JP2021135828A - Request processing system and request processing method - Google Patents

Abstract

To process a write request from a client at high speed.SOLUTION: A plurality of client terminals transmits a write request for data and a read request for data to each of a first server and a second server. The first server executes writing processing corresponding to the write request when receiving the write request. The first server transmits order notification including order of write requests received before the read request to the second server when receiving the read request. The second server receives and holds the write request, and executes the writing processing corresponding to the held write request according to the order included in the order notification when receiving the order notification.SELECTED DRAWING: Figure 6

Description

The present invention relates to a request processing system and a request processing method.

When data is created in a computer system, it is common practice to create duplicates to reduce the risk of data loss. For example, "Passive replication" and "Active replication" are known as replication creation methods. "Passive replication" is a method in which only the primary server receives a write request and the primary server forwards the write request to the remaining backup servers. On the other hand, "Active replication" is a method in which all servers directly receive write requests. Since "Active replication" can execute processing in parallel as compared with "Passive replication", the time required for writing can be shortened.

Also, in an environment where there are multiple clients that are the source of write requests, an event may occur in which consistency is not maintained between each server. For example, in "Passive replication", since there is only one request entrance, write requests are processed in the order of reception on the server side without providing a special mechanism in the client device or between the client and the server. Can be done.

On the other hand, in "Active replication", control of the order of write requests may be required in order to maintain consistency between servers. For example, assume that the client A sends a write request A1 and the client B sends a write request B1 to each of the servers X, Y, and Z. At this time, depending on the transmission timing and the network route, the servers X and Z may receive the write request in the order of A1 and B1, and the server Y may receive the write request in the order of B1 and A1. In this case, the data after writing matches between the servers X and Z, but the data after writing does not match between the servers X and Z and the server Y, and the consistency cannot be maintained between the servers. Cannot meet the requirement to keep a copy of.

Therefore, in order to realize "Active replication" in a multiple client environment, a mechanism called "atomic broadcast" or "total order broadcast" is used to "receive requests in the same order at all recipients". Be done. For example, a system configuration in which an order control server is installed between a client and a server is known. The client sends a write request to the order control server, and the order control server controls the order of the write requests sent from the client and then sends the write request to each server.

Japanese Unexamined Patent Publication No. 2001-101062

By the way, high-speed writing is one of the values that users demand, and there are systems in which the higher the writing speed, the higher the value. For example, auction bids and stock trading are required to be processed at the fastest speed without delay in response to a request from a client.

However, when the above technique is applied in a plurality of client environments, it is difficult to execute high-speed write processing in response to a write request from a client. For example, in the case of "Passive replication", since the representative server writes to another server after the client writes to the representative server, the processes are executed sequentially, and the write process cannot be executed at high speed. In the case of "Active replication", write processing can be executed in parallel, but in order to guarantee consistency between servers, it is necessary to control the order of write requests by the order control server, which takes time.

One aspect is to provide a request processing system and a request processing method capable of processing a write request from a client at high speed.

In the first plan, the request processing system has a first server and a second server that execute processing according to a request transmitted from each of the plurality of client terminals. Each of the plurality of client terminals has a transmission unit that transmits a data write request or a data read request to each of the first server and the second server. When the first server receives the write request, the write processing unit executes a write process in response to the write request, and when the first server receives the read request, the write request received before the read request. It has a read processing unit that transmits an order notification including the order of the above to the second server. The second server receives and holds the write request, and when the order notification is received, the second server executes a write process according to the write request to be held in accordance with the order included in the order notification. It has a processing unit.

On one side, write requests from clients can be processed at high speed.

FIG. 1 is a diagram showing an overall configuration example of the request processing system according to the first embodiment. FIG. 2 is a functional block diagram showing a functional configuration of the request processing system according to the first embodiment. FIG. 3 is a diagram showing an example of request information of the representative server. FIG. 4 is a diagram showing an example of request information of the replication server. FIG. 5 is a diagram illustrating processing at the time of a write request for the entire system. FIG. 6 is a diagram illustrating process 1 at the time of a read request for the entire system. FIG. 7 is a diagram illustrating process 2 at the time of a read request for the entire system. FIG. 8A is a sequence diagram showing a flow of processing according to the first embodiment. FIG. 8B is a sequence diagram showing a flow of processing according to the first embodiment. FIG. 9 is a diagram showing an overall configuration example of the request processing system according to the second embodiment. FIG. 10 is a functional block diagram showing a functional configuration of the request processing system according to the second embodiment. FIG. 11 is a diagram illustrating request processing according to the second embodiment. FIG. 12 is a diagram illustrating a hardware configuration example.

Hereinafter, examples of the request processing system and the request processing method disclosed in the present application will be described in detail with reference to the drawings. The present invention is not limited to this embodiment. In addition, each embodiment can be appropriately combined within a consistent range.

[overall structure]
FIG. 1 is a diagram showing an overall configuration example of the request processing system according to the first embodiment. As shown in FIG. 1, the request processing system is a system in which a client terminal A10, a client terminal B20, a representative server 30, a replication server 40, and a replication server 50 are connected to each other so as to be able to communicate with each other via a network N. The number of servers and the number of client terminals are examples and can be changed arbitrarily. Further, as the network N, various communication networks such as the Internet and a dedicated line can be used regardless of whether they are wired or wireless.

In this request processing system, each of the representative server 30, the duplication server 40, and the duplication server 50 stores the data to be written or read by each of the client terminal A10 and the client terminal B20. Further, the data stored in each replication server is for reducing the risk of data loss in the representative server 30, and it is required that the data are synchronized with each other.

Such a request processing system is a system that realizes "Active replication" in a plurality of client environments. Specifically, the client terminal A10 or the client terminal B20 transmits a write request and a read request to each of the representative server 30, the replication server 40, and the replication server 50. Then, each of the representative server 30, the duplication server 40, and the duplication server 50 executes processing according to each request received from each client terminal.

Here, in general, in "Active replication" in a plurality of client environments, write processing can be executed in parallel, but in order to guarantee consistency between servers, the order control server controls the order of write requests. It is necessary and the write process cannot be executed at high speed. For this reason, a system that is required to process requests from clients at the fastest speed without delay, such as bidding for auctions and buying and selling stocks, cannot meet the user's requirements, resulting in a decrease in system reliability. Connect.

Therefore, in the request processing system according to the first embodiment, the request is transmitted to the plurality of replication servers without adjusting the order between the client terminals, and the write order already executed by the representative server 30 before reading is left as the remaining write order. By notifying the replication server, speed and reliability will be improved.

Specifically, the client terminal A10 and the client terminal B20 transmit a data write request or a data read request to the representative server 30, the replication server 40, and the replication server 50, respectively. When the representative server 30 receives the write request, the representative server 30 executes the write process according to the write request. Further, when the representative server 30 receives the read request, the representative server 30 sends an order notification including the order of the write requests received before the read request to each replication server. Then, when each replication server receives the write request, it holds the request without executing the write process, and when it receives the order notification from the representative server 30, it holds the write in the order included in the order notification. Executes write processing according to the request.

By doing so, in the request processing system according to the first embodiment, high-speed write completion notification to the user is realized by minimizing the processing to be executed at the time of writing. As a result, in "Active replication" in a plurality of client environments, write requests from clients can be processed at high speed without using a special mechanism for order control.

[Functional configuration]
FIG. 2 is a functional block diagram showing a functional configuration of the request processing system according to the first embodiment. Since the client terminal A10 and the client terminal B20 have the same configuration, only the client terminal A10 will be described. Similarly, since the replication server 40 and the replication server 50 have the same configuration, only the replication server 40 will be described.

(Client terminal configuration)
As shown in FIG. 2, the client terminal A10 has a communication unit 11, a storage unit 12, and a control unit 13.

The communication unit 11 is a processing unit that controls communication between the representative server 30 and each replication server, and is realized by, for example, a communication interface. For example, the communication unit 11 transmits various requests to the representative server 30 and each replication server, and receives responses corresponding to the various requests from the representative server 30 and each replication server.

The storage unit 12 is a processing unit that stores various data, a program executed by the control unit 13, and the like, and is realized by, for example, a memory or a hard disk.

The control unit 13 is a processing unit that controls the entire client terminal A10, and is realized by, for example, a processor. The control unit 13 has a write request unit 13a and a read request unit 13b. The write request unit 13a and the read request unit 13b can be realized by an electronic circuit such as a processor, or can be realized as a process executed by the processor.

The write request unit 13a is a processing unit that transmits a write request to the representative server 30 and each replication server. For example, the write request unit 13a transmits a write request to the representative server 30 and each replication server at an arbitrary trigger such as a user operation or an operation by an application. Then, the write request unit 13a receives the processing result from the server that has executed the writing process in response to the write request.

Here, it is assumed that the write request is transmitted by assigning the order of the requests to each client terminal. For example, the write request transmitted from the client terminal A is assigned an A as an ID and a number indicating the transmission order, and the write request A1, the write request A2, the write request A3, and the like are given in the order of transmission. Similarly, the write request transmitted from the client terminal B is assigned a B as an ID and a number indicating the transmission order, and becomes a write request B1, a write request B2, a write request B3, and the like in the order of transmission. The write request A1 or the like may be described by simply omitting "A1" or the like.

The read request unit 13b is a processing unit that transmits a read request to the representative server 30 and each replication server. For example, the read request unit 13b transmits a read request to the representative server 30 and each replication server at an arbitrary trigger such as a user operation or an operation by an application. Then, the read request unit 13b receives the processing result from the server that has executed the reading process in response to the read request. As for the description method of the read request, an ID and a number indicating the transmission order are assigned as in the above write request.

(Configuration of representative server 30)
As shown in FIG. 2, the representative server 30 has a communication unit 31, a storage unit 32, and a control unit 33. The communication unit 31 is a processing unit that controls communication between each client terminal and each replication server, and is realized by, for example, a communication interface. For example, the communication unit 31 receives a write request and a read request from each client terminal. In addition, the communication unit 31 transmits responses corresponding to various requests to each client terminal, and transmits an order notification to be described later to each replication server.

The storage unit 32 is a processing unit that stores various data, a program executed by the control unit 33, and the like, and is realized by, for example, a memory or a hard disk. For example, the storage unit 32 stores the request information 32a and the database 32b.

The request information 32a is information about various requests received from each client terminal. For example, the request information 32a is information that manages the type and reception order of the requests transmitted by each client terminal. The request information 32a can also be managed so that it can be distinguished whether each request has been executed or not executed.

FIG. 3 is a diagram showing an example of request information 32a of the representative server 30. As shown in FIG. 3, the request information 32a is associated with a "order" indicating the order of reception and a "request" indicating the type of received request. In the example of FIG. 3, the write request A1 transmitted first by the client terminal A10, the write request A2 transmitted second by the client terminal A10, the write request B1 transmitted first by the client terminal B20, and the client terminal A10 are three. It indicates that the representative server 30 has received the write request A3 transmitted second and the write request B2 transmitted second by the client terminal B20 in this order.

The database 32b stores data to be processed by each client terminal. Specifically, the database 32b is managed for each user, and stores data to be written or read by the user of each client terminal.

The control unit 33 is a processing unit that controls the entire representative server 30, and is realized by, for example, a processor. The control unit 33 includes a write processing unit 33a, an order notification unit 33b, and a read processing unit 33c. The write request unit 13a and the read request unit 13b can be realized by an electronic circuit such as a processor, or can be realized as a process executed by the processor.

The write processing unit 33a is a processing unit that executes a write process according to the write request without delay when a write request is received from the client terminal A10 or the client terminal B20. Specifically, when the write processing unit 33a receives the write request, the write processing unit 33a updates the request information 32a. Then, the write processing unit 33a executes the write process to the database 32b in the order in which the write requests are received, and transmits the process result (execution result) to the client terminal of the request source.

For example, when the requests are received in the order shown in FIG. 3, the write processing unit 33a executes the write process in the order of write request A1, write request A2, write request B1, write request A3, and write request B2. Here, when the write processing unit 33a receives the request, the write processing unit 33a executes the write process corresponding to the request without waiting for the reception of another request or the like. That is, when the write processing unit 33a receives the write request A1, the write process corresponding to the write request A1 is immediately executed, and when the write request A2 is received thereafter, the write process corresponding to the write request A2 is immediately executed. Run.

The order notification unit 33b is a processing unit that, when a read request is received from the client terminal A10 or the client terminal B20, transmits an order notification including the order of the write requests received before the read request to each replication server.

For example, when the order notification unit 33b receives the read request from the client terminal A after receiving the write request shown in FIG. 3, the order notification unit 33b refers to the request information 32a and the reception order of the write request is "write request A1, write request". It is specified that it is "A2, write request B1, write request A3, write request B2". Then, the order notification unit 33b provides information in the order of "write request A1, write request A2, write request B1, write request A3, write request B2" (for example, "A1 → A2 → B1 → A3 → B2"). ) Is generated and transmitted to the replication server 40 and the replication server 50.

When the order notification unit 33b receives the read request, the order notification unit 33b registers the read request in the request information 32a and notifies the read processing unit 33c of the read request.

The read processing unit 33c is a processing unit that executes read processing in response to a read request transmitted from the client terminal A10 or the client terminal B20. Specifically, the read processing unit 33c executes the read process after the order notification is transmitted by the order notification unit 33b and the reception response indicating the normal reception of the order notification is received from each replication server. Then, the read processing unit 33c transmits the processing result including the read data and the like to the request source.

(Configuration of replication server 40)
As shown in FIG. 2, the replication server 40 includes a communication unit 41, a storage unit 42, and a control unit 43. The communication unit 41 is a processing unit that controls communication with each client terminal and the representative server 30, and is realized by, for example, a communication interface. For example, the communication unit 41 receives a write request and a read request from each client terminal. Further, the communication unit 41 receives the order notification from the representative server 30.

The storage unit 42 is a processing unit that stores various data, a program executed by the control unit 33, and the like, and is realized by, for example, a memory or a hard disk. For example, the storage unit 42 stores the request information 42a and the database 42b.

The request information 42a is information about various requests received from each client terminal. For example, the request information 42a is information that manages the type and reception order of the requests transmitted by each client terminal. The request information 42a can also be managed so that it can be distinguished whether each request has been executed or not executed.

FIG. 4 is a diagram showing an example of request information 42a of the replication server 40. As shown in FIG. 4, the request information 42a is information in which a "order" indicating a reception order and a "request" indicating a type of received request are associated with each other. In the example of FIG. 4, the write request A1 transmitted first by the client terminal A10, the write request B1 transmitted first by the client terminal B20, the write request A2 transmitted second by the client terminal A10, and the client terminal A10 are three. It indicates that the replication server 40 has received the write request A3 transmitted second and the write request B2 transmitted second by the client terminal B20 in this order.

Here, when the request information 42a of the replication server 40 shown in FIG. 4 and the request information 32a of the representative server 30 shown in FIG. 3 are compared, although the requests are transmitted from each client terminal in the same order, the requests are transmitted. The order of receiving requests is different between the replication server 40 and the representative server 30. This is due to various factors such as the communication path, the mixed communication status, and the load on the server. Therefore, the replication server 40 is required to execute the write process in the same order as the representative server 30 in order to maintain the consistency of the database with the representative server 30.

The database 42b stores data to be processed by each client terminal. Specifically, the database 42b is managed for each user, and stores data to be written or read by the user of each client terminal. Since the representative server 30 and each replication server have a relationship (maintenance of consistency) for managing data replication, each database needs to be synchronized.

The control unit 43 is a processing unit that controls the entire replication server 40, and is realized by, for example, a processor. The control unit 43 has a receiving unit 43a and a writing processing unit 43b. The receiving unit 43a and the writing processing unit 43b can be realized by an electronic circuit such as a processor, or can be realized as a process executed by the processor.

The receiving unit 43a is a processing unit that holds a write request when a write request is received from the client terminal A10 or the client terminal B20. Specifically, when the receiving unit 43a receives a write request, it registers in the request information 42a in the order of reception without executing the write process, and manages the reception order. At this time, although the receiving unit 43a has not executed the writing process, since the writing request has been received normally, the receiving unit 43a transmits a response corresponding to the writing request to the client terminal of the transmitting source.

When the order notification is received from the representative server 30, the write processing unit 43b is a processing unit that executes the write process according to the write request to be held according to the order included in the order notification. For example, the writing processing unit 43b receives “A1 → A2 → B1 → A3 → B2” as an order notification. Then, the write processing unit 43b is notified by the order notification even though "write request A1, write request B1, write request A2, write request A3, write request B2" is registered in the request information 42a. The writing process is executed in the order of "A1-> A2-> B1-> A3-> B2". The writing processing unit 43b transmits a reception response indicating that the order notification has been normally received to the representative server 30 after receiving the order notification or after the writing is completed. Further, after the writing is completed, the writing processing unit 43b may send a notification of the writing completion to the representative server 30.

[Specific example of processing]
Next, with reference to FIGS. 5 to 7, the flow at the time of issuing each request in the entire system will be specifically described. FIG. 5 is a diagram illustrating processing at the time of a write request for the entire system. FIG. 6 is a diagram for explaining the process 1 at the time of a read request for the entire system, and FIG. 7 is a diagram for explaining the process 2 at the time of a read request for the entire system.

As shown in FIG. 5, the client terminal A10 sends a write request A1 to the representative server 30 and each replication server, then sends a write request A2 to the representative server 30 and each replication server, and then further. , The write request A3 is transmitted to the representative server 30 and each replication server. Further, the client terminal B20 transmits the write request B1 to the representative server 30 and each replication server, and then transmits the write request B2 to the representative server 30 and each replication server.

Then, when the representative server 30 receives the write request A1, it executes the write process to the database 32b, and then when it receives the write request A2, it executes the write process to the database 32b. Further, when the representative server 30 receives the write request B1, it executes the write process to the database 32b, and then when it receives the write request A3, it executes the write process to the database 32b and further receives the write request B2. Then, the writing process to the database 32b is executed. That is, the representative server 30 receives the write requests A1, A2, B1, A3, and B2 in this order, and executes the write process in the received order.

On the other hand, unlike the representative server 30, the replication server 40 receives the write requests A1, B1, A2, A3, and B2 in this order, but suppresses the write request without executing the write process. Further, unlike the representative server 30 and the replication server 40, the replication server 50 receives the write requests A1, A2, A3, B1, and B2 in this order, but suppresses the write request without executing the write process.

After that, as shown in FIG. 6, the client terminal A10 transmits a read request A1 to the representative server 30 and each replication server, and the representative server 30 and each replication server receive the read request A1.

Then, the representative server 30 specifies A1, A2, B1, A3, and B2 as the order of the write requests received by the own device so far, and uses this order as the order information, respectively, of the replication server 40 and the replication server 50. Send to.

After that, as shown in FIG. 7, the representative server 30 that has transmitted the order notification executes the read process according to the read request A1, and transmits the data read from the database 32b to the request source client terminal A10.

On the other hand, the replication server 40 holds the write requests in the order of "A1, B1, A2, A3, B2", but sorts the write requests in the order of "A1, A2, B1, A3, B2" notified by the order notification. Then, writing to the database 42b is sequentially executed. Similarly, the replication server 50 holds the write requests in the order of "A1, A2, A3, B1, B2", but sorts the write requests in the order of "A1, A2, B1, A3, B2" notified by the order notification. Then, writing to the database 52b is sequentially executed.

[Processing sequence]
8A and 8B are sequence diagrams showing a flow of processing according to the first embodiment. As shown in FIG. 8A, the client terminal A10 (or the client terminal B20) transmits a write request to each of the representative server 30, the replication server 40, and the replication server 50 (S101 to S104).

Then, the representative server 30 executes a write process to the database 32b (S105), and transmits the process result as a write response to the client terminal A10 (S106 and S107).

On the other hand, the replication server 40 does not execute the write process and holds the received write request (S108). After that, the replication server 40 transmits a write response to the client terminal A10 as a response to the write request (S109 and S110).

Similarly, the replication server 50 does not execute the write process and holds the received write request (S111). After that, the replication server 50 transmits a write response to the client terminal A10 as a response to the write request (S112 and S113).

Subsequently, the client terminal A10 or the client terminal B20 transmits a write request to each of the representative server 30, the replication server 40, and the replication server 50 (S114 to S117).

Then, the representative server 30 executes a write process to the database 32b (S118), and transmits the process result as a write response to the client terminal A10 (S119 and S120).

On the other hand, the replication server 40 does not execute the write process and holds the received write request (S121). After that, the replication server 40 transmits a write response to the client terminal A10 as a response to the write request (S122 and S123).

Similarly, the replication server 50 does not execute the write process and holds the received write request (S124). After that, the replication server 50 transmits a write response to the client terminal A10 as a response to the write request (S125 and S126).

Subsequently, as shown in FIG. 8B, the client terminal A10 or the client terminal B20 transmits a read request to each of the representative server 30, the replication server 40, and the replication server 50 (S127 to S130).

Then, the representative server 30 generates an order notification and transmits it to the replication server 40 and the replication server 50 (S131 to S133). Then, the replication server 40 executes a write process to the database 42b according to the order notification (S134), and transmits a reception response indicating that the order notification has been normally received to the representative server 30 (S135 and S136). Similarly, the replication server 50 executes a write process to the database 52b according to the order notification (S137), and transmits a reception response indicating that the order notification has been normally received to the representative server 30 (S138 and S139). ..

On the other hand, the representative server 30 receives the received response to the transmitted order notification from each replication server, executes the read process (S140), and transmits the data read from the database 32b to the client terminal A10 as the read response. (S141 and S142).

[effect]
As described above, the request processing system according to the first embodiment does not execute the order control at the time of transmitting the write request in "Active replication", and executes the order control at the time of reading the data. Further, at the time of a write request, only the representative server 30 executes the database update process, and the other replication servers only receive and record the write request, and do not immediately execute the update process. That is, sharing of the update order on the representative server 30 to the replication server is executed before reading the data, and the update process on the replication server is executed when the update order is received.

In this way, the request processing system according to the first embodiment adopts "Active replication", but does not control the order of write requests at the time of transmission, so that the order is controlled by using the general "Atomic broadcast". It is possible to realize high-speed writing as compared with the method of performing. Further, the request processing system according to the first embodiment realizes high-speed write processing as compared with the method of performing order control, and is faster than "Passive replication" in which the processing is sequentially executed. Processing can be realized.

By the way, in the request processing system shown in FIG. 1, when a failure occurs in the representative server 30, another replication server is promoted to the representative server to maintain the continuity of the system. In that case, it is required to maintain the consistency of the database. Therefore, in the second embodiment, an example of maintaining the consistency of the database when a failure of the representative server 30 occurs will be described.

[overall structure]
FIG. 9 is a diagram showing an overall configuration example of the request processing system according to the second embodiment. As shown in FIG. 9, in the request processing system, as in FIG. 1, the client terminal A10, the client terminal B20, the representative server 30, the replication server 40, and the replication server 50 are connected to each other so as to be able to communicate with each other via the network N. It is a system.

In such a state, if a failure occurs in the representative server 30, the replication server 40 is promoted to the representative server to maintain the continuity of the system. The failure includes various failures such as a failure of the communication path between each client terminal and the representative server 30, and a failure of the representative server 30 itself. Further, as a method of promoting the replication server 40 to a representative server in the event of a failure, a general technique used in a redundant system or the like can be adopted.

[Functional configuration]
FIG. 10 is a functional block diagram showing a functional configuration of the request processing system according to the second embodiment. Since the client terminal A10 and the client terminal B20 have the same configuration, only the client terminal A10 will be described, but since the configuration is the same as that of the first embodiment, detailed description thereof will be omitted. Further, since the representative server 30 has the same configuration as that of the first embodiment, detailed description thereof will be omitted.

(Configuration of replication server 40)
As shown in FIG. 10, the replication server 40 promoted to the representative server has a communication unit 41, a storage unit 42, and a control unit 43. Since the communication unit 41, the storage unit 42, the reception unit 43a of the control unit 43, and the write processing unit 43b of the control unit 43 have the same functions as those of the first embodiment, detailed description thereof will be omitted. Here, the fault handling unit 43c and the transmitting unit 43d, which are different from those in the first embodiment, will be described.

The failure response unit 43c is a processing unit that executes write processing in the order of write requests received by the replication server 40 when a failure occurs in the representative server 30. For example, when the failure response unit 43c causes a failure in the representative server 30 and the replication server 40 is promoted to the representative server, the replication server 40 sends the received and unexecuted write requests in the reception order of the replication server 40. Execute with. At this time, the failure handling unit 43c can also transmit the result of the writing process to each request source.

The transmission unit 43d is a processing unit that transmits the order of the write requests executed by the failure response unit 43c to the replication server 50. For example, the transmission unit 43d reflects a write request that was not executed at the time of the failure to the replication server 40 in order to maintain consistency between the replication server 40 and the replication server 50 after the failure of the representative server 30 occurs. The replication server 50 is notified of the order in which they are made.

(Configuration of replication server 50)
As shown in FIG. 10, the replication server 50, which is in charge of replication even after the failure of the representative server 30, has a communication unit 51, a storage unit 52, and a control unit 53. The communication unit 51, the storage unit 52, the reception unit 53a of the control unit 53, and the write processing unit 53b of the control unit 53 are the communication unit 41, the storage unit 42, the reception unit 43a of the control unit 43, and the control unit of the replication server 40. Since the function is the same as that of the write processing unit 53b of 43, detailed description thereof will be omitted. Here, the failure handling unit 53c, which is different from the replication server 40, will be described.

When the failure response unit 53c receives the order notification from the replication server 40, the failure response unit 53c is a processing unit that executes a write process according to the write request to be held according to the order included in the order notification. For example, the failure response unit 53c executes the write process in the same order as the write request executed by the copy server 40, not in the write request order registered in the request information 52a of the copy server 50 which is its own device.

[Concrete example]
Next, a method of reflecting a write request when a failure occurs in the representative server 30 will be described with reference to FIG. FIG. 11 is a diagram illustrating request processing according to the second embodiment. FIG. 11 shows the order of the write requests received by the representative server 30, the replication server 40, and the replication server 50. After the representative server 30 determines the processing up to B2 and notifies the remaining replication servers, It represents a state in which the representative server 30 is stopped before the notification of the processing order of A4, A5, and B3.

That is, the writing process to the database is completed in the order of A1, A2, B1, A3, and B2 among the representative server 30, the replication server 40, and the replication server 50. However, after that, since it is before the notification from the representative server 30, the reception order of A4, A5, and B3 does not match among the three servers.

In this case, neither the replication server 40 nor the replication server 50 knows the execution order on the representative server 30, so that the processes cannot be executed in the same order as the execution order on the representative server 30. Therefore, the replication server 40 or the replication server 50 is promoted to the representative server, and the processing is executed in the order of reception on the promoted representative server. For example, the replication server 40 is promoted to a representative server, and the replication server 40 determines the execution order in the reception order "A4, B3, A5" of the requests that have not been executed in the replication server 40. Then, the replication server 40 executes the write requests in the determined execution order, and notifies the replication server 50 of the execution order of execution.

Here, on the system, in order to reflect the write request from the client terminal and quickly generate a consistent state, various requests from each client terminal are requested until the request processing of A4, B3, and A5 is completed. Suppress acceptance and stop processing.

[effect]
As described above, when a failure occurs in the representative server 30 and it becomes necessary to switch, the write requests for which the notification has not come from the representative server 30 at the switching destination are in the order received by the switching destination. Perform update processing. As a result, it is possible to quickly reflect unreflected write requests on each replication server, suppress inconsistencies between systems, and maintain consistency between replication servers. It is necessary to maintain consistency only when the execution result of the process is observed externally by performing the read process. At the stage where only writing is performed, the writing source does not know in what order the processing is performed, so the states do not necessarily have to match between the servers.

By the way, although the examples of the present invention have been described so far, the present invention may be implemented in various different forms other than the above-mentioned examples.

[Numerical values, etc.]
The number of servers, the number of requests, etc. used in the above embodiment are merely examples and can be changed arbitrarily.

[system]
Information including processing procedures, control procedures, specific names, various data and parameters shown in the above documents and drawings can be arbitrarily changed unless otherwise specified. The representative server 30 corresponds to the first server, the replication server 40 corresponds to the second server, and the replication server 50 corresponds to the third server. The write request unit 13a and the read request unit 13b of the client terminal A correspond to the transmission unit.

Further, each component of each of the illustrated devices is a functional concept, and does not necessarily have to be physically configured as shown in the figure. That is, the specific forms of distribution and integration of each device are not limited to those shown in the figure. That is, all or a part thereof can be functionally or physically distributed / integrated in any unit according to various loads, usage conditions, and the like.

Further, each processing function performed by each device may be realized by a CPU and a program analyzed and executed by the CPU, or may be realized as hardware by wired logic.

[hardware]
Next, a hardware configuration example will be described. Since each device shown in FIG. 1 and the like has a similar hardware configuration, it will be described here as a computer 1. FIG. 12 is a diagram illustrating a hardware configuration example. As shown in FIG. 12, the computer 1 includes a communication device 1a, an HDD (Hard Disk Drive) 1b, a memory 1c, and a processor 1d. Further, the parts shown in FIG. 12 are connected to each other by a bus or the like.

The communication device 1a is a network interface card or the like, and communicates with another server. The HDD 1b stores a program or DB that operates the functions shown in FIGS. 2 and 10.

The processor 1d executes each function described in FIGS. 2 and 10 by reading a program that executes the same processing as each processing unit shown in FIGS. 2 and 10 from the HDD 1b or the like and expanding the program into the memory 1c. Run the process. For example, this process executes the same function as each processing unit of the representative server 30. Specifically, the processor 1d reads a program having the same functions as the write processing unit 33a, the order notification unit 33b, the read processing unit 33c, and the like from the HDD 1b and the like. Then, the processor 1d executes a process of executing the same processing as the writing processing unit 33a, the order notification unit 33b, the reading processing unit 33c, and the like.

In this way, the computer 1 operates as an information processing device that executes the request processing method by reading and executing the program. Further, the computer 1 can realize the same function as that of the above-described embodiment by reading the program from the recording medium by the medium reading device and executing the read program. The program referred to in the other embodiment is not limited to being executed by the computer 1. For example, the present invention can be similarly applied when another computer or server executes a program, or when they execute a program in cooperation with each other.

10 Client terminal A
11 Communication unit 12 Storage unit 13 Control unit 13a Write request unit 13b Read request unit 20 Client terminal B
30 Representative server 31 Communication unit 32 Storage unit 32a Request information 32b Database 33 Control unit 33a Write processing unit 33b Order notification unit 33c Read processing unit 40, 50 Replication server 41 Communication unit 42 Storage unit 42a Request information 42b Database 43 Control unit 43a Reception Unit 43b Writing processing unit

Claims (6)

In a request processing system having a first server and a second server that execute processing according to a request transmitted from each of a plurality of client terminals.
Each of the plurality of client terminals
Each of the first server and the second server has a transmission unit for transmitting a data write request or a data read request.
The first server is
When the write request is received, the write processing unit that executes the write process in response to the write request and the write processing unit
When the read request is received, it has a read processing unit that transmits an order notification including the order of write requests received before the read request to the second server.
The second server is
A receiver that receives and holds the write request,
A request processing system including a write processing unit that executes write processing according to a write request to be held in accordance with the order included in the order notification when the order notification is received. The first server has a storage unit that stores a first database including data to be processed by each of the plurality of client terminals.
The second server has a storage unit that stores a second database that is required to maintain consistency with the first database.
When the write processing unit of the first server receives the write request, the write processing unit executes the write processing on the first database.
The write processing unit of the second server suppresses the write process for the second database when the write request is received, and when the order notification is received, the order is The request processing system according to claim 1, wherein the write process is executed for the second database according to the order included in the notification. The read processing unit according to claim 1 or 2, wherein the read processing unit of the first server executes a read process in response to the read request after transmitting the order notification to the second server. Request processing system. A claim characterized in that, when a failure occurs in the first server, the write processing unit of the second server executes the write process in the order of the write requests received by the second server. The request processing system according to any one of items 1 to 3. The request processing system has a third server having the receiving unit of the second server and the writing process.
The second server is
It has a transmission unit that transmits a second order notification including the order of write requests executed by the second server to the third server.
The third server is
The fourth aspect of claim 4 is characterized in that, when the second order notification is received, it has a failure handling unit that executes a write process according to a write request to be held in accordance with the order included in the second order notification. Request processing system. In a request processing method executed by a request processing system having a first server and a second server that execute processing according to a request transmitted from each of a plurality of client terminals.
Each of the plurality of client terminals
A data write request or a data read request is transmitted to each of the first server and the second server, respectively.
The first server is
When the write request is received, the write process corresponding to the write request is executed, and the write process is executed.
When the read request is received, an order notification including the order of the write requests received before the read request is sent to the second server.
The second server is
Receive and hold the write request
When the order notification is received, the write process according to the write request to be held is executed according to the order included in the order notification.
A request processing method characterized by including processing.

Images