JP2015108930A - Switch method between direct and sub systems - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for switching flexibly a direct system and a sub system without generating an application error.SOLUTION: When a direct system is operated by normal operation, a transfer destination of a direct system load balancer 12 is switched to a destination for sub system transfer time in which a fictitious address is set in primary and a sub system load balancer 22 is set in secondary, from a destination for normal time in which a direct system WEB/AP server 14 is set in primary, and by the time when the primary becomes time out, replication between the direct system DB server 16→sub system DB server 26 is completed and reverse setting in a replication direction is completed. After time out, request from a client terminal 40 is transferred to the direct system load balancer 12→the sub system load balancer 22→the sub system WEB/AP server 24, and a processing result by the sub system WEB/AP server 24 is reflected in the order of the sub system DB server 26→the direct system DB server 16.

Description

  The present invention relates to a method for switching between a primary system and a secondary system, and in particular, in addition to a primary system that executes various information processing in normal times, a backup secondary system that acts as a substitute for the primary system when a failure occurs is provided. The present invention relates to a technology for switching a processing subject from a primary system to a secondary system on a trial basis in a redundant computer system, or switching back from a secondary system to a primary system after operation confirmation.

In addition to the primary system responsible for business processing in normal times, a secondary system with the same function is prepared at a remote location, and when the primary system goes down, the secondary system is started quickly. There is a technology that allows business processes to be taken over.
JP 2006-330782 A System redundancy and disaster countermeasures Internet URL: http://www.networld.co.jp/vmware/solution/sol5.htm Search date: November 20, 2013

  For example, in Patent Document 1 described above, a redundant configuration of data processing includes a primary business server 1 in charge of business processing in a normal state and a backup secondary business server 2 having a configuration substantially similar to this. A system is disclosed. In this data processing system, the input message 7 transmitted from the data transmission device 5 is simultaneously transmitted to the primary business server 1 and the secondary business server 2 and stored in the respective entry queues 12 and 22. Further, the processing result in the primary business system is transferred to the secondary business system, stored in the business data storage unit 24, and in the input message 7 stored in the entry queue 22, the primary business server 1 A read flag is set by the entry queue management unit 27 for the completed processing. For this reason, the secondary business server 2 that has received a notification that a failure has occurred in the primary business server 1 can immediately continue processing the unprocessed input message 7 for which the read flag is not set. It becomes possible.

  By adopting such a redundant configuration of the system, even if the primary system goes down, processing is continued by the secondary system, so it can be said that it is extremely effective as a preparation for a large-scale disaster or the like.

By the way, even during normal times when the primary system is operating normally, the processing entity is temporarily transferred to the secondary business system for the purpose of confirming whether the secondary system has failed or as part of disaster drills. There is a need to switch to.
However, when the processing entity is switched from the primary system to the secondary system, an application error occurs on the client terminal side as the primary system stops, resulting in a problem in providing services.
For this reason, conventionally, it has been practically difficult to switch from the primary system to the secondary system while maintaining the service.

  The present invention has been devised to solve the problems associated with switching between such primary and secondary systems, and can be processed between the primary system and the secondary system without causing an application error on the client terminal side. The purpose is to provide technology that allows the subject to be flexibly changed.

  In order to achieve the above object, the switching method between the primary and secondary systems described in claim 1 includes a primary system request distribution unit for transferring a received request to a predetermined address, and a request received from the primary system request distribution unit. A secondary system that consists of a primary system that consists of a primary business server that performs the necessary processing according to the process, a primary database server that updates the corresponding table according to the processing result, and a received request to a predetermined address. A sub-system consisting of a sub-system server that executes necessary processing according to the request received from the sub-request distribution means, and a sub-database server that updates the corresponding table according to the processing result. In normal times, after the main request distribution means receives the request transmitted from the client terminal The primary business server executes processing according to the request, and the primary database server updates the corresponding table according to the processing result by the primary business server and is transmitted from the primary database server. The secondary database server updates the corresponding table according to the processing result of the primary business server, and when the primary system failure occurs, the secondary request distribution means receives the request transmitted from the client terminal. After that, the subordinate business server executes processing corresponding to the request, the subordinate database server updates the corresponding table according to the processing result by the subordinate business server, and from the subordinate database server. Corresponding to the primary database server according to the processing result of the secondary business server sent In a redundant configuration system having a mechanism for updating a table, a method of switching a processing subject from a primary system to a secondary system in a normal state while maintaining a service for a client terminal, and transferring the above-mentioned primary request distribution means From the normal setting where the primary business server is set as the primary, the fictitious address is set as the primary, and the secondary system transition where the address of the secondary request distribution means is set as the secondary A first step of switching to a time setting; a second step in which the primary request distribution means transfers a new request transmitted from a client terminal to the fictitious address; and the primary database server , The processing result for the previous request executed by the main business server A third step for transmitting to the subordinate database server, a fourth step for the subordinate database server to update a corresponding table based on the processing result, and after the elapse of a preset timeout period, The primary request distribution means includes a fifth step of transferring the new request to the secondary request distribution means set to the secondary.

  The switching method between the primary and secondary systems described in claim 2 is the method of claim 1, further comprising a sixth step in which the secondary system request distribution means transfers the request to the secondary business server. The seventh sub-operation server executes processing according to the request and transmits the processing result to the sub-database server, and the sub-database server responds based on the processing result. An eighth step of updating the table and transmitting the processing result to the main database server, and a ninth step of updating the corresponding table based on the processing result by the main database server. It is characterized by that.

  The switching method between the primary and secondary systems described in claim 3 is the method according to claim 2, and further, the transfer destination of the primary request distribution means is set as a fictitious from the setting for the transition to the secondary system. Is sent from the client terminal, and the tenth step of switching to the setting for the transition to the primary system in which the address of the primary business server is set as the secondary, and the primary request distribution means is sent from the client terminal. An eleventh step of transferring the new request to the fictitious address, and a twelfth step in which the secondary database server transmits a processing result for the previous request executed by the secondary business server to the primary database server. And a thirteenth step in which the main database server updates the corresponding table based on the processing result, After the elapse of the set timeout period, the main system request distribution means transfers the new request to the main business server set as the secondary, and the main business server A fifteenth step of executing processing according to the request and transmitting the processing result to the primary database server, and the primary database server updating the corresponding table based on the processing result, and the processing result 16 to the subordinate database server, a seventeenth step in which the subordinate database server updates the corresponding table based on the processing result, and a transfer destination of the primary request allocating means Is provided with an eighteenth step of switching from the setting for transition to the normal system to the setting for normal time.

  The switching method between the primary and secondary systems described in claim 4 is the method according to claims 1 to 3, further comprising the primary database server and the secondary system between the fourth step and the fifth step. It is characterized in that a step of reversing the replication direction between database servers is provided.

  The switching method between the primary and secondary systems described in claim 5 is the method according to claim 3 or 4, and further, between the 13th step and the 14th step, the primary database server and the secondary system. It is characterized in that a step of reversing the replication direction between database servers is provided.

  According to the switching method between the primary and secondary systems described in claim 1, the request is first transferred from the primary system request distribution means to the fictitious IP address, and the primary database server uses the time difference until the primary side times out. After the processing result is sent to the secondary database server and the update of the table on the secondary system side is completed, it has a mechanism to distribute requests to the secondary system, so without causing an application error on the client terminal side, Therefore, the processing subject can be flexibly switched from the primary system to the secondary system while continuing the service.

  According to the switching method between the primary and secondary systems described in claim 3, even when returning the processing subject from the secondary system to the primary system, first, the request is transmitted from the primary system request distribution means to the fictitious IP address, A system that distributes requests to the primary business server after the processing results are sent from the secondary database server to the primary database server using the time difference until the primary side times out and the update of the table on the primary system side is completed. Therefore, it is possible to smoothly return to normal operation without causing an application error on the client terminal side.

FIG. 1 is a conceptual diagram showing a redundant configuration system 30 comprising a primary system 10 and a secondary system 20. The primary system 10 includes a primary load balancer 12 as a primary request distribution means, and a primary service. It has a primary WEB / AP server 14 as a server and a primary DB server 16 as a primary database server. The secondary system 20 includes a secondary load balancer 22 as a secondary request distribution means, A secondary WEB / AP server 24 as a server and a secondary DB server 26 as a secondary database server are provided.
The secondary system 20 is installed at a remote location with respect to the primary system 10, and the two systems are connected by a dedicated line.

In normal times, the data transfer destination of the primary load balancer 12 is set to the primary WEB / AP server 14, so when a request is sent from the client terminal 40 via the network, the primary load balancer 12 The request is transferred to the system WEB / AP server 14.
Receiving this, the primary WEB / AP server 14 executes processing according to the content of the request and transmits the processing result to the primary DB server 16.

The primary DB server 16 updates the corresponding table based on the processing result data received from the WEB / AP server 14, and simultaneously transmits the data to the secondary DB server 26 via the network.
The secondary DB server 26 updates the corresponding table based on the received processing result.

  Here, for example, when a large-scale earthquake occurs and the main system 10 actually goes down, even if a temporary interruption occurs in the provision of services, the client terminal 40 is changed by changing the setting of the DNS server (not shown). Since the connection destination is switched to the secondary load balancer 22, the provision of services by the secondary system 20 can be resumed within a relatively short time.

Next, a processing procedure for temporarily switching data processing to the secondary system 20 under the situation where the primary system 10 is operating normally will be described based on the flowchart of FIG.
First, as shown in FIG. 3, when a switchover command is sent from the management terminal 42 to the primary load balancer 12 to instruct the setting change to the secondary operation mode, the primary load balancer that receives this command is sent. 12 switches the transfer destination of the received data from the normal use to the sub system transfer (S10).

  As a transmission destination for the sub system migration, a fictitious IP address that does not actually exist in “primary” is set, and the IP address of the sub system load balancer 22 is set in “secondary”.

  When a new request is transmitted from the client terminal 40 in this state, the primary load balancer 12 forwards the request toward the fictitious IP address set as primary (S12).

During this time, the primary DB server 16 transfers untransmitted data to the secondary DB server 26 and completes replication between the primary and secondary DB servers (S14).
Next, as shown in FIG. 4, the primary DB server 16 and the secondary DB server 26 that have received a command from the management terminal 42 change the replication direction between “primary → secondary” to “secondary → The setting is changed so as to be reversed to “main system” (S16).

  On the other hand, since there is no response from the primary side even if a preset timeout period (for example, 90 seconds) elapses (S18), as shown in FIG. 5, the primary load balancer 12 has the secondary load set as secondary. The above request is transferred to the IP address of the balancer 22 (S20).

Then, necessary processing is executed in the secondary WEB / AP server 24 that has received this request from the secondary load balancer 22 (S22), and the processing result is stored in the correspondence table in the secondary DB server 26 (S24). ).
The processing result data is transferred from the secondary DB server 26 to the primary DB server 16 via the network, and the replication between the primary and secondary DB servers is completed (S26).

  As described above, in this redundant configuration system 30, first, a request is transmitted from the primary load balancer 12 to a fictitious IP address, and the time difference until the primary side times out is referred to as “primary DB server 16 → sub system DB server 26 ”replication and reversal of the replication direction are completed, and then the request is distributed to the secondary system 20, so there is no application error on the client terminal side, so the service can be The processing subject can be temporarily switched from the primary system 10 to the secondary system 20 while continuing.

In the primary load balancer 12, the timeout time for switching the transfer destination from primary to secondary is sufficient to complete replication between the primary DB server 16 and secondary DB server 26, but the application on the client terminal 40 side It is set to a short value that does not cause an error.
If the primary DB server 16 and the secondary DB server 26 are equipped with a function that enables bi-directional replication, the step of “S16” can be omitted. The setting of the timeout time can be shortened.

When a sufficient operation check for the sub system 20 is completed, the redundant configuration system 30 is switched again to the main system operation mode.
First, as shown in FIG. 6, when a switchback command is sent from the management terminal 42 to the primary load balancer 12 to instruct the return to the primary operation mode, the primary load balancer that has received this command is sent. 12 switches the transfer destination of the received data from the time of sub system shift to the time of main system shift.

  As the destination for the primary transition, a fictitious IP address that does not actually exist is set as “primary”, and the IP address of the primary WEB / AP server 14 is set as “secondary”.

  When a new request is transmitted from the client terminal 40 in this state, the primary load balancer 12 forwards the request toward the fictitious IP address set as the primary.

During this time, the secondary DB server 26 transfers unsent data to the primary DB server 16 and completes replication between the primary and secondary DB servers.
Next, the primary DB server 16 and the secondary DB server 26 that have received a command from the management terminal 42 are set to reverse the replication direction from “subsystem → primary system” to “primary system → subsystem”. Make changes.

  On the other hand, even if a preset timeout period (for example, 90 seconds) elapses, there is no response from the primary side, so the primary load balancer 12 is directed to the IP address of the primary WEB / AP server 14 set as secondary. Forward the above request.

As a result, necessary processing is executed in the primary WEB / AP server 14, and the processing result is stored in the correspondence table in the primary DB server 16.
Further, the data of the processing result is transmitted from the primary DB server 16 to the secondary DB server 26 via the network and stored in the corresponding table.

After that, the primary load balancer 12 switches the setting of the transfer destination at an appropriate timing from that for transitioning to the normal system to normal use.
In this normal address setting, the IP address of the primary WEB / AP server 14 is set to “Primary”, so that requests sent from the client terminal 40 will be forwarded to a fictitious IP address thereafter. Instead, it is immediately transferred to the main system WEB / AP server 14 and necessary processing is executed without delay.

  In a general redundant configuration system, as described above, after the update to the main DB server is completed, the processing result is transmitted to the backup DB server. ”Has achieved replication between databases.

  Of course, in order to guarantee the complete identity of the data, first send the processing result from the main DB server to the backup DB server, and when the data update completion notification is returned, Ideally, the so-called “synchronous mode” that executes the update is adopted, but the network load is inevitably increased. Therefore, the “asynchronous mode” should be adopted in normal times assuming peak service usage. The current situation is unavoidable.

On the other hand, when operating this redundant configuration system 30 in the secondary operation mode, it is possible to intentionally select a time zone when the service usage peak time has been excluded. Replication between DB servers 16 can be executed in "synchronous mode".
As a result, when switching back from the secondary system operation mode to the primary system operation mode, there is no need to wait for the replication between the "secondary DB server 26 → the primary system DB server 16" to be completed. In the system load balancer 12, it is possible to shorten the setting of the timeout time for switching the transfer destination from “primary” to “secondary”.

  In the above description, the primary load balancer 12 and the secondary load balancer 22 are used as request distribution means. However, a primary proxy server and a secondary proxy server can be used instead.

It is a conceptual diagram which shows the state at the time of normal operation | movement of the redundantly configured system which concerns on this invention. It is a flowchart which shows the process sequence at the time of switching the said redundant structure system to sub-system operation mode. It is a conceptual diagram which shows the state in the middle of switching the said redundant configuration system to sub-system operation mode. It is a conceptual diagram which shows the state in the middle of switching the said redundant configuration system to sub-system operation mode. It is a conceptual diagram which shows the state at the time of the sub system operation mode of the said redundant configuration system. It is a conceptual diagram which shows the state in the middle of switching the said redundantly configured system to the normal operation mode.

10 Main system
12 Main load balancer
14 Main web / AP server
16 Primary database server
20 Subsystem
22 Subordinate load balancer
24 Subordinate WEB / AP server
26 Subordinate database server
30 Redundant system
40 client terminals
42 Management terminal

Claims (5)

Main system request distribution means for transferring received requests to a predetermined address, a main business server for executing necessary processing according to the request received from the main system request distribution means, and a table corresponding to the processing result A primary system consisting of a primary database server that updates the request, a secondary request distribution unit that forwards the received request to a predetermined address, and executes necessary processing according to the request received from the secondary request distribution unit A subordinate system consisting of a subordinate business server, a subordinate database server that updates the corresponding table according to the processing result,
In normal times, after the request sent from the client terminal is received by the primary request distribution means, the primary business server executes processing according to the request, and the primary database server is the primary business server. The corresponding table is updated according to the processing result by the main database server, and the sub database server updates the corresponding table according to the processing result of the main business server transmitted from the main database server. When a system failure occurs, after the request sent from the client terminal is received by the sub system request distribution means, the sub system business server executes processing corresponding to the request, and the sub system database server executes the sub system request server. The corresponding table is updated according to the processing result of the secondary business server, and this secondary database is also updated. In redundant system having a mechanism in which the main system database server updates the corresponding table according to the processing result of the transmitted said secondary system business server from Susaba,
A method for switching a processing subject from a primary system to a secondary system in a normal state while maintaining a service for a client terminal,
From the setting for normal use in which the primary business server is set as the primary, the fictitious address is set as the primary, and the secondary request distribution means as the secondary. A first step of switching to the setting for transition to the sub system in which the address of
A second step in which the primary request distribution means transfers a new request transmitted from the client terminal to the fictitious address;
A third step in which the primary database server transmits a processing result for a previous request executed by the primary business server to the secondary database server;
A fourth step in which the secondary database server updates the corresponding table based on the processing result;
A fifth step in which the primary request distribution means transfers the new request to the secondary request distribution means set to the secondary after elapse of a preset timeout period;
Switching method between primary and secondary systems. A sixth step in which the sub system request distribution means transfers the request to the sub business server;
A seventh step in which the subordinate business server executes processing according to the request and transmits a processing result to the subordinate database server;
The secondary database server updates the corresponding table based on the processing result, and transmits the processing result to the primary database server,
A ninth step in which the primary database server updates the corresponding table based on the processing result;
The switching method between the primary and secondary systems according to claim 1, comprising: When transferring the primary request distribution means to the primary system from which the fictitious address is set as the primary and the address of the primary business server is set as the secondary from the settings for the secondary system transition A tenth step of switching to a setting for
An eleventh step in which the primary request distribution means transfers the new request transmitted from the client terminal to the fictitious address;
A twelfth step in which the secondary database server transmits a processing result for a previous request executed by the secondary business server to the primary database server;
A thirteenth step in which the primary database server updates a corresponding table based on the processing result;
A fourteenth step in which the primary request distribution means transfers the new request to the primary business server set as a secondary after the elapse of a preset timeout period;
A fifteenth step in which the main business server executes a process according to the request and transmits a processing result to the main database server;
The primary database server updates a corresponding table based on the processing result, and a sixteenth step of transmitting the processing result to the secondary database server;
A seventeenth step in which the sub-database server updates a corresponding table based on the processing result;
An eighteenth step of switching the transfer destination of the main system request distribution means from the setting for transition to the main system to the setting for normal time;
The switching method between the primary and secondary systems according to claim 2, comprising:   4. The step of inverting the replication direction between the primary database server and the secondary database server is provided between the fourth step and the fifth step. Switching method between the primary and secondary systems described.   5. The primary / secondary system according to claim 3, wherein a step of inverting the replication direction between the primary database server and the secondary database server is provided between the thirteenth step and the fourteenth step. How to switch between systems.

Images