JP2015176203A - database system, database server, database server program, database client, and database client program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten processing time in the case of performing data update accompanied by commitment from a database client to a plurality of database servers.SOLUTION: In a database system including a plurality of database servers and a database client, the database servers other than the database server in the last stage are characterized to transfer a data update request supplied from the device in the previous stage to the database server in the following stage, and the database client is characterized to transmit the data update request to the database server in the foremost stage, and to receive all the data update results of the plurality of database servers from the database server in the foremost stage.

Description

  The present invention relates to a database system, a database server, a database server program, a database client, and a database client program, and can be applied to, for example, a database server / client system capable of updating data by two-phase commit.

  In the data update in the conventional database distributed transaction environment, in order to maintain the consistency of data between the distributed database servers, the database server (hereinafter also referred to as “DB server”) and the database are used in the data update. Data may be updated with a client (hereinafter also referred to as “DB client”) according to a two-phase commit (two-phase commitment) protocol. In a two-phase commit in a database distributed transaction environment, a two-phase commit is possible only when all DB servers are in a commitable state, and the committed data is officially released by completing the two-phase commit transaction. Can be used as simple data (see Patent Document 1).

  Further, in an environment that conventionally employs middleware such as ORACLE (registered trademark), when performing a two-phase commit from a DB client to two DB servers, a transaction from the DB client to the first DB server, Transaction processing to the second DB server via the DB client first DB server is required. Specifically, a context (DB update management information) used by the DB client when updating data is generated, and the first DB server and the second DB server are sequentially updated using the context. Thus, the first DB server recognizes that the transaction is a two-phase commit, and at the time of a commit request for the corresponding context, the commit request is transmitted from the first DB server to the second DB server. It has become. That is, when two-phase commit is performed from a DB client to a plurality of DB servers, multistage connection to a plurality of DB server groups is required.

JP 2009-122935 A

  However, as described above, when multi-phase connection is performed when performing a two-phase commit from a DB client to a plurality of DB servers, the number of sequence steps required for data update increases, and the two-phase commit starts. It takes a long time to finish. Also, communication between DB servers may take a long time depending on the network environment.

  Therefore, there is a demand for a database system, a database server, a database server program, a database client, and a database client program that can reduce processing time when performing data update with commit from a database client to a plurality of database servers. .

  The first aspect of the present invention is a database system including a plurality of database servers and a database client. (1) Among the plurality of database servers, the database server excluding the last stage database server is (1-1) a preceding apparatus. (1-2) data update request transfer means for transferring the data update request supplied from the preceding apparatus to the subsequent database server; 1-3) When there is a data update result response to the data update request from the subsequent database server, the data update that transmits the data update result that is a combination of the data update result and the data update result of the own device to the previous device Responding to the data commit request supplied from the response means and (1-4) the preceding apparatus. Data commit means for performing data commit, (1-5) data commit request transfer means for transferring the data commit request supplied from the preceding apparatus to the subsequent database server, and (1-6) the subsequent database server. And a data commit response means for transmitting a data commit result obtained by combining the data commit result and the data commit result of the own device to the preceding device when there is a response of the data commit result to the data commit request from ( 2) The last-stage database server includes (2-1) data update response means for performing data update in accordance with the data update request supplied from the previous-stage database server and responding to the previous-stage database server with the data update result. (2-2) Data communication supplied from the previous database server Data commit responding means for performing data commit in accordance with the data request and responding with the data commit result to the preceding database server. (3) The database client is (3-1) Of the plurality of database servers A data update request means for transmitting a data update request to the front-stage database server and receiving all data update results of the plurality of database servers from the front-stage database server; (3-2) the front-stage database server A data commit request means for transmitting a data commit request to the first database server and receiving all data commit results of the plurality of database servers from the first database server.

  The database server according to the second aspect of the present invention includes (1) data update means for updating data in accordance with a data update request supplied from the preceding apparatus, and (2) a data update request supplied from the preceding apparatus. Data update request transfer means for transferring to the database server, and (3) if a data update result response to the data update request is received from the subsequent database server, the data update result and the data update result of the own device are combined. Data update response means for transmitting the update result to the preceding apparatus, (4) data commit means for performing data commit according to the data commit request supplied from the preceding apparatus, and (5) supplied from the preceding apparatus. A data commit request transfer means for transferring a data commit request to a subsequent database server; and (6) subsequent data. A data commit response means for transmitting a data commit result, which is a combination of the data commit result and the data commit result of its own device, to the previous device when there is a response from the data server to the data commit request. Features.

  The database server program according to the third aspect of the present invention includes: (1) data updating means for updating data in accordance with a data update request supplied from the preceding apparatus; and (2) data updating supplied from the preceding apparatus. A data update request transfer means for transferring the request to the subsequent database server; and (3) if there is a response of the data update result to the data update request from the subsequent database server, the data update result and the data update result of the own device A data update response means for transmitting a data update result combining the above to the preceding apparatus, (4) a data commit means for performing data commit in accordance with a data commit request supplied from the preceding apparatus, and (5) a preceding apparatus. Data commit request transfer that forwards the data commit request supplied from the server to the subsequent database server (6) When there is a response of the data commit result to the data commit request from the database server in the subsequent stage, the data commit result in which the data commit result and the data commit result of the own apparatus are combined is transmitted to the preceding apparatus. It functions as a data commit response means.

  The database client of the fourth aspect of the present invention (1) transmits a data update request to the first database server among the plurality of database servers, and all the data of the plurality of database servers is transmitted from the first database server. Data update request means for receiving update results; (2) Sending a data commit request to the first database server and receiving all data commit results of the plurality of database servers from the first database server. And a data commit requesting means.

  According to a fifth aspect of the present invention, the database client program sends a data update request to the first database server among the plurality of database servers (1), and the plurality of database servers are transmitted from the first database server. (2) a data commit request is transmitted to the foremost database server, and all data of the plurality of database servers are transmitted from the foremost database server. It is characterized by functioning as a data commit request means for receiving a commit result.

  According to the present invention, processing time can be shortened when data update accompanied by commit is performed from a database client to a plurality of database servers.

It is the sequence diagram shown about operation | movement of DB system which concerns on 1st Embodiment. It is the block diagram shown about the whole structure of the DB system which concerns on 1st Embodiment. It is explanatory drawing shown about the structural example of the message (message of a data update request and a data commit request) produced | generated with the DB client which concerns on 1st Embodiment. It is the sequence diagram shown about operation | movement when a DB client produces | generates a message for each DB server tentatively in the DB system which concerns on 1st Embodiment. It is the sequence diagram shown about operation | movement of DB system which concerns on 2nd Embodiment.

(A) First Embodiment Hereinafter, a first embodiment of a database system, a database server, a database server program, a database client, and a database client program according to the present invention will be described in detail with reference to the drawings.

(A-1) Configuration of the First Embodiment FIG. 2 is a block diagram showing the overall configuration of the DB system 1 of the first embodiment and the functional configuration of each device.

  The DB system 1 includes two DB servers 10 (10-1, 10-2) and a DB client 20. Note that the number of DB servers 10 constituting the DB system 1 may be three or more.

  Functionally, each DB server 10 includes a DB server process 11 that performs information processing as a DB server and a DB file system 12 that stores data (database). The DB server 10 can be realized, for example, by installing a database server application (including the database server program of the embodiment) in a computer (for example, a server machine such as a PC or a workstation). Further, the DB server 10 may be realized by using a function of database middleware (for example, database middleware such as ORACLE). In the following, the identifiers on the databases (identifiers on the application) of the DB servers 10-1 and 10-2 are represented as “DB # 1” and “DB # 2”, respectively.

  Here, it is assumed that the DB servers 10-1 and 10-2 constitute a redundant system, and each DB file system 12 stores synchronized data. The data of the DB servers 10-1 and 10-2 can be updated by a two-phase commit (two-phase commitment) method (protocol) in response to a request from the DB client 20. In the DB system 1, the data update method is not limited to the two-phase commit, and the data update that does not require a commit may be selectively enabled.

  The DB client 20 acquires and updates data on the DB servers 10-1 and 10-2, and has a DB client process 21. The DB client 20 may be realized, for example, by installing an application (a program including the database client program of the embodiment) corresponding to the DB client process 21 in a computer (for example, a PC or a workstation).

  The DB client process 21 uses a type indicating that the data update request is a two-phase commit and a list of DB servers 10 (hereinafter referred to as “server list”) that are targets of the two-phase commit as a data update request or data It has an interface that can be defined in messages such as commit requests. As a result, the DB client process 21 can select the update method used when updating data and the target DB server 10.

  For example, the configuration of a message M (message of a control command related to two-phase commit) generated by the DB client process 21 can be conceptually shown as in FIG. Note that the configuration of the actual message M needs to follow the format on the platform to be used.

  As shown in FIG. 3, the message M includes at least the type of update method based on the message, a server list to which the message is to be transferred, and definition data of the message (data update content or data commit content definition data). It shall be assumed.

  The message M type (update method type) shown in FIG. 3 indicates that two-phase commit is set.

  In the server list shown in FIG. 3, the identifiers of the DB servers 10 are described side by side in accordance with the message transfer order. For example, in FIG. 3, since “DB # 1, DB # 2” is described as the server list, the message is first supplied to the DB server 10-1, and then from the DB server 10-1 to the DB server 10 -2 indicates an instruction to transfer to -2. In other words, the message M shown in FIG. 3 indicates that the transfer destination of the first stage is the DB server 10-1, and the transfer destination (termination) of the next second stage is the DB server 10-2.

  When the DB server 10 (DB server process 11) receives a message M as shown in FIG. 3, for example, the DB server 10 (DB server process 11) recognizes that the data update request is a two-phase commit request, and sends the data update request to the next stage. This corresponds to a function that enables transmission (that is, message transfer) to the two-phase commit DB server 10. In addition, the DB server 10 (DB server process 11) corresponds to a function that enables the process of updating or committing the data (data on the file system 12) of its own device in parallel according to the received message definition data. It shall be. Further, in the DB server 10 (DB server process 11), a transaction for updating (commiting) the data of the own device and a transaction for transferring the message and performing data update (commit) performed on the DB server 10 side in the next stage are the same. It is assumed that it also supports a function for managing

  That is, in the DB system 1, it can be said that the DB client 20 is connected to the DB servers 10-1 and 10-2 in multiple stages (connected in series). In the following description, the DB server that is directly accessed by the DB client 20 is also referred to as the foremost DB server 10. In the following description, the DB base server 10 corresponding to the end of message transfer (the tail of the server list) is also referred to as the last DB server 10. In this embodiment, the DB server 10-1 corresponds to the foremost DB server 10, and the DB server 10-2 corresponds to the last DB server. It is assumed that the preceding device of the foremost DB server 10-1 is the DB client 20.

(A-2) Operation of the First Embodiment Next, the operation of the DB system 1 of the first embodiment having the above configuration will be described.

  Here, the update method for the DB servers 10-1 and 10-2 (DB # 1 and DB # 2) is sent from the DB client 20 (DB client process 21) to the DB servers 10-1 and 10-2 (DB # 1 and DB # 2) using a message as shown in FIG. The operation when a data update request for phase commit is performed will be described with reference to the sequence diagram of FIG.

  First, it is assumed that the DB client 20 (DB client process 21) generates a data update request message configured as shown in FIG. 3 and transmits it to the DB server 10-1 (S101).

  At this time, the time required for message transmission from the DB client 20 (DB client process 21) to the DB server 10-1 (time from the start to the end of transmission) is set to “t1”.

  Next, the DB server 10-1 (DB server process 11) analyzes the received data update request message, recognizes that the transaction is a two-phase commit transaction, and further executes the next-stage DB server on the server list. 10-2, a data update request message similar to the message is transmitted (S103), and at the same time, data update of the own device (that is, I / O to the DB file system 12 of the own device) is started ( S102).

  At this time, the time required to transmit the data update request (message transmission) from the DB server 10-1 to the DB server 10-2 (time from the start of message transmission to the completion of transmission) is “t3”, and the DB server 10− The time required for data update in 1 (time from the start of data update to the end of update) is assumed to be “t2”.

  When receiving the data update request message from the DB server 10-1, the DB server 10-2 (DB # 2) recognizes from the contents of the server list of the message that the own device is the end server of the two-phase commit. The data of the own device is updated (S104).

  At this time, the time required for data update in the DB server 10-2 is set to “t4”.

  Then, when the data update is completed, the DB server 10-2 transmits a data update response (processing result) to the DB server 10-1 that is the message transmission source (first stage on the server list) of the data update request (S105). ).

  At this time, the time taken to transmit a data update response (message transmission) from the DB server 10-2 to the DB server 10-1 is “t5”.

  When the reception of the data update response from the DB server 10-2 is completed, the DB server 10-1 completes the data update of its own device and the DB that is the message transfer destination (next stage on the server list) of the data update request. The data update response (processing result) from the server 10-2 is transmitted to the message transmission source (DB client 20) of the data update request (S106).

  At this time, the time required to transmit a data update response (message transmission) from the DB server 10-1 to the DB client 20 is represented as “t6”.

  In this embodiment, since there are two DB servers 10, the data update response received by the DB server 10-1 is only for the DB server 10-2. When the DB server 10 is also present in the subsequent stage, the DB server 10-1 collectively transfers the data update responses of those DB servers 10 to the previous stage.

  Upon receiving the data update response from the DB server 10-1, the DB client 20 generates a message for requesting data commit (data commit request message having a configuration as shown in FIG. 3), and this time, the DB server 10- 1 (S107).

  At this time, the time taken for the data commit request transmission (message transmission) from the DB client 20 to the DB server 10-1 is assumed to be “t7”.

  Upon receiving the data commit request message, the DB server 10-1 executes the data commit of its own device according to the message (S108), and then transfers the data commit request message to the next DB server 10-2 ( S109).

  At this time, the time required for transmission (message transmission) of the data commit request from the DB server 10-1 to the DB server 10-2 is “t9”, and the time required for data commit in the DB server 10-1 is “t8”. .

  Upon receiving the data commit request message from the DB server 10-1, the DB server 10-2 executes the data commit of its own device according to the content of the message (S110), and sends a data commit response (processing result). It transmits to DB server 10-1 (S111).

At this time, the time taken for the data commit in the DB server 10-2 is “t10”,
The time taken to transmit a data commit response (message transmission) from the DB server 10-1 to the DB client 20 is “t11”.

  When the reception of the data commit response from the DB server 10-2 is completed, the DB server 10-1 terminates the data commit of its own device and the DB that is the message transfer destination (next stage on the server list) of the data commit request. Data commit responses (processing results) from the server 10-2 are collectively transmitted to the message commit source (DB client 20) of the data commit request (S112).

  At this time, the time taken to transmit a data commit response (message transmission) from the DB server 10-1 to the DB client 20 is set to “t12”.

  As described above, in the DB system 1, data is updated by two-phase commit.

  Here, since message transmission / reception between the same nodes is considered to pass the same route on the network, t1, t6, t7, t12, t3, t5, t9, t11 are expressed by the following formula (1) and (2 It is assumed that an equation such as

t1 = t6 = t7 = t12 (1)
t3 = t5 = t9 = t11 (2)
Further, if the performances of the DB servers 10-1 and 10-2 (time required for data update and time required for data commit) are equal, the following equations (3) and (4) are established.

t2 = t4 (3)
t8 = t10 (4)
There are two points to note here. First, the DB system 1 inserts a server list into a data update request message, and a data update request message sent from the DB client 20 is transferred (transmitted) to each DB server 10 to update the data. Is the point where is executed. The second precaution is that in the DB system 1, the data update of the DB server 10-1, the message transfer to the DB server 10-2, and the data update process of the DB server 10-2 are performed in parallel. In other words, only the value with the longer processing time needs to be counted. Therefore, in the DB system 1 of the first embodiment, the required time x1 from the start to the end of data update by two-phase commit (from the start of step S101 to the end of step S112) is expressed by the following equation (5) It can be shown as follows. Then, t1 = t3 = t4 = t8 = t9 = t14 = 1 ms, t5 = t7 = t11 = t13 = 5 ms, t2 = t6 = 10 ms, and t10 = t12 = 5 ms. x1 is 44 ms as shown in the following equation (6).

x1 = t1 + (t2, t3 + t4, whichever is larger) + t5 + t6
+ T7 + t8 + t9 + t10 + t11 + t12 (5)
x1 = 1 + 15 + 5 + 1 + 1 + 5 + 5 + 5 + 5 + 1 = 44 ms (6)
(A-3) Effects of First Embodiment According to the first embodiment, the following effects can be achieved.

(A-3-1) As described above, in the DB system 1, the server list is inserted into the data update request message, and the data update request message sent from the DB client 20 is transferred (transmitted) to each DB server 10. Then, the data update is executed.

  Temporarily, in the DB system 1, it is necessary to transmit separate messages (data update request message and data commit message) to the DB server 10-1 and the DB server 10-2 from the DB client 20. In this case, the sequence diagram is as shown in FIG.

  In the sequence diagram of FIG. 4, first, the DB client 20 transmits a data update request message to the DB server 10-1 (S201), and the DB server 10-1 updates the data (S201). -1 transmits a data update response to the DB client 20 (S203). In the sequence diagram of FIG. 4, next, the DB client 20 transmits a data update request message to the DB server 10-2 via the DB server 10-1 (S204, S205). Data update is performed (S206), and a data update response is transmitted from the DB server 10-1 to the DB client 20 via the DB server 10-2 (S207, S208). Further, in the sequence diagram of FIG. 4, the DB client 20 transmits a data commit request message to the DB server 10-1 (S209), the DB server 10-1 executes data commit (S210), and the DB server 10 -1 transmits a data commit request to the DB server 10-2 (S211), the DB server 10-2 executes data commit (S212), and the DB server 10-2 sends a data commit response to the DB server 10-1. To the DB client 20 (S213, S214). In FIG. 4, the time required for each of steps S201 to S214 is t201 to t214, respectively.

  If the DB system 1 performs two-phase commit data update in the sequence shown in FIG. 4, the required time x2 can be expressed as the following equation (7). In FIG. 4, it is assumed that message transmission / reception between the same nodes follows the same route on the network, and the following equations (8) and (9) are established. Further, it is assumed that the performance of each DB server 10 is equivalent, and the following formulas (10) and (11) hold. Further, similarly to the example of FIG. 1 described above, the message transmission time between the DB client 20 and the DB server 10-1 is 1 ms, and the message transmission between the DB server 10-1 and the DB server 10-2 is performed. Is 5 ms, the data update time is 10 ms, and the data commit time is 5 ms. As shown in the following equation (12), the time required for the two-phase commit data update in the sequence of FIG. 4 (step S201). (Required time of step S214) is 56 ms. Therefore, it can be seen that the time required for the processing of the sequence of FIG. 4 is longer than that of the processing of the sequence of FIG.

x3 = t201 + t202 + t203 + t204 + t205
+ T206 + t207 + t208 + t209 + t210
+ T211 + t212 + t213 + t214 (7)
t201 = t203 = t204 = t208 = t209 = t214 (8)
t205 = t207 = t211 = t213 (9)
t202 = t206 (10)
t210 = t212 (11)
x = 1 + 10 + 1 + 1 + 5 + 10 + 5 + 1 + 1
+ 5 + 5 + 5 + 5 + 1 = 56 ms (12)
That is, in the DB system 1, when updating the data of the two-phase comet, the server list is inserted into the data update request message, and the data update request message sent from the DB client 20 is transferred to each DB server 10 ( Transmission), the sequence (step) can be shortened and the time required for data update can be shortened.

(A-3-2) As described above, data update of the DB server 10-1 (data update of the own device), message transfer to the DB server 10-2 (message transfer to the next stage), and DB server The data update process 10-2 (data update process in the next stage) is performed in parallel.

  Thereby, in the DB system 1, a partial sequence (step) of data update by two-phase commit can be executed in parallel, and the required time can be further shortened. Even if the delay time between the DB server 10-1 and the DB server 10-2 is zero, the DB system 1 can reduce the required time compared to the conventional method.

(B) Second Embodiment Hereinafter, a second embodiment of a database system, a database server, a database server program, a database client, and a database client program according to the present invention will be described in detail with reference to the drawings.

  The configuration of the DB system 1 of the second embodiment can also be shown using FIG. Below, only the difference with 1st Embodiment is demonstrated about DB system 1 of 2nd Embodiment.

  In the DB system 1 of the first embodiment, the data update of the DB server 10-1 (data update of the own device), the message transfer to the DB server 10-2 (message transfer to the next stage), and the DB server Although the data update process 10-2 (data update process in the next stage) is executed in parallel, the data commit is not such a structure. On the other hand, in the DB system 1 of the second embodiment, even in the data commit, the data commit of the DB server 10-1 (data commit of the own device) and the message transfer to the DB server 10-2 (next) Message transfer to the stage) and data commit of the DB server 10-2 (data commit in the next stage) are performed in parallel with the first embodiment.

  Therefore, the operation of the DB system 1 of the second embodiment can be expressed using the sequence diagram of FIG. In the sequence diagram of FIG. 5, the same steps (step numbers) are assigned to the same steps as those in the sequence diagram 1 of the first embodiment.

  As shown in FIG. 5, in the DB system 1 of the second embodiment, the data commit of the DB server 10-1 (Step S 108), the message transfer to the DB server 10-2 (Step S 109), and the DB server 10. -2 data commit (step S110) is executed in parallel.

  Therefore, in the DB system 1 of the second embodiment, since the data update of the two-phase commit is performed in the sequence as shown in FIG. 5, the required time x3 can be expressed as the following equation (13). As in the example of FIG. 1, the message transmission time between the DB client 20 and the DB server 10-1 is 1 ms, and the message transmission between the DB server 10-1 and the DB server 10-2 is performed. The required time x3 is 5 ms, the data update time is 10 ms, and the data commit time is 5 ms. The required time x3 for two-phase commit data update in the DB system 1 of the second embodiment is expressed by the following equation (14). As shown, the time is 39 ms, and the required time can be shortened compared to the processing in the first embodiment (sequence of FIG. 1). That is, in the DB system 1 of the second embodiment, a part of the sequence (steps) of data commit is executed in parallel, thereby further reducing the required time compared to the first embodiment. be able to.

x3 = t1 + (the larger of t2, t3 + t4)
+ T5 + t6 + t7 + (the larger of t8 and t9 + t10)
+ T11 + t12 (13)
x3 = 1 + 15 + 5 + 1 + 1 + 10 + 5 + 1 = 39 ms (14)
(C) Other Embodiments The present invention is not limited to the above-described embodiments, and may include modified embodiments as exemplified below.

(C-1) In each of the above embodiments, for simplicity of explanation, there is no DB server 10 (node) that relays between the DB server 10-1 and the DB server 10-2. Of course, the DB server 10 may be inserted between them.

(C-2) In each of the above embodiments, all the DB servers 10 have been described as having the same configuration. However, since the latter DB server 10 does not exist for the last DB server 10, the latter DB A configuration that does not correspond to a function of transferring a message to the server 10 or a function of processing a processing result received from the subsequent DB server 10 may be adopted.

(C-3) In each of the above embodiments, the DB system 1 has been described as a configuration corresponding to two-phase commit data update. However, the DB system 1 has a configuration corresponding to a three-phase commit and a similar sequence (see FIG. 3 or FIG. Data update and data commit may be performed by a process including the sequence 5).

  DESCRIPTION OF SYMBOLS 1 ... DB system 1, 10-1, 10-2 ... DB server 10-1, 10 ... DB server, 11 ... DB server process, 12 ... DB file system, 20 ... DB client, 21 ... DB client process.

Claims (7)

In a database system comprising a plurality of database servers and a database client,
Database servers excluding the last database server among the plurality of database servers are
In accordance with a data update request supplied from the preceding apparatus, data update means for performing data update,
A data update request transfer means for transferring the data update request supplied from the preceding apparatus to the subsequent database server;
When there is a data update result response to the data update request from the subsequent database server, a data update response means for transmitting the data update result that combines the data update result and the data update result of the own device to the previous device;
Data commit means for performing data commit in accordance with the data commit request supplied from the preceding device;
A data commit request transfer means for transferring the data commit request supplied from the preceding apparatus to the subsequent database server;
When there is a response of the data commit result to the data commit request from the database server at the latter stage, a data commit response means for sending the data commit result that combines the data commit result and the data commit result of the own apparatus to the preceding apparatus Have
The last database server is
Data update response means for performing data update according to the data update request supplied from the previous database server, and responding to the previous database server with the data update result;
Data commit response means for performing data commit in accordance with the data commit request supplied from the previous database server, and responding to the previous database server with the data commit result;
The database client
A data update request means for transmitting a data update request to the frontmost database server among the plurality of database servers, and receiving all the data update results of the plurality of database servers from the frontmost database server;
A data commit request means for transmitting a data commit request to the frontmost database server and receiving all data commit results of the plurality of database servers from the frontmost database server; system.   In each of the database servers other than the last database server among the plurality of database servers, the data update request transfer process to the subsequent database by the data update response unit and the data update process by the data update unit are performed in parallel. The database system according to claim 1, wherein the database system is executed.     The transfer processing of the data commit request to the subsequent database by the data commit response means and the data commit processing by the data commit means in parallel with each of the database servers other than the last database server among the plurality of database servers. The database system according to claim 1 or 2, wherein the database system is executed. In accordance with a data update request supplied from the preceding apparatus, data update means for performing data update,
A data update request transfer means for transferring the data update request supplied from the preceding apparatus to the subsequent database server;
When there is a data update result response to the data update request from the subsequent database server, a data update response means for transmitting the data update result that combines the data update result and the data update result of the own device to the previous device;
Data commit means for performing data commit in accordance with the data commit request supplied from the preceding device;
A data commit request transfer means for transferring the data commit request supplied from the preceding apparatus to the subsequent database server;
When there is a data commit result response to the data commit request from the database server at the latter stage, a data commit response means for sending the data commit result, which is a combination of the data commit result and the data commit result of its own device, to the former stage A database server characterized by comprising: Computer
In accordance with a data update request supplied from the preceding apparatus, data update means for performing data update,
A data update request transfer means for transferring the data update request supplied from the preceding apparatus to the subsequent database server;
When there is a data update result response to the data update request from the subsequent database server, a data update response means for transmitting the data update result that combines the data update result and the data update result of the own device to the previous device;
Data commit means for performing data commit in accordance with the data commit request supplied from the preceding device;
A data commit request transfer means for transferring the data commit request supplied from the preceding apparatus to the subsequent database server;
When there is a response to the data commit request for the data commit request from the database server at the latter stage, the data commit response means to send the data commit result that combines the data commit result and the data commit result of its own device to the previous stage device. A database server program characterized by being made to function. A data update request means for transmitting a data update request to the frontmost database server among the plurality of database servers, and receiving all data update results of the plurality of database servers from the frontmost database server;
A data commit request means for transmitting a data commit request to the frontmost database server and receiving all data commit results of the plurality of database servers from the frontmost database server; client. Computer
A data update request means for transmitting a data update request to the frontmost database server among the plurality of database servers, and receiving all data update results of the plurality of database servers from the frontmost database server;
A data commit request is transmitted to the foremost database server and functions as data commit request means for receiving all data commit results of the plurality of database servers from the foremost database server. Database client program.

Images