JP2000250771A - Server duplication system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To fast and also reliably operate a duplication system for a network server and database server which poses a problem in a 24-hour operation non- stop system. SOLUTION: In this server duplication system, a server 1 in operation and a server 2 in a standby mode respectively have each process connected to main processes 11 and 21 and are monitored all the time, In the synchronization of the servers of respective databases 15 and 25, updation history is always stored in the updation history database 16 of the self-server 1 when the database updation process 14 of the server 1 updates the database 15 of the self-server. Further, the updation history is notified to the database updation process 24 of the standby server 2 and updates the database 25 of the self-server 2 and an updation history database 26. When a failure takes place in the server 1 and communication between the processes 11 and 21 is interrupted, the server 2 is made to operate and continues subsequent processing in an instant. Thus, system central servers are switched in a non-stop state.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dual server system for constructing a non-stop communication system.

[0002]

2. Description of the Related Art Conventionally, a database server system of a non-stop redundant system uses a system configuration for switching between an active server and a standby server. in this case,
In general, a cross-call that does not synchronize transaction data between the active server and the standby server is used. As a method that does not use cross calls,
For example, there is JP-A-9-305558. In this method, when the active server is updated, the data of the standby server is also updated at the same time, and the active server is switched to the standby server when the active server becomes abnormal without synchronizing the transaction data.

[0003]

However, in the method using the cross call, the main process of the server monitors the process, but the server is switched without restarting when the process is abnormally terminated. Switching may occur due to a bug in a business process or the like. In many cases, the load increases, and in some cases, the operation ends abnormally. The way of handling the database includes a configuration in which a disk is mounted from two servers by cross-call, but in this method, o
When using a database such as a rake, for example, the steps of “stopping the database of the active server” → “unmounting the disk of the active server” → “mounting the disk of the spare server” → “starting the database of the spare server” must be performed. Because of that, it takes a lot of time to switch servers.

[0004] In addition, the terminal is connected to the server from the terminal side, but in this method, the terminal side does not know the state of the server. On the other hand, only a method of alternately transmitting connection requests at regular intervals can be adopted. Therefore, a considerable switching time is required for the whole system. Also, in the method disclosed in JP-A-9-305558,
Since the method of connecting to the server from the terminal side is adopted, the problem of the switching time of the entire system has not been solved. As described above, in the conventional server switching method, there is a problem that the switching time between the active and the standby cannot be shortened because a factor that delays the system switching exists in the process configuration.

The present invention has been made in view of such circumstances, and an object of the present invention is to provide a high-speed and high-speed system, which is a dual system of a network server and a database server, which is a problem in a 24-hour non-stop system. It is to provide a system configuration that can be reliably performed.

[0006]

According to the server duplication method of the present invention, the database of the active server and the database of the standby server are always synchronized, and the active server fails and the main server of both servers is synchronized. As soon as the communication between the processes is interrupted, the standby server becomes the active server, and the standby server takes over the communication without interruption. As a result, the system central server can be instantaneously switched, and a highly reliable server duplex system can be constructed.

That is, in the server redundancy system according to the first aspect, in a server redundancy system in which a plurality of servers are used as an active system and a standby system to configure a non-stop system, each server is provided with a process in its own server. Connected and constantly monitor each process, and also a main process that is interconnected with other servers, a database that stores information of the entire system including business contents based on user requests, and when updating the contents of the database Update history database that stores the update information and notification information results to other servers, and updates the database based on the business content, and stores the update information and notification information results in the update history database, A database update process for notifying other servers of update information, the active server and the standby server Characterized by being configured uninterrupted system by the.

[0008] According to a second aspect of the present invention, the main process monitors each process of its own server, and when a failure occurs in any of the processes, the main process monitors its own process. It is characterized in that processing for restarting the server is performed.

According to a third aspect of the present invention, when the database update process of the active server updates the database of its own server, the database update process includes the update information. In the update history database of its own server, and notifies the update information to the database update process of the standby server, and the standby server updates its own database and the update history database based on the received update information, The database of the active server and the database of the standby server are synchronized, and when the active server fails, the communication state is immediately taken over to the standby server immediately.

[0010] According to a fourth aspect of the present invention, in the server dual mode according to any one of the first to third aspects, each server is connected to a client terminal by a switching notification from a main process. The terminal interface process and the terminal are connected to each other by setting a server process to a client connection and a terminal to a server connection.

According to a fifth aspect of the present invention, there is provided a server duplication method according to any one of the first to fourth aspects, further comprising a monitoring process for monitoring a main process as a higher-level device. It monitors each process in its own server.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a system configuration diagram of a redundant server system according to the first embodiment of this invention. First, a first embodiment of a server redundancy system according to the present invention will be described with reference to FIG. This system includes an active server 1, a standby server 2, and a client terminal group (hereinafter, referred to as terminals) 3. The running server 1
It comprises a main process 11, a business process 12, a terminal interface process 13, a database update process 14, a database 15, and an update history database 16.

The main process 11 mainly monitors each process in the running server 1 and keeps a connection with the main process 21 of the standby server 2 at all times. When each process in the running server 1 ends abnormally, a process restart process is performed. The business process 12 is a process for operating the terminal interface process 13 and the database update process 14, and differs depending on the business content. Further, the terminal interface process 13
Is a process for relaying communication between the business process 12 and the terminal 3, and receives a request from the terminal 3 to the business process 12 and transmits a result from the business process 12 to the terminal 3.

The database update process 14 updates the database 15 in response to a request from the business process 12, notifies the update information to the database update process 24 of the standby server 2, and updates the communication result and the update information with the update history. Store it in the database 16. The database 15 is an information source of the entire system and differs in content depending on the contents of business. However, the database 15 needs to be always synchronized with the database 25 of the standby server 2 in order to realize a duplex system. The update history database 16 stores the update history of the database 15 and the notification result to the waiting server 2. The role of the notification result is that “not yet notified” is added to the database update history information while the standby server 2 is stopped.
Is added, update information not reflected in the database 25 is taken over when the standby server 2 is restored.

On the other hand, the standby server 2 has the same configuration as the active server 1, and includes a main process 21, a business process 22, a terminal interface process 23, a database update process 24, a database 25, and an update history database. 26.

The main process 21 is always connected to the main process 11 of the running server 1, and switches the standby server 2 from the standby state to the active state when the failure notification of the active server 1 is received. . Business process 22
Keeps the standby state until an operation switching notification is received from the main process 21. Further, upon receiving the switching notification from the main process 21, the terminal interface process 23 performs a connection process with the terminal 3. Further, the database update process 24 is always connected to the database update process 14 of the active server 1 and, upon receiving the update information, reflects the update information in the database 25 of the standby server 2. The database 25 and the update history database 26 are stored in a database update process 24.
As a result, the server 1 is always synchronized with the operating server 1. Further, the terminal 3 is always connected to the running server 1 and issues a request to the running server 1.

As for the connection method, when a socket connection is taken as an example, the terminal 3 is connected as a server, and the terminal interface process 13 of the active server 1 is connected as a client. That is, by using the terminal 3 as a server, connection from two servers becomes possible. That is, when the connection state is reversed, it is not possible to determine which server is operating and connectable.

Next, the operation of the server redundant system having such a configuration will be described. That is, in FIG. 1, the operating server 1 is operating, and the standby server 2 is in standby. The processes in the servers 1 and 2 are connected to the main processes 11 and 21, respectively, and the processes are constantly monitored. Each main process 11, 21
Respectively perform only simple processing such as monitoring of each process in the own servers 1 and 2, and perform processing such as restarting when each process ends abnormally.

The synchronization between the servers of the databases 15 and 25 is performed by updating the update history of the server 1 whenever the database update process 14 of the operating server 1 updates the database 15 of the server 1. The information is stored in the history database 16 and notified to the database update process 24 of the standby server 2, and the database update process 24 of the standby server 2 updates the received update information to the database 25 and the update history database 26 of the own server 2. Thereby, the database 15 of the active server 1 and the database 25 of the standby server 2 are always synchronized.

In this situation, if a failure occurs in the running server 1 and communication between the main processes 11 and 21 is interrupted,
The waiting server 2 is in operation, and the subsequent processing can be continued instantaneously. In this way, each main process 11 and 21 is made to monitor each process, and furthermore, the system central server (active server 1) is instantaneously switched to the system spare server (standby server 2), thereby improving reliability. A high server redundancy system can be constructed.

Further, the operation of the first embodiment shown in FIG. 1 will be described in detail with reference to a flowchart. FIG.
9 is a flowchart showing the flow of a database synchronization process in a normal state. First, the flow of processing in a normal state will be described with reference to FIGS. First, the active server 1 receives a data request from the terminal 3 through the terminal interface process 13 (step A).
1. The following steps are omitted), and it is determined whether or not the database 15 needs to be updated (A2). If the database 15 needs to be updated, the database 15
Is updated (A3), and the update information is notified to the waiting server 2 (A4).

When the update information is normally notified (A4, normal), the update information is registered as "notified" in the update history database 16 (A5), and the process ends.
Also, if the update information cannot be transmitted normally (A
(4, abnormal), the standby server 2 recognizes that the server 2 is stopped (A6), registers the update information in the update history database 16 as "not notified" (A7), and terminates.

Next, a process of switching to the standby server 2 when a failure occurs in the active server 1 will be described. FIG. 3 is a flowchart illustrating the flow of the switching process when a failure occurs in the operating server. First, the running server 1
If a failure occurs in a process inside the server due to abnormal termination,
The restart process of the abnormal process is performed in the main process 11 (B1). When recovery is performed by the restart process (B
1, recovery), and the abnormal processing ends. However, if the recovery is not successful (B1, restart processing failed), the active server 1 recognizes that a failure has occurred (B2), and notifies the standby server main process 21 that the active server 1 has a failure (B2).
3).

Here, when the main process 21 of the standby server 2 can recognize the abnormality of the main process 11 of the active server 1, the standby server 2 switches itself to the "active" processing. Transition. If the connection between the main process 11 and the main process 21 is normal (B3, connecting), the main process 1 of the running server 1
1 requests each process of its own server 1 to switch to "waiting" (B4), and at the same time, the terminal interface process 13 disconnects the connection with the terminal 3 (B5). Then, after the switching of the running server 1 to “standby” is completed,
It notifies the main process 21 of the waiting server 2 that the switching has been completed (B6).

The main process 21 of the waiting server 2
At the time when the connection error with the main process 11 of the running server 1 is received (B3, connection error), or at the time when the switch completion notification is received in the above-described step B6, each process of the own server 2 is set to the "running". Request to switch to "Medium" (B
7) At the same time, the terminal interface process 23 performs connection processing with the terminal 3 (B8). When the switching of each process of the standby server 2 is completed, the switching of the server redundant system is completed (B9), and the process ends. By switching between the two servers in this manner, the communication stop time can be minimized, and the entire system can be duplicated without intervening the operation on the terminal 3 side.

Next, a description will be given of a second embodiment of the dual server system according to the present invention. FIG. 4 is a system configuration diagram of a dual server system according to the second embodiment of this invention. In the configuration of this embodiment, a monitoring process 201 and a plurality of business processes 203 are added to the first embodiment shown in FIG. The monitoring process 201 is arranged as a higher-level device of the main process 202, and the main process 2
02 and each process in the server 200 through the main process 202. Since the monitoring process 201 only monitors the process, the monitoring process 201 becomes a more stable process as compared with the first embodiment shown in FIG. 1, and higher reliability can be obtained. In addition, business process 2
In the first embodiment shown in FIG. 1, there is only one process, but in this embodiment, a plurality of business processes 203 are configured for each business content.

[0027]

As described above, in the server duplication method of the present invention, the main process is in a state of connection with each process in its own server. Therefore, if the connection status becomes abnormal or each process becomes abnormal, the main process detects the failure promptly and notifies the failure, restarts its own server, and minimizes the failure in a minimum time. Hold down. Therefore, the processing load of the main process is reduced,
By monitoring the process, the durability of the server itself can be improved. Further, in the event of a failure, by synchronizing the databases, there is no need to perform correction processing of the database, and the system central server can be instantaneously switched.

Further, in the conventional system in which a terminal connects to a server, each terminal must be aware of two connection destination servers, and the timing of switching is difficult to understand from the terminal side. Although there were problems, these problems are solved according to the dual server system of the present invention. That is, according to the present invention, the terminal interface process of the server and the connection method of the terminal adopt a method of connecting the terminal from the server to the terminal, that is, by connecting the server process to the client, the terminal to the server connection, The side can switch the entire system without being aware of the switching.

[Brief description of the drawings]

FIG. 1 is a system configuration diagram of a redundant server system according to a first embodiment of this invention.

FIG. 2 is a flowchart showing a flow of a database synchronization process in a normal state in the system of FIG. 1;

FIG. 3 is a flowchart showing a flow of a switching process when a failure occurs in a running server in the system of FIG. 1;

FIG. 4 is a system configuration diagram of a dual server system according to a second embodiment of this invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Active server, 2 ... Standby server, 3 ... Client terminal group (terminal), 11, 21, 202 ... Main process, 12, 22, 203 ... Business process, 13, 23 ...
Terminal interface process, 14, 24 ... database update process, 15, 25 ... database, 16,
26: Update history database, 200: Server, 201
… Monitoring process

Claims (5)

[Claims] A plurality of servers serving as an active system and a standby system;
In the server duplication system constituting the non-stop system, each of the servers is connected to each process in its own server, constantly monitors each process, and is connected to a main process interconnected with another server; A database that stores information of the entire system including the business content based on the request, and an update history database that stores update information when updating the content of the database and that stores notification information results to other servers. A database update process for updating the database based on business contents, storing update information and notification information results in the update history database, and notifying the update information to another server; Server non-stop system characterized by comprising a non-stop system with system servers 2. The server redundancy system according to claim 1, wherein the main process monitors each process of its own server, and performs a process of restarting its own server when a failure occurs in any of the processes. Server duplication method characterized by performing. 3. The dual server system according to claim 1, wherein when the database update process of the active server updates the database of its own server, the database update process updates the update information of its own server. Storing the update information in a database, and notifying the update information to a database update process of the standby server, wherein the standby server updates its own database and update history database based on the received update information; A server redundancy system, wherein the database of the active server is synchronized with the database of the standby server, and when the active server fails, the communication state is continuously taken over to the standby server immediately. 4. The server duplication method according to claim 1, wherein each of the servers executes a terminal interface process for connecting to a client terminal in response to a switching notification from the main process. The server duplex system, wherein the connection between the terminal interface process and the terminal is such that the server process is a client connection and the terminal is a server connection. 5. The dual server system according to claim 1, further comprising: a monitoring process for monitoring the main process as a higher-level device, wherein the monitoring process is a process within its own server. Server duplication method characterized in that the server is monitored.