation states of the respective data servers and making them publicly viewable, and a client device. Obtain data server operating state information, select a data server that is writable based on the operating state information of each data server, and select the selected data server against the operating state information of each data server. When data cannot be written to the data server, an operating state update procedure for updating the operating state of the data server selected by the client device from writable to non-writable based on the notification to that effect is sequentially executed on the computer. Is a management server program.

  Further, the present invention is based on an operating state acquisition procedure for acquiring information on the operating state of each data server from the management server that monitors the operating states of a plurality of data servers, and information on the operating state of each data server, A data server selection procedure for selecting a writable data server, and when the data server selection procedure cannot write data to the data server selected by the data server selection procedure, the data server selection procedure. Is a client device program for causing a computer to sequentially execute a write disable notification procedure for causing the management server to update the operation state of the selected data server from write enabled to write disabled.

  According to this configuration, in order to reduce the load on the management server, even if the frequency of monitoring of the data server by the management server is reduced, the client monitors the operating status of the data server together with the management server. Can not increase. Furthermore, since the frequency of updating the information on the operating state of the data server can be increased while reducing the frequency of monitoring of the data server by the management server, it is possible to reduce client write failures.

  In order to achieve the above object, the proxy server accepts data writing on behalf of the selected data server, and the management server is selected for the proxy server when confirming the operational recovery of the selected data server. The data server was made to write back data.

  Specifically, the present invention provides a plurality of data servers, a management server that monitors the operating status of each data server, creates information on the operating status of each data server, and makes it available for viewing, and each data server A client device that obtains information on the operating state of the data server from the management server, selects a writable data server based on the information on the operating state of each data server, and writes data to the selected data server; When the client device cannot write the data to the selected data server against the information on the operating state of each data server, the client device accepts the writing of the data instead of the selected data server, and the client device When the data server selected by is restored to the writable state, the client device selects A proxy server to the data server writes back the data, a data distribution system comprising: a.

  Further, the present invention provides an operation state monitoring unit that monitors the operation states of a plurality of data servers, an operation state creation unit that creates and displays information on the operation states of each data server, and a client device. Obtain data server operating state information, select a data server that is writable based on the operating state information of each data server, and select the selected data server against the operating state information of each data server. When it is not possible to write data to the data server, when confirming the recovery of the operation of the data server selected by the client device, the client device sends the data to the proxy server that accepts writing of the data instead of the data server selected by the client device. The device notifies the data server selected for recovery, and the client device selects the data selected. A management server, characterized in that it comprises a working recovery notification unit for writing back the data to the server.

  Further, the present invention is based on an operating state acquisition unit that acquires information on the operating state of each data server from a management server that monitors the operating states of a plurality of data servers, and information on the operating state of each data server, A data server selection unit for selecting a writable data server, and the data server selection unit when data cannot be written to the data server selected by the data server selection unit, contrary to information on the operating state of each data server. And a data writing unit that writes the data to a proxy server that accepts the writing of the data instead of the data server selected by the client device.

  Further, the present invention provides an operation state monitoring procedure for monitoring the operation states of a plurality of data servers, an operation state creation procedure for creating information on the operation states of the respective data servers and making them publicly viewable, and a client device. Obtain data server operating state information, select a data server that is writable based on the operating state information of each data server, and select the selected data server against the operating state information of each data server. When it is not possible to write data to the data server, when confirming the recovery of the operation of the data server selected by the client device, the client device sends the data to the proxy server that accepts writing of the data instead of the data server selected by the client device. Notification of recovery of operation of the data server selected by the device, and the client device selected Running a recovery notification procedure the data to write back to Tasaba, to the computer is a management server program to be executed sequentially.

  Further, the present invention is based on an operating state acquisition procedure for acquiring information on the operating state of each data server from the management server that monitors the operating states of a plurality of data servers, and information on the operating state of each data server, A data server selection procedure for selecting a writable data server, and when the data server selection procedure cannot write data to the data server selected by the data server selection procedure, the data server selection procedure. Is a client device program for causing a computer to sequentially execute a data writing procedure for writing the data to a proxy server that receives the writing of the data instead of the data server selected.

  According to this configuration, in order to reduce the load on the management server, even if the frequency of monitoring the data server by the management server is lowered, the data is written back to the original server after the data is saved to the proxy server. Reselection of data server can be eliminated.

  According to the present invention, in order to reduce the load on the management server, even if the frequency of monitoring of the data server by the management server is reduced, it is possible to eliminate reselection of a separate data server without increasing client write failures.

It is a block diagram which shows the structure of a data distribution system. It is a block diagram which shows the process of writing data in a data distribution system. It is a sequence which shows the process of the monitoring of the operating state in a data distribution system. It is a flowchart which shows a process of a client. It is a block diagram which shows the process of data write-back in a data distribution system. It is a sequence which shows the process of data write-back in a data distribution system. It is a figure which shows the 1st data structure of a proxy server. It is a figure which shows the 2nd data structure of a proxy server.

  Embodiments of the present invention will be described with reference to the accompanying drawings. The embodiments described below are examples of the present invention, and the present invention is not limited to the following embodiments. In the present specification and drawings, the same reference numerals denote the same components.

(Outline of data distribution system configuration)
A block diagram showing the configuration of the data distribution system is shown in FIG. The data distribution system includes a distributed database 1, a management server 2, a client 3, and a proxy server 4. The distributed database 1 includes a plurality of database units 11, 12, and 13.

  The management server 2 includes an alive monitoring unit 21, an operation state management unit 22, an operation state list disclosure unit 23, a report reception unit 24, and a database control unit 25.

  The life and death monitoring unit 21 monitors the operating states of the plurality of database units 11, 12, and 13 as an operating state monitoring unit. As the operation state creation unit, the operation state management unit 22 creates information on the operation states of the database units 11, 12, and 13, and the operation state list disclosure unit 23 displays the operation state information of the database units 11, 12, and 13. Make information available for viewing.

  The report accepting unit 24 is one of the clients 3 that can acquire information on the operating state of each of the database units 11, 12, and 13 and write based on the operating state information of each of the database units 11, 12, and 13. When the database unit is selected and data cannot be written to any of the selected database units, contrary to the information on the operating state of each of the database units 11, 12, and 13, a notification to that effect is received. As the operation state update unit, the operation state management unit 22 updates the operation state of any database unit selected by the client 3 from writable to non-writable based on the notification to that effect.

  When the database control unit 25 confirms the operation recovery of any database unit selected by the client 3 as the operation recovery notification unit, the proxy server 4 accepts data writing instead of any database unit selected by the client 3. In response to this, the recovery of the operation of any database unit selected by the client 3 is notified, and the data is written back to any database unit selected by the client 3.

  The client 3 includes a management server corresponding unit 31, an operation state list management unit 32, a server selection unit 33, and a data input / output unit 34.

  The management server corresponding unit 31 acquires information on the operating states of the database units 11, 12, and 13 from the management server 2 that monitors the operating states of the plurality of database units 11, 12, and 13 as an operating state acquisition unit. The operating state list management unit 32 holds information on the operating states of the database units 11, 12, and 13. The server selection unit 33 selects, as a data server selection unit, one of the database units that can be written on the basis of the operating state information of each of the database units 11, 12, and 13. The data input / output unit 34 writes data to any writable database unit as a data writing unit.

  When the management server corresponding unit 31 cannot write data to any of the database units selected by the server selection unit 33 against the information on the operating state of each of the database units 11, 12, and 13, as a non-writable notification unit, The management server 2 is made to update the operation state of any database unit selected by the server selection unit 33 from writable to non-writable.

  The data input / output unit 34 writes data to the proxy server 4 that accepts data writing instead of the database unit selected by the server selection unit 33 as a data writing unit.

  The proxy server 4 accepts data writing in place of any of the database units selected by the client 3, and includes a data input / output unit 41, a database unit 42, a management server corresponding unit 43, and a table management unit 44.

  The data input / output unit 41 inputs data from the client 3 when any of the database units selected by the client 3 becomes unwritable, and when any of the database units selected by the client 3 becomes writable The data is output to the database part. The database unit 42 stores data from the client 3.

  When one of the database units selected by the client 3 becomes writable, the management server corresponding unit 43 receives a notification from the database control unit 25 of the management server 2. The management server corresponding unit 43 notifies the report receiving unit 24 of the management server 2 to that effect when writing back to any database unit selected by the client 3 is completed. The table management unit 44 manages the data output by the client 3 in an appropriate area of the database unit 42 depending on which database unit the client 3 has selected.

  Each apparatus described in FIG. 1 can be realized by a computer and a program, and the program can be recorded on a recording medium or provided through a network.

  As described in the problem to be solved by the invention, if the operating state of the data server changes before the expiration date of the public list elapses, the database section that is said to be active writes Sometimes dead.

  However, as shown in FIG. 1, the client 3 selects the management server 2 when data cannot be written to any of the selected database units, contrary to the operational status information of the database units 11, 12, and 13. The operating state of any database unit is updated from writable to non-writable. That is, since the client 3 monitors the operating state of the database units 11, 12, and 13 together with the management server 2, it is possible to reduce the writing failure of the client 3 without increasing the load on the management server 2.

  Then, as shown in FIG. 1, the proxy server 4 accepts data writing instead of any selected database unit, and the management server 2 confirms the operational recovery of any selected database unit. Data is written back to one of the database units selected for the proxy server 4. That is, since the data is written back to the original server after the data is saved in the proxy server, it is possible to eliminate the need to select a separate data server.

(Process when data writing is successful)
Next, processing when data writing is successful will be described. FIG. 2 is a block diagram showing data write processing in the data distribution system.

  The management server 2 transmits a ping to each database unit 11, 12, 13 and monitors the operating state of each database unit 11, 12, 13. Then, the management server 2 publishes a disclosure list P indicating information on the operating state of each database unit 11, 12, 13. Here, {1, 2, 3} in the public list P indicates that data can be written in the database parts 11, 12, and 13 marked DB1, DB2, and DB3. The client 3 requests the public list P from the management server 2 and acquires the public list P from the management server 2. Then, the client 3 confirms that data can be written in the database units 11, 12, and 13 based on the public list P, and writes the data D in the database unit 13.

  FIG. 3 shows a sequence showing an operation state monitoring process in the data distribution system. The operating state management unit 22 holds two operating state lists, one is an internal list I and the other is a public list P.

  The life and death monitoring unit 21 periodically checks the operating state of the database units 11, 12, and 13. There are three types of operating states: “Normal”, “Preparing”, and “Abnormal”. “Normal” is operating normally and writing from the client 3 is accepted, “Preparing” is operating normally but writing from the client 3 is not accepted, and “Abnormal” is It represents a state in which it is not operating normally and writing from the client 3 cannot be accepted.

  The alive monitoring unit 21 throws a ping to the database units 11, 12, and 13. If it returns correctly, the database section that has been “normal” until now is determined to be “normal”, and the database section that has been “abnormal” is determined to be “in preparation”. However, if it does not return correctly, the database part that has been “abnormal” until now continues to be determined as “abnormal”, and the database part that has been “normal” until now is determined to be “abnormal”. The operating state management unit 22 updates the internal list I. This check is performed every 5 minutes, for example.

  Further, as described above, the client 3 may report an abnormality in the database unit. In this case, the report reception unit 24 receives the report, and the operation state management unit 22 updates the corresponding operation state of the internal list I to “abnormal”.

  Further, as described above, the proxy server 4 may report the recovery of the database unit. In this case, the report receiving unit 24 receives the report, and the operation state management unit 22 updates the corresponding operation state of the internal list I to “normal”.

  The operating state management unit 22 creates a list of database units in the “normal” state in the internal list I as a public list P every five minutes, for example. The operating state list publishing unit 23 publishes the publishing list P with an expiration date of, for example, 5 minutes later.

  In FIG. 3, in the initial state, the internal list I and the public list P1 indicate {1, 2, 3}. First, the operating state management unit 22 updates the internal list I from {1, 2, 3} to {1, 2} upon receiving an abnormality report from the database unit 13, and creates the internal list I1. To do. However, the operating state list disclosing unit 23 does not update the disclosing list P1 because the expiration date has not elapsed. Next, since the expiration date has passed, the operating state list publicizing unit 23 updates the public list P1 from {1, 2, 3} to {1, 2} based on the internal list I1. Publish the public list P2. Hereinafter, the internal lists I2, I3, I4, and I5 and the public lists P3, P4, and P5 are created and released in the same manner as the internal list I1 and the public lists P1 and P2.

  As described above, the operating state lists are not always the same for internal use and for public use. Here, the reason why the management server 2 publishes not the internal list I but the public list P to the client 3 will be described. The management server 2 needs to know from which set of database parts the client 3 has selected the database part, and at least needs to know from which set of database parts the client 3 has selected the database part There is. This is because the management server 2 performs statistical processing later based on how many distributed databases 1 that are a set of database units 13 are selected.

  First, consider a case where the management server 2 discloses the internal list I to the client 3. Since the internal list I is updated each time the operating state changes, the operating state list held by each client 3 at the same time may be different. Therefore, it becomes a burden for the management server 2 to grasp at which point the database unit is selected based on the internal list I updated by the client 3.

  Next, consider a case where the management server 2 publishes the public list P to the client 3. Since the public list P is updated after the expiration date, the operating state lists held by the clients 3 are guaranteed to be the same at the same time. Therefore, it becomes easy for the management server 2 to grasp that the client 3 has selected the database unit based on the public list P updated when the expiration date has expired.

  Therefore, the management server 2 publishes not the internal list I but the public list P to the client 3. However, there is a possibility that the operating state of the data server will change before the expiration date of the public list elapses, and there is a possibility that the database part which is supposed to be alive is dead at the time of writing. Therefore, the present invention is applied.

  FIG. 4 is a flowchart showing client processing. Data is generated in the client 3 (step S1). The operating state list management unit 32 checks whether the operating state list is within the expiration date or whether the operating state list is held (step S2).

  A case will be described in which the operating state list management unit 32 confirms that the operating state list is within the expiration date (YES in step S2). The server selection unit 33 selects a database unit based on the operating state list (step S5). The data input / output unit 34 attempts to write data in the database unit (step S6).

  A case will be described where the operating state list management unit 32 confirms that the operating state list is not within the expiration date or that the operating state list is not held (NO in step S2). The management server corresponding unit 31 acquires an operation state list from the management server 2 (step S3). The operating state list management unit 32 holds the operating state list (step S4). Steps S5 and S6 are executed in the same manner as described above.

  When data input / output unit 34 has been able to write data to the database unit (YES in step S7), the series of processing ends. When the data input / output unit 34 cannot write data to the database unit (NO in step S7), the management server corresponding unit 31 reports to that effect to the management server 2 (step S8), and the data input / output unit 34 Writes data into the proxy server 4 (step S9). Steps S8 and S9 will be described in detail in the item of processing when data writing fails.

(Processing when data writing fails)
Next, processing when data writing fails will be described. FIG. 5 is a block diagram showing data write-back processing in the data distribution system.

  The client 3 confirms that data can be written in the database units 11, 12, and 13 based on the public list P, and tries to write data D into the database unit 13. However, the client 3 cannot write the data D into the database unit 13 against the public list P. Therefore, the client 3 notifies the management server 2 that writing to the database unit 13 is impossible. Furthermore, the client 3 writes data D associated with the identifier ID of the database unit 13 in the proxy server 4.

  The management server 2 updates the internal list I. That is, in the internal list I, {1, 2, 3} is updated to {1, 2}. The proxy server 4 stores data D. The management server 2 transmits a ping to the database unit 13 and monitors the operating state of the database unit 13. When the management server 2 confirms that the data can be written in the database unit 13, the management server 2 notifies the proxy server 4 to that effect. When the proxy server 4 receives the notification to that effect, the proxy server 4 writes the data D back to the database unit 13. Client 3 is not involved in writing back.

  FIG. 6 shows a sequence showing data write-back processing in the data distribution system. The management server corresponding unit 31 requests the operating state list from the operating state list disclosing unit 23 (step S11). The operating state list publishing unit 23 publishes the operating state list to the management server corresponding unit 31 (step S12). At this time, the internal list I11 indicates {1, 2, 3}. Thereafter, the database unit 13 is in an abnormal state. However, the internal list I11 still shows {1, 2, 3}.

  The data input / output unit 34 attempts to write data to the database unit 13, but cannot write data to the database unit 13 (step S13). The management server corresponding unit 31 reports the fact to the report receiving unit 24 (step S14). The operating state management unit 22 updates the internal list I11 from {1, 2, 3} to {1, 2}, and creates the internal list I12. The report receiving unit 24 notifies the management server corresponding unit 31 of the address of the proxy server 4 (step S15), but the operating state list disclosure unit 23 may disclose the address of the proxy server 4 to the management server corresponding unit 31. it can. The data input / output unit 34 writes the data associated with the identifier ID of the database unit 13 in the proxy server 4 (step S16).

  The operating state management unit 22 indicates that the alive monitoring unit 21 has received a ping from the database unit 13 that is “abnormal” in the internal list I12 or that the operator has restored the database unit 13. Is directly registered in the operation state management unit 22, the operation state of the database unit 13 in the internal list I12 is updated to “in preparation” (step S17). The database control unit 25 notifies the management server corresponding unit 43 that the operation state of the database unit 13 is “in preparation” (step S18).

  The table management unit 44 selects a table associated with the database unit 13, reads all the data from the database unit 42, and writes it back to the database unit 13 through the data input / output unit 41 (step S19). Thereafter, the table management unit 44 deletes the table linked to the database unit 13 from the table management unit 44 and the database unit 42. Then, the management server corresponding unit 43 reports the completion of the process to the report receiving unit 24 (Step S20). The operating state management unit 22 receives the report of processing end from the proxy server 4 and determines that the database unit 13 has been restored, and updates the internal list I12 from {1,2} to {1,2,3}. Then, the internal list I13 is created.

  If an abnormality occurs in the database unit 13 during the write-back, the data input / output unit 41 stops the write-back process, the table management unit 44 does not delete the data, and the management server corresponding unit 43 uses the report reception unit. The operation state management unit 22 returns the operation state of the database unit 13 to “abnormal” in the internal list I, and the alive monitoring unit 21 continues monitoring.

  Further, the abnormality of the database unit 13 is noticed not by the report from the client 3 but by the check of the life and death monitoring unit 21, and after that, the database unit 13 has never reported the abnormality of the database unit 13 from the client 3. May be restored. In this case, it is possible to determine that the proxy server 4 has no data associated with the database unit 13. Therefore, when the life and death monitoring unit 21 notices that the database unit 13 is restored, the operating state management unit 22 sets the operating state of the database unit 13 in the internal list I to “normal” without executing the above-described write-back process. It may be possible to return to "."

  A diagram showing the first data structure of the proxy server is shown in FIG. Data associated with the database units 11, 12, and 13 is managed by a collective table. There is a column for identifying the database part, and it is necessary to filter all data with the identifier of the database part before reading and deleting data, but the table management part 44 is not necessary.

  FIG. 8 shows a second data structure of the proxy server. Data associated with the database units 11, 12, and 13 are managed by separate tables. A table is managed for each database unit, and the table management unit 44 is necessary, but a process of filtering all data with the identifier of the database unit before reading and deleting data is not necessary. It is also possible to make a database separate from the table.

  The data distribution system, the management server, the client device, the management server program, and the client device program according to the present invention can reduce the load on the management server and the bias in writing regardless of the explosive increase in the amount of data to be distributed. Can be applied.

1: distributed database 2: management server 3: client 4: proxy servers 11, 12, 13: database unit 21: alive monitoring unit 22: operation state management unit 23: operation state list disclosure unit 24: report reception unit 25: database control Unit 31: Management server corresponding unit 32: Operating state list management unit 33: Server selection unit 34: Data input / output unit 41: Data input / output unit 42: Database unit 43: Management server corresponding unit 44: Table management unit I: For internal use List P: Public list D: Data ID: Identifier

Claims (10)

Multiple data servers,
A management server that monitors the operational status of each data server, creates information on the operational status of each data server and publishes it for viewing;
Information on the operating state of each data server is acquired from the management server, a writable data server is selected based on the information on the operating state of each data server, and the information on the operating state of each data server is selected. On the other hand, when data cannot be written to the selected data server, the client device causes the management server to update the operating state of the selected data server from writable to non-writable.
A data distribution system comprising: Multiple data servers,
A management server that monitors the operational status of each data server, creates information on the operational status of each data server and publishes it for viewing;
A client that acquires information on the operating state of each data server from the management server, selects a writable data server based on the information on the operating state of each data server, and writes data to the selected data server Equipment,
When the client device cannot write the data to the selected data server against the information on the operating state of each data server, the client device accepts the writing of the data instead of the selected data server, and the client device A proxy server that writes back the data to the data server selected by the client device when the data server selected by the device recovers to be writable;
A data distribution system comprising: An operating state monitoring unit for monitoring the operating states of a plurality of data servers;
An operational state creation unit that creates information on the operational state of each data server and publishes it for viewing;
The client device obtains information on the operating state of each data server, selects a writable data server based on the information on the operating state of each data server, and sets the information on the operating state of each data server. On the other hand, when data cannot be written to the selected data server, an operation state update unit that updates the operation state of the data server selected by the client device from writable to non-writable based on the notification to that effect ,
A management server comprising: An operating state monitoring unit for monitoring the operating states of a plurality of data servers;
An operational state creation unit that creates information on the operational state of each data server and publishes it for viewing;
The client device obtains information on the operating state of each data server, selects a writable data server based on the information on the operating state of each data server, and sets the information on the operating state of each data server. On the other hand, when data cannot be written to the selected data server, the proxy server accepts writing of the data instead of the data server selected by the client device when the operation recovery of the data server selected by the client device is confirmed. An operation recovery notifying unit for notifying operation recovery of the data server selected by the client device, and writing back the data to the data server selected by the client device;
A management server comprising: An operating state acquisition unit that acquires information on the operating state of each data server from a management server that monitors the operating states of a plurality of data servers;
A data server selection unit that selects a data server that is writable based on information on the operating state of each data server;
When data cannot be written to the data server selected by the data server selection unit against the information on the operation state of each data server, the operation state of the data server selected by the data server selection unit is changed from writable to non-writable. And a non-writable notification unit that causes the management server to update,
A client device comprising: An operating state acquisition unit that acquires information on the operating state of each data server from a management server that monitors the operating states of a plurality of data servers;
A data server selection unit that selects a data server that is writable based on information on the operating state of each data server;
A proxy server that accepts the writing of the data instead of the data server selected by the data server selection unit when data cannot be written to the data server selected by the data server selection unit, contrary to the operating state information of each data server A data writing unit for writing the data to
A client device comprising: An operational status monitoring procedure for monitoring the operational status of multiple data servers;
Operational state creation procedure for creating operational state information for each data server and making it available for viewing;
The client device obtains information on the operating state of each data server, selects a writable data server based on the information on the operating state of each data server, and sets the information on the operating state of each data server. On the other hand, when data cannot be written to the selected data server, an operating state update procedure for updating the operating state of the data server selected by the client device from writable to non-writable based on the notification to that effect; ,
Server program for causing computers to execute in order. An operational status monitoring procedure for monitoring the operational status of multiple data servers;
Operational state creation procedure for creating operational state information for each data server and making it available for viewing;
The client device obtains information on the operating state of each data server, selects a writable data server based on the information on the operating state of each data server, and sets the information on the operating state of each data server. On the other hand, when data cannot be written to the selected data server, the proxy server accepts writing of the data instead of the data server selected by the client device when the operation recovery of the data server selected by the client device is confirmed. An operation recovery notification procedure for notifying operation recovery of the data server selected by the client device and causing the data server selected by the client device to write back the data;
Server program for causing computers to execute in order. An operation status acquisition procedure for acquiring information on the operation status of each data server from a management server that monitors the operation status of a plurality of data servers;
A data server selection procedure for selecting a writable data server based on the operating state information of each data server;
When data cannot be written to the data server selected by the data server selection procedure, contrary to the operation status information of each data server, the operation status of the data server selected by the data server selection procedure is changed from writable to non-writable. And a non-writable notification procedure for updating the management server;
Client device program for causing a computer to execute the programs in order. An operation status acquisition procedure for acquiring information on the operation status of each data server from a management server that monitors the operation status of a plurality of data servers;
A data server selection procedure for selecting a writable data server based on the operating state information of each data server;
A proxy server that accepts writing of data instead of the data server selected by the data server selection procedure when the data server selection procedure cannot write data to the data server selected by the data server selection procedure, contrary to the operating state information of each data server A data writing procedure for writing the data to
Client device program for causing a computer to execute the programs in order.

Images